deps: bump dvc-data to 0.29.0

Per https://github.com/iterative/dvc.org/pull/3934

.codecov.yml

@@ -1,10 +0,0 @@
-comment: false
-coverage:
-  status:
-    project:
-      default:
-        target: 90
-        threshold: 10%
-    patch:
-      default:
-        threshold: 25%

.coveragerc

@@ -1,16 +0,0 @@
-[run]
-branch = True
-source = dvc
-omit =
-    *__main__.py
-
-[report]
-exclude_lines =
-    if __name__ == .__main__.:
-    if typing.TYPE_CHECKING:
-    if TYPE_CHECKING:
-    pass
-    raise NotImplementedError
-    # pragma: no cover
-    #  pragma: no cover
-    \.\.\.$

.dvcignore

@@ -1,4 +1,3 @@
 /tests
 /dvc
 /.github
-/bin

.flake8

@@ -0,0 +1,12 @@
+[flake8]
+ignore=
+    E203, # Whitespace before ':'
+    E266, # Too many leading '#' for block comment
+    W503, # Line break occurred before a binary operator
+    P1,  # unindexed parameters in the str.format, see:
+    # https://pypi.org/project/flake8-string-format/
+max_line_length = 79
+max-complexity = 15
+select = B,C,E,F,W,T4,B902,T,P
+show_source = true
+count = true

.github/ISSUE_TEMPLATE/bug_report.md

@@ -8,7 +8,7 @@ about: Create a bug report to help us improve DVC
 <!--
 ## Issue name
 
-Issue names must follow the pattern `command: description` where the command is the dvc command that you are trying to run. The description should describe the consequence of the bug. 
+Issue names must follow the pattern `command: description` where the command is the dvc command that you are trying to run. The description should describe the consequence of the bug.
 
 Example: `repro: doesn't detect input changes`
 -->

.github/ISSUE_TEMPLATE/feature_request.md

@@ -2,4 +2,3 @@
 name: "\U0001F680 Feature Request"
 about: Suggest an idea for this project
 ---
-

.github/codecov.yml

@@ -0,0 +1,12 @@
+codecov:
+  notify:
+    wait_for_ci: true
+coverage:
+  status:
+    project:
+      default:
+        target: auto
+        threshold: 2%
+    patch: off
+github_checks:
+  annotations: false

.github/release-drafter.yml

@@ -1,49 +0,0 @@
-# Config for https://github.com/apps/release-drafter
-name-template: "$RESOLVED_VERSION 🦉 "
-tag-template: "$RESOLVED_VERSION"
-categories:
-  - title: "🚀 New Features and Enhancements "
-    labels:
-      - "feature"
-      - "enhancement"
-      - "ui"
-  - title: "💣 Breaking Changes "
-    labels:
-      - "breaking-change"
-  - title: "🏇 Optimizations "
-    labels:
-      - "optimize"
-      - "performance"
-  - title: "🐛 Bug Fixes "
-    labels:
-      - "bugfix"
-      - "bug"
-  - title: "🔨 Maintenance "
-    labels:
-      - "maintenance"
-      - "refactoring"
-      - "chore"
-      - "build"
-      - "ci"
-change-template: "- $TITLE (#$NUMBER) @$AUTHOR"
-version-resolver:
-  major:
-    labels:
-      - "major"
-  minor:
-    labels:
-      - "minor"
-  patch:
-    labels:
-      - "patch"
-  default: patch
-exclude-labels:
-  - "skip-changelog"
-template: |
-  > Refer to https://dvc.org/doc/install for installation instructions.
-
-  ## Changes
-
-  $CHANGES
-
-  Thanks again to $CONTRIBUTORS for the contributions! 🎉

.github/release.yml

@@ -0,0 +1,16 @@
+changelog:
+  exclude:
+    labels: ["skip-changelog"]
+  categories:
+    - title: 🛠 Breaking Changes
+      labels: ["breaking-change"]
+    - title: 🚀 New Features and Enhancements
+      labels: ["feature", "enhancement", "ui"]
+    - title: ⚡ Optimizations
+      labels: ["optimize", "performance"]
+    - title: 🐛 Bug Fixes
+      labels: ["bugfix", "bug"]
+    - title: 🔨 Maintenance
+      labels: ["maintenance", "refactoring", "chore", "build", "ci"]
+    - title: Other Changes
+      labels: ["*"]

.github/workflows/benchmarks.yaml

@@ -12,7 +12,7 @@ jobs:
     steps:
       - uses: actions/setup-python@v4
         with:
-            python-version: "3.10"
+            python-version: "3.11"
       - uses: iterative/dvc-bench@main
         with:
             pytest_options: "-k 'test_init or test_help'"

.github/workflows/codeql.yml

@@ -0,0 +1,39 @@
+name: "CodeQL"
+
+on:
+  push:
+    branches: ["main"]
+  pull_request:
+    branches: ["main"]
+  schedule:
+    - cron: '32 19 * * 2'
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  analyze:
+    name: Analyze
+    runs-on: ubuntu-latest
+    permissions:
+      actions: read
+      contents: read
+      security-events: write
+
+    strategy:
+      fail-fast: false
+      matrix:
+        language: ['python']
+
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v3
+
+    - name: Initialize CodeQL
+      uses: github/codeql-action/init@v2
+      with:
+        languages: ${{ matrix.language }}
+
+    - name: Perform CodeQL Analysis
+      uses: github/codeql-action/analyze@v2

.github/workflows/packages.yaml

@@ -7,8 +7,13 @@ on:
     types: [released, prereleased]
   workflow_dispatch:
 
+permissions:  # added using https://github.com/step-security/secure-workflows
+  contents: read
+
 jobs:
   binary:
+    permissions:
+      contents: write  # for actions/upload-release-asset to upload release asset
     strategy:
       matrix:
         include:
@@ -24,13 +29,13 @@ jobs:
       with:
           fetch-depth: 0
 
-    - name: Set up Python 3.9
+    - name: Set up Python 3.10
       uses: actions/setup-python@v4
       with:
-        python-version: 3.9
+        python-version: "3.10"
         cache: pip
         cache-dependency-path: |
-          setup.cfg
+          pyproject.toml
           scripts/build-requirements.txt
 
     - name: Set up Ruby 2.6
@@ -45,7 +50,7 @@ jobs:
 
     - name: Install deps
       run: |
-        pip install -U wheel pip setuptools
+        pip install --upgrade pip wheel setuptools
         pip install .[all]
         pip install -r scripts/build-requirements.txt
 
@@ -77,21 +82,31 @@ jobs:
       uses: actions/setup-python@v4
       with:
         python-version: 3.8
-        cache: pip
-        cache-dependency-path: setup.cfg
 
     - name: Install
+      run: pip install --upgrade pip wheel
+
+    - name: Force version for Test PyPI uploads
+      if: ${{ !startsWith(github.ref, 'refs/tags') }}
       run: |
-        pip install -U wheel pip build setuptools
-        pip install ".[all,tests]"
+        pip install setuptools_scm
+        echo version=$(python -m setuptools_scm | awk -F+ '{print $1}' | tail -1) >> $GITHUB_ENV
 
     - name: Build packages
-      run: |
-        pip install -U pip
-        ./scripts/build_package.sh
+      run: ./scripts/build_package.sh
+      env:
+        SETUPTOOLS_SCM_PRETEND_VERSION: ${{ env.version }}
 
     - name: Publish packages to PyPI
       if: github.event_name == 'release'
       uses: pypa/gh-action-pypi-publish@release/v1
       with:
         password: ${{ secrets.PYPI_TOKEN }}
+
+    - name: Publish to Test PyPI
+      if: ${{ github.event_name == 'release' || (github.event_name == 'push' && github.ref == 'refs/heads/main') }}
+      uses: pypa/gh-action-pypi-publish@release/v1
+      with:
+        password: ${{ secrets.TEST_PYPI_TOKEN }}
+        repository_url: https://test.pypi.org/legacy/
+        skip_existing: true

.github/workflows/tests.yaml

@@ -8,15 +8,13 @@ on:
     - cron: '5 1 * * *'  # every day at 01:05
   workflow_dispatch:
 
-env:
-  DVC_TEST: "true"
-  HOMEBREW_NO_AUTO_UPDATE: 1
-  SHELL: /bin/bash
-
 concurrency:
   group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
   cancel-in-progress: true
 
+permissions:  # added using https://github.com/step-security/secure-workflows
+  contents: read
+
 jobs:
   lint:
     timeout-minutes: 10
@@ -29,13 +27,12 @@ jobs:
       uses: actions/setup-python@v4
       with:
         python-version: 3.8
+        cache: pip
+        cache-dependency-path: pyproject.toml
     - name: Install requirements
       run: |
-        pip install --upgrade pip setuptools wheel
-        pip install collective.checkdocs==0.2
+        pip install --upgrade pip wheel
         pip install ".[dev]"
-    - name: Check README
-      run: python setup.py checkdocs
     - uses: pre-commit/[email protected]
   tests:
     timeout-minutes: 45
@@ -43,13 +40,16 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-20.04, windows-latest, macos-latest]
-        pyv: ["3.8", "3.9", "3.10"]
-        pytest-filter: 
+        os: [ubuntu-20.04, windows-latest]
+        pyv: ["3.8", "3.9", "3.10", "3.11"]
+        pytest-filter:
         - "import or plot or live or experiment"
         - "not (import or plot or live or experiment)"
         include:
-        - {os: ubuntu-latest, pyv: "3.11-dev"}
+        - {os: macos-latest, pyv: "3.8"}
+        - {os: macos-latest, pyv: "3.9"}
+        - {os: macos-latest, pyv: "3.10"}
+        - {os: macos-latest, pyv: "3.11"}
 
     steps:
     - uses: actions/checkout@v3
@@ -60,25 +60,33 @@ jobs:
       with:
         python-version: ${{ matrix.pyv }}
         cache: pip
-        cache-dependency-path: setup.cfg
+        cache-dependency-path: pyproject.toml
     - name: install
-      if: steps.cache.pip-cache-dir.cache-hit != 'true'
       run: |
-        pip install --upgrade pip setuptools wheel
+        pip install --upgrade pip wheel
         pip install -e ".[dev]"
     - name: run tests
-      timeout-minutes: 40
-      run: pytest -nauto -ra --durations 100 --cov=dvc --cov-report=xml --cov-report=term -k "${{ matrix.pytest-filter }}"
+      timeout-minutes: 30
+      run: >
+        pytest -nauto --timeout=300 --durations=100
+        --cov --cov-report=xml --cov-report=term
+        -k "${{ matrix.pytest-filter }}"
     - name: upload coverage report
       uses: codecov/codecov-action@v3
       with:
         file: ./coverage.xml
         fail_ci_if_error: false
+  check:
+    if: always()
+    needs: [lint, tests]
+    runs-on: ubuntu-latest
+    steps:
+      - uses: re-actors/alls-green@release/v1
+        with:
+          jobs: ${{ toJSON(needs) }}
   notify:
     if: github.ref == 'refs/heads/main' && failure()
-    needs:
-      - lint
-      - tests
+    needs: [lint, tests]
     runs-on: ubuntu-latest
     steps:
     - name: Slack Notification

.pre-commit-config.yaml

@@ -1,39 +1,62 @@
+default_language_version:
+  python: python3
+
 ci:
   skip: [mypy, pylint, dvc-pre-commit]
 
 repos:
-  - hooks:
-      - id: black
-        language_version: python3
-    repo: https://github.com/ambv/black
-    rev: 22.10.0
   - repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v4.3.0
     hooks:
+      - id: check-added-large-files
+      - id: check-case-conflict
+      - id: check-executables-have-shebangs
+      - id: check-json
+      - id: check-merge-conflict
+        exclude: "tests/func/utils/test_strict_yaml.py"
+        args: ['--assume-in-merge']
       - id: check-toml
       - id: check-yaml
-  - hooks:
-      - id: codespell
-        args: 
-          - --ignore-words-list
-          - ba,datas,fo,uptodate,cachable,falsy
-    repo: https://github.com/codespell-project/codespell
-    rev: v2.2.2
-  - hooks:
-      - id: isort
-        language_version: python3
-    repo: https://github.com/timothycrosley/isort
+      - id: debug-statements
+        exclude: "dvc/_debug.py"
+      - id: end-of-file-fixer
+      - id: mixed-line-ending
+      - id: sort-simple-yaml
+      - id: trailing-whitespace
+  - repo: https://github.com/lovesegfault/beautysh
+    rev: v6.2.1
+    hooks:
+      - id: beautysh
+        args: ["-i", "2"]
+  - repo: https://github.com/asottile/pyupgrade
+    rev: v3.2.0
+    hooks:
+      - id: pyupgrade
+        args:
+          - --py38-plus
+  - repo: https://github.com/psf/black
+    rev: 22.10.0
+    hooks:
+      - id: black
+  - repo: https://github.com/PyCQA/isort
     rev: 5.10.1
-  - hooks:
+    hooks:
+      - id: isort
+  - repo: https://github.com/codespell-project/codespell
+    rev: v2.2.2
+    hooks:
+      - id: codespell
+        additional_dependencies: ["tomli"]
+  - repo: https://github.com/PyCQA/flake8
+    rev: 5.0.4
+    hooks:
       - id: flake8
-        language_version: python3
         additional_dependencies:
+          - flake8-broken-line
           - flake8-bugbear
           - flake8-comprehensions
           - flake8-debugger
           - flake8-string-format
-    repo: https://gitlab.com/pycqa/flake8
-    rev: 3.9.2
   - repo: local
     hooks:
     - id: mypy
@@ -47,14 +70,6 @@ repos:
       entry: pylint
       language: system
       types: [python]
-  - hooks:
-      - args:
-          - -i
-          - "2"
-        id: beautysh
-        language_version: python3
-    repo: https://github.com/lovesegfault/beautysh
-    rev: v6.2.1
   - hooks:
       - args:
           - git-hook

CODE_OF_CONDUCT.md

@@ -74,4 +74,3 @@ available at https://www.contributor-covenant.org/version/1/4/code-of-conduct.ht
 
 For answers to common questions about this code of conduct, see
 https://www.contributor-covenant.org/faq
-

MANIFEST.in

@@ -1,3 +0,0 @@
-include LICENSE
-recursive-include requirements *.txt
-recursive-include dvc/repo/plots/templates *.json

README.rst

@@ -13,7 +13,7 @@
 
 |CI| |Python Version| |Coverage| |VS Code| |DOI|
 
-|PyPI| |Packages| |Brew| |Conda| |Choco| |Snap|
+|PyPI| |PyPI Downloads| |Packages| |Brew| |Conda| |Choco| |Snap|
 
 |
 
@@ -276,6 +276,10 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://pypi.org/project/dvc
    :alt: PyPI
 
+.. |PyPI Downloads| image:: https://img.shields.io/pypi/dm/dvc.svg?color=blue&label=Downloads&logo=pypi&logoColor=gold
+   :target: https://pypi.org/project/dvc
+   :alt: PyPI Downloads
+
 .. |Packages| image:: https://img.shields.io/github/v/release/iterative/dvc?label=deb|pkg|rpm|exe&logo=GitHub
    :target: https://github.com/iterative/dvc/releases/latest
    :alt: deb|pkg|rpm|exe
@@ -292,6 +296,6 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://github.com/iterative/dvc/graphs/contributors
    :alt: Contributors
 
-.. |VS Code| image:: https://vsmarketplacebadge.apphb.com/version/Iterative.dvc.svg
+.. |VS Code| image:: https://img.shields.io/visual-studio-marketplace/v/Iterative.dvc?color=blue&label=VSCode&logo=visualstudiocode&logoColor=blue
    :target: https://marketplace.visualstudio.com/items?itemName=Iterative.dvc
    :alt: VS Code Extension

bin/dvc

@@ -1,3 +0,0 @@
-#!/bin/bash
-
-PYTHONPATH=$DVC_HOME python -m dvc $@

bin/dvc.bat

@@ -1,3 +0,0 @@
-@echo OFF
-set PYTHONPATH=%DVC_HOME%
-python -m dvc %*

dvc/__init__.py

@@ -4,6 +4,6 @@ DVC
 Make your data science projects reproducible and shareable.
 """
 import dvc.logger
-from dvc.version import __version__  # noqa: F401
+from dvc.version import __version__, version_tuple  # noqa: F401
 
 dvc.logger.setup()

dvc/_debug.py

@@ -27,7 +27,9 @@ def viztracer_profile(path: Union[Callable[[], str], str], depth: int = -1):
 
 @contextmanager
 def yappi_profile(
-    path: Union[Callable[[], str], str] = None, wall_clock: bool = True
+    path: Union[Callable[[], str], str] = None,
+    wall_clock: Optional[bool] = True,
+    separate_threads: Optional[bool] = False,
 ):
     try:
         import yappi  # pylint: disable=import-error
@@ -42,14 +44,26 @@ def yappi_profile(
     yield
     yappi.stop()
 
-    # pylint:disable=no-member
-    if path:
-        stats = yappi.get_func_stats()
-        fpath = path() if callable(path) else path
-        stats.save(fpath, "callgrind")
+    threads = yappi.get_thread_stats()
+    stats = {}
+    if separate_threads:
+        for thread in threads:
+            ctx_id = thread.id
+            stats[ctx_id] = yappi.get_func_stats(ctx_id=ctx_id)
     else:
-        yappi.get_func_stats().print_all()
-        yappi.get_thread_stats().print_all()
+        stats[None] = yappi.get_func_stats()
+
+    fpath = path() if callable(path) else path
+    for ctx_id, st in stats.items():
+        if fpath:
+            out = f"{fpath}-{ctx_id}" if ctx_id is not None else fpath
+            st.save(out, type="callgrind")
+        else:
+            if ctx_id is not None:
+                print(f"\nThread {ctx_id}")
+            st.print_all()  # pylint:disable=no-member
+            if ctx_id is None:
+                threads.print_all()  # pylint:disable=no-member
 
     yappi.clear_stats()
 
@@ -155,7 +169,12 @@ def debugtools(args: "Namespace" = None, **kwargs):
             stack.enter_context(show_stack())
         if kw.get("yappi"):
             path_func = _get_path_func("callgrind", "out")
-            stack.enter_context(yappi_profile(path=path_func))
+            stack.enter_context(
+                yappi_profile(
+                    path=path_func,
+                    separate_threads=kw.get("yappi_separate_threads"),
+                )
+            )
         if kw.get("viztracer") or kw.get("viztracer_depth"):
             path_func = _get_path_func("viztracer", "json")
             depth = kw.get("viztracer_depth") or -1
@@ -173,6 +192,12 @@ def add_debugging_flags(parser):
     parser.add_argument(
         "--yappi", action="store_true", default=False, help=SUPPRESS
     )
+    parser.add_argument(
+        "--yappi-separate-threads",
+        action="store_true",
+        default=False,
+        help=SUPPRESS,
+    )
     parser.add_argument(
         "--viztracer", action="store_true", default=False, help=SUPPRESS
     )

dvc/cli/parser.py

@@ -2,8 +2,8 @@
 import argparse
 import logging
 import os
-import sys
 
+from dvc import __version__
 from dvc.commands import (
     add,
     cache,
@@ -128,17 +128,6 @@ class DvcParser(argparse.ArgumentParser):
         return args
 
 
-class VersionAction(argparse.Action):  # pragma: no cover
-    # pylint: disable=too-few-public-methods
-    """Shows DVC version and exits."""
-
-    def __call__(self, parser, namespace, values, option_string=None):
-        from dvc import __version__
-
-        print(__version__)
-        sys.exit(0)
-
-
 def get_parent_parser():
     """Create instances of a parser containing common arguments shared among
     all the commands.
@@ -186,13 +175,11 @@ def get_main_parser():
         help="Show this help message and exit.",
     )
 
-    # NOTE: On some python versions action='version' prints to stderr
-    # instead of stdout https://bugs.python.org/issue18920
     parser.add_argument(
         "-V",
         "--version",
-        action=VersionAction,
-        nargs=0,
+        action="version",
+        version=__version__,
         help="Show program's version.",
     )
 

dvc/commands/experiments/__init__.py

@@ -18,6 +18,7 @@ from dvc.commands.experiments import (
     queue_worker,
     remove,
     run,
+    save,
     show,
 )
 
@@ -34,6 +35,7 @@ SUB_COMMANDS = [
     queue_worker,
     remove,
     run,
+    save,
     show,
 ]
 

dvc/commands/experiments/ls.py

@@ -3,6 +3,7 @@ import logging
 
 from dvc.cli.command import CmdBase
 from dvc.cli.utils import append_doc_link
+from dvc.ui import ui
 
 logger = logging.getLogger(__name__)
 
@@ -16,22 +17,13 @@ class CmdExperimentsList(CmdBase):
             num=self.args.num,
             git_remote=self.args.git_remote,
         )
-        tags = self.repo.scm.describe(exps)
-        remained = {baseline for baseline, tag in tags.items() if tag is None}
-        base = "refs/heads/"
-        ref_heads = self.repo.scm.describe(remained, base=base)
 
         for baseline in exps:
-            name = (
-                tags[baseline]
-                or ref_heads[baseline][len(base) :]
-                or baseline[:7]
-            )
             if not name_only:
-                print(f"{name}:")
+                ui.write(f"{baseline}:")
             for exp_name in exps[baseline]:
                 indent = "" if name_only else "\t"
-                print(f"{indent}{exp_name}")
+                ui.write(f"{indent}{exp_name}")
 
         return 0
 

dvc/commands/experiments/pull.py

@@ -68,7 +68,7 @@ def add_parser(experiments_subparsers, parent_parser):
         "-f",
         "--force",
         action="store_true",
-        help="Replace local experiment already exists.",
+        help="Replace local experiment if it already exists.",
     )
     experiments_pull_parser.add_argument(
         "--no-cache",

dvc/commands/experiments/save.py

@@ -0,0 +1,79 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.exceptions import DvcException
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsSave(CmdBase):
+    def run(self):
+
+        try:
+            ref = self.repo.experiments.save(
+                name=self.args.name,
+                force=self.args.force,
+                include_untracked=self.args.include_untracked,
+            )
+        except DvcException:
+            logger.exception("failed to save experiment")
+            return 1
+
+        if self.args.json:
+            ui.write_json({"ref": ref})
+        else:
+            name = self.repo.experiments.get_exact_name([ref])[ref]
+            ui.write(f"Experiment has been saved as: {name}")
+            ui.write(
+                "\nTo promote an experiment to a Git branch run:\n\n"
+                "\tdvc exp branch <exp> <branch>\n"
+            )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    EXPERIMENTS_SAVE_HELP = "Save current workspace as a dvc experiment."
+    save_parser = experiments_subparsers.add_parser(
+        "save",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_SAVE_HELP, "exp/save"),
+        help=EXPERIMENTS_SAVE_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    save_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Replace experiment if it already exists.",
+    )
+    save_parser.add_argument(
+        "--json",
+        "--show-json",
+        action="store_true",
+        default=False,
+        help="Show output in JSON format.",
+    )
+    save_parser.add_argument(
+        "-n",
+        "--name",
+        default=None,
+        help=(
+            "Human-readable experiment name. If not specified, a name will "
+            "be auto-generated."
+        ),
+        metavar="<name>",
+    )
+    save_parser.add_argument(
+        "-I",
+        "--include-untracked",
+        action="append",
+        default=[],
+        help="List of untracked paths to include in the experiment.",
+        metavar="<path>",
+    )
+    save_parser.set_defaults(func=CmdExperimentsSave)

dvc/commands/get.py

@@ -4,6 +4,7 @@ import logging
 from dvc.cli import completion
 from dvc.cli.command import CmdBaseNoRepo
 from dvc.exceptions import DvcException
+from dvc.scm import CloneError
 
 from ..cli.utils import append_doc_link
 
@@ -44,6 +45,9 @@ class CmdGet(CmdBaseNoRepo):
                 jobs=self.args.jobs,
             )
             return 0
+        except CloneError:
+            logger.exception(f"failed to get '{self.args.path}'")
+            return 1
         except DvcException:
             logger.exception(
                 "failed to get '{}' from '{}'".format(

dvc/commands/imp.py

@@ -5,6 +5,7 @@ from dvc.cli import completion
 from dvc.cli.command import CmdBase
 from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
+from dvc.scm import CloneError
 
 logger = logging.getLogger(__name__)
 
@@ -26,6 +27,9 @@ class CmdImport(CmdBase):
                 meta=self.args.meta,
                 jobs=self.args.jobs,
             )
+        except CloneError:
+            logger.exception(f"failed to import '{self.args.path}'")
+            return 1
         except DvcException:
             logger.exception(
                 "failed to import '{}' from '{}'.".format(

dvc/commands/imp_url.py

@@ -73,6 +73,18 @@ def add_parser(subparsers, parent_parser):
         help="Specify name of the .dvc file this command will generate.",
         metavar="<filename>",
     ).complete = completion.DIR
+    import_parser.add_argument(
+        "--to-remote",
+        action="store_true",
+        default=False,
+        help="Download it directly to the remote",
+    )
+    import_parser.add_argument(
+        "-r",
+        "--remote",
+        help="Remote storage to download to",
+        metavar="<name>",
+    )
     no_download_exec_group = import_parser.add_mutually_exclusive_group()
     no_download_exec_group.add_argument(
         "--no-exec",
@@ -87,18 +99,6 @@ def add_parser(subparsers, parent_parser):
         help="Create .dvc file including target data hash value(s)"
         " but do not actually download the file(s).",
     )
-    import_parser.add_argument(
-        "--to-remote",
-        action="store_true",
-        default=False,
-        help="Download it directly to the remote",
-    )
-    import_parser.add_argument(
-        "-r",
-        "--remote",
-        help="Remote storage to download to",
-        metavar="<name>",
-    )
     import_parser.add_argument(
         "-j",
         "--jobs",

dvc/commands/queue/kill.py

@@ -11,13 +11,18 @@ class CmdQueueKill(CmdBase):
     """Kill exp task in queue."""
 
     def run(self):
-        self.repo.experiments.celery_queue.kill(revs=self.args.task)
+        self.repo.experiments.celery_queue.kill(
+            revs=self.args.task, force=self.args.force
+        )
 
         return 0
 
 
 def add_parser(queue_subparsers, parent_parser):
-    QUEUE_KILL_HELP = "Kill actively running experiment queue tasks."
+    QUEUE_KILL_HELP = (
+        "Gracefully interrupt running experiment queue tasks "
+        "(equivalent to Ctrl-C)"
+    )
     queue_kill_parser = queue_subparsers.add_parser(
         "kill",
         parents=[parent_parser],
@@ -25,6 +30,13 @@ def add_parser(queue_subparsers, parent_parser):
         help=QUEUE_KILL_HELP,
         formatter_class=argparse.RawDescriptionHelpFormatter,
     )
+    queue_kill_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Forcefully and immediately kill running experiment queue tasks",
+    )
     queue_kill_parser.add_argument(
         "task",
         nargs="*",

dvc/commands/stage.py

@@ -238,26 +238,6 @@ def _add_common_args(parser):
         help="Declare output plot file (do not put into DVC cache).",
         metavar="<path>",
     )
-    parser.add_argument(
-        "--live", help="Declare output as dvclive.", metavar="<path>"
-    )
-    parser.add_argument(
-        "--live-no-cache",
-        help="Declare output as dvclive (do not put into DVC cache).",
-        metavar="<path>",
-    )
-    parser.add_argument(
-        "--live-no-summary",
-        action="store_true",
-        default=False,
-        help=argparse.SUPPRESS,
-    )
-    parser.add_argument(
-        "--live-no-html",
-        action="store_true",
-        default=False,
-        help="Signal dvclive logger to not produce training report.",
-    )
     parser.add_argument(
         "-w",
         "--wdir",

dvc/commands/update.py

@@ -31,7 +31,8 @@ class CmdUpdate(CmdBase):
 def add_parser(subparsers, parent_parser):
     UPDATE_HELP = (
         "Update data artifact imported (via dvc import or dvc import-url) "
-        "from an external DVC repository or URL."
+        "from an external DVC repository or URL, or update data from a "
+        "DVC worktree remote."
     )
     update_parser = subparsers.add_parser(
         "update",

dvc/config_schema.py

@@ -97,6 +97,8 @@ HTTP_COMMON = {
     "ask_password": Bool,
     "ssl_verify": Any(Bool, str),
     "method": str,
+    "connect_timeout": All(Coerce(float), Range(0, min_included=True)),
+    "read_timeout": All(Coerce(float), Range(0, min_included=True)),
     Optional("verify", default=False): Bool,
 }
 WEBDAV_COMMON = {

dvc/data_cloud.py

@@ -5,6 +5,7 @@ from typing import TYPE_CHECKING, Iterable, Optional
 
 from funcy import cached_property
 
+from dvc.config import NoRemoteError, RemoteConfigError
 from dvc_data.hashfile.db import get_index
 
 if TYPE_CHECKING:
@@ -23,6 +24,14 @@ class Remote:
 
         self.worktree = config.pop("worktree", False)
         self.config = config
+        if self.worktree:
+            version_aware = self.config.get("version_aware")
+            if version_aware is False:
+                raise RemoteConfigError(
+                    "worktree remotes require version_aware cloud"
+                )
+            if version_aware is None:
+                self.fs.version_aware = True
 
     @cached_property
     def odb(self):
@@ -54,8 +63,6 @@ class DataCloud:
         name: Optional[str] = None,
         command: str = "<command>",
     ) -> "Remote":
-        from dvc.config import NoRemoteError
-
         if not name:
             name = self.repo.config["core"].get("remote")
 
@@ -88,6 +95,10 @@ class DataCloud:
         command: str = "<command>",
     ) -> "HashFileDB":
         remote = self.get_remote(name=name, command=command)
+        if remote.fs.version_aware or remote.worktree:
+            raise NoRemoteError(
+                f"'{command}' is unsupported for cloud versioned remotes"
+            )
         return remote.odb
 
     def _log_missing(self, status: "CompareStatusResult"):

dvc/dependency/__init__.py

@@ -1,7 +1,7 @@
 from collections import defaultdict
 from typing import Any, Mapping
 
-from dvc.output import ARTIFACT_SCHEMA, Output
+from dvc.output import ARTIFACT_SCHEMA, DIR_FILES_SCHEMA, Output
 
 from .base import Dependency
 from .param import ParamsDependency
@@ -15,10 +15,11 @@ SCHEMA: Mapping[str, Any] = {
     **ARTIFACT_SCHEMA,
     **RepoDependency.REPO_SCHEMA,
     **ParamsDependency.PARAM_SCHEMA,
+    Output.PARAM_FILES: [DIR_FILES_SCHEMA],
 }
 
 
-def _get(stage, p, info, fs_config=None):
+def _get(stage, p, info, **kwargs):
     if info and info.get(RepoDependency.PARAM_REPO):
         repo = info.pop(RepoDependency.PARAM_REPO)
         return RepoDependency(repo, stage, p, info)
@@ -27,14 +28,15 @@ def _get(stage, p, info, fs_config=None):
         params = info.pop(ParamsDependency.PARAM_PARAMS)
         return ParamsDependency(stage, p, params)
 
-    return Dependency(stage, p, info, fs_config=fs_config)
+    return Dependency(stage, p, info, **kwargs)
 
 
 def loadd_from(stage, d_list):
     ret = []
     for d in d_list:
         p = d.pop(Output.PARAM_PATH, None)
-        ret.append(_get(stage, p, d))
+        files = d.pop(Output.PARAM_FILES, None)
+        ret.append(_get(stage, p, d, files=files))
     return ret
 
 

dvc/dependency/base.py

@@ -62,3 +62,8 @@ class Dependency(Output):
             self.fs_path = self.fs.path.version_path(
                 self.fs_path, self.meta.version_id
             )
+
+    def dumpd(self, **kwargs):
+        if self.fs.version_aware:
+            kwargs["with_files"] = True
+        return super().dumpd(**kwargs)

dvc/dvcfile.py

@@ -1,7 +1,18 @@
 import contextlib
 import logging
 import os
-from typing import TYPE_CHECKING, Any, Callable, Tuple, TypeVar, Union
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Callable,
+    Dict,
+    List,
+    Tuple,
+    TypeVar,
+    Union,
+)
+
+from funcy import cached_property
 
 from dvc.exceptions import DvcException
 from dvc.parsing.versions import LOCKFILE_VERSION, SCHEMA_KWD
@@ -180,6 +191,10 @@ class SingleStageFile(FileMixin):
     from dvc.schema import COMPILED_SINGLE_STAGE_SCHEMA as SCHEMA
     from dvc.stage.loader import SingleStageLoader as LOADER
 
+    metrics: List[str] = []
+    plots: Dict[str, Any] = {}
+    params: List[str] = []
+
     @property
     def stage(self):
         data, raw = self._load()
@@ -223,6 +238,11 @@ class PipelineFile(FileMixin):
     def _lockfile(self):
         return Lockfile(self.repo, os.path.splitext(self.path)[0] + ".lock")
 
+    def _reset(self):
+        self.__dict__.pop("contents", None)
+        self.__dict__.pop("lockfile_contents", None)
+        self.__dict__.pop("resolver", None)
+
     def dump(
         self, stage, update_pipeline=True, update_lock=True, **kwargs
     ):  # pylint: disable=arguments-differ
@@ -278,11 +298,36 @@ class PipelineFile(FileMixin):
             "PipelineFile has multiple stages. Please specify it's name."
         )
 
+    @cached_property
+    def contents(self):
+        return self._load()[0]
+
+    @cached_property
+    def lockfile_contents(self):
+        return self._lockfile.load()
+
+    @cached_property
+    def resolver(self):
+        from .parsing import DataResolver
+
+        wdir = self.repo.fs.path.parent(self.path)
+        return DataResolver(self.repo, wdir, self.contents)
+
     @property
     def stages(self):
-        data, _ = self._load()
-        lockfile_data = self._lockfile.load()
-        return self.LOADER(self, data, lockfile_data)
+        return self.LOADER(self, self.contents, self.lockfile_contents)
+
+    @property
+    def metrics(self):
+        return self.contents.get("metrics", [])
+
+    @property
+    def plots(self):
+        return self.contents.get("plots", {})
+
+    @property
+    def params(self):
+        return self.contents.get("params", [])
 
     def remove(self, force=False):
         if not force:

dvc/env.py

@@ -2,11 +2,10 @@ DVC_CHECKPOINT = "DVC_CHECKPOINT"
 DVC_DAEMON = "DVC_DAEMON"
 DVC_PAGER = "DVC_PAGER"
 DVC_ROOT = "DVC_ROOT"
-DVCLIVE_PATH = "DVCLIVE_PATH"
-DVCLIVE_SUMMARY = "DVCLIVE_SUMMARY"
-DVCLIVE_HTML = "DVCLIVE_HTML"
 DVCLIVE_RESUME = "DVCLIVE_RESUME"
 DVC_IGNORE_ISATTY = "DVC_IGNORE_ISATTY"
+DVC_EXP_BASELINE_REV = "DVC_EXP_BASELINE_REV"
+DVC_EXP_NAME = "DVC_EXP_NAME"
 DVC_EXP_GIT_REMOTE = "DVC_EXP_GIT_REMOTE"
 DVC_EXP_AUTO_PUSH = "DVC_EXP_AUTO_PUSH"
 DVC_NO_ANALYTICS = "DVC_NO_ANALYTICS"

dvc/exceptions.py

@@ -132,10 +132,17 @@ class NotDvcRepoError(DvcException):
 class CyclicGraphError(DvcException):
     def __init__(self, stages):
         assert isinstance(stages, list)
-        msg = "Pipeline has a cycle involving: {}.".format(
-            ", ".join(s.addressing for s in stages)
+        stage_part = "stage" if len(stages) == 1 else "stages"
+        msg = (
+            "Same item(s) are defined as both a dependency and an output "
+            "in {stage_part}: {stage}."
+        )
+        super().__init__(
+            msg.format(
+                stage_part=stage_part,
+                stage=", ".join(s.addressing for s in stages),
+            )
         )
-        super().__init__(msg)
 
 
 class ConfirmRemoveError(DvcException):

dvc/fs/dvc.py

@@ -276,43 +276,38 @@ class _DVCFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         key = self._get_key_from_relative(path)
         repo, dvc_fs, subkey = self._get_subrepo_info(key)
 
-        names = set()
+        dvc_infos = {}
         if dvc_fs:
             with suppress(FileNotFoundError):
                 dvc_path = _get_dvc_path(dvc_fs, subkey)
-                for entry in dvc_fs.ls(dvc_path, detail=False):
-                    names.add(dvc_fs.path.name(entry))
+                for info in dvc_fs.ls(dvc_path, detail=True):
+                    dvc_infos[dvc_fs.path.name(info["name"])] = info
 
+        fs_infos = {}
         ignore_subrepos = kwargs.get("ignore_subrepos", True)
         if not dvc_only:
             fs = self.repo.fs
             fs_path = self._from_key(key)
             try:
-                for entry in repo.dvcignore.ls(
-                    fs, fs_path, detail=False, ignore_subrepos=ignore_subrepos
+                for info in repo.dvcignore.ls(
+                    fs, fs_path, detail=True, ignore_subrepos=ignore_subrepos
                 ):
-                    names.add(fs.path.name(entry))
+                    fs_infos[fs.path.name(info["name"])] = info
             except (FileNotFoundError, NotADirectoryError):
                 pass
 
         dvcfiles = kwargs.get("dvcfiles", False)
-        if not dvcfiles:
-            names = (name for name in names if not _is_dvc_file(name))
 
         infos = []
         paths = []
+        names = set(dvc_infos.keys()) | set(fs_infos.keys())
         for name in names:
-            entry_path = self.path.join(path, name)
-            entry_key = key + (name,)
-            try:
-                info = self._info(
-                    entry_key,
-                    entry_path,
-                    ignore_subrepos=ignore_subrepos,
-                    check_ignored=False,
-                )
-            except FileNotFoundError:
+            if not dvcfiles and _is_dvc_file(name):
                 continue
+
+            entry_path = self.path.join(path, name)
+            info = _merge_info(repo, fs_infos.get(name), dvc_infos.get(name))
+            info["name"] = entry_path
             infos.append(info)
             paths.append(entry_path)
 

dvc/machine/backend/base.py

@@ -26,7 +26,7 @@ class BaseMachineBackend(ABC):
     ) -> Iterator[dict]:
         """Iterate over status of all instances of the specified machine."""
 
-    def close(self):
+    def close(self):  # noqa: B027
         pass
 
     @abstractmethod

dvc/output.py

@@ -87,10 +87,10 @@ def loadd_from(stage, d_list):
         plot = d.pop(Output.PARAM_PLOT, False)
         persist = d.pop(Output.PARAM_PERSIST, False)
         checkpoint = d.pop(Output.PARAM_CHECKPOINT, False)
-        live = d.pop(Output.PARAM_LIVE, False)
         remote = d.pop(Output.PARAM_REMOTE, None)
         annot = {field: d.pop(field, None) for field in ANNOTATION_FIELDS}
         files = d.pop(Output.PARAM_FILES, None)
+        push = d.pop(Output.PARAM_PUSH, True)
         ret.append(
             _get(
                 stage,
@@ -101,10 +101,10 @@ def loadd_from(stage, d_list):
                 plot=plot,
                 persist=persist,
                 checkpoint=checkpoint,
-                live=live,
                 remote=remote,
                 **annot,
                 files=files,
+                push=push,
             )
         )
     return ret
@@ -118,8 +118,8 @@ def loads_from(
     plot=False,
     persist=False,
     checkpoint=False,
-    live=False,
     remote=None,
+    push=True,
 ):
     return [
         _get(
@@ -131,8 +131,8 @@ def loads_from(
             plot=plot,
             persist=persist,
             checkpoint=checkpoint,
-            live=live,
             remote=remote,
+            push=push,
         )
         for s in s_list
     ]
@@ -166,29 +166,21 @@ def load_from_pipeline(stage, data, typ="outs"):
         stage.PARAM_OUTS,
         stage.PARAM_METRICS,
         stage.PARAM_PLOTS,
-        stage.PARAM_LIVE,
     ):
         raise ValueError(f"'{typ}' key is not allowed for pipeline files.")
 
     metric = typ == stage.PARAM_METRICS
     plot = typ == stage.PARAM_PLOTS
-    live = typ == stage.PARAM_LIVE
-    if live:
-        # `live` is single object
-        data = [data]
 
     d = _merge_data(data)
 
     for path, flags in d.items():
-        plt_d, live_d = {}, {}
+        plt_d = {}
         if plot:
             from dvc.schema import PLOT_PROPS
 
             plt_d, flags = _split_dict(flags, keys=PLOT_PROPS.keys())
-        if live:
-            from dvc.schema import LIVE_PROPS
 
-            live_d, flags = _split_dict(flags, keys=LIVE_PROPS.keys())
         extra = project(
             flags,
             [
@@ -196,6 +188,7 @@ def load_from_pipeline(stage, data, typ="outs"):
                 Output.PARAM_PERSIST,
                 Output.PARAM_CHECKPOINT,
                 Output.PARAM_REMOTE,
+                Output.PARAM_PUSH,
                 *ANNOTATION_FIELDS,
             ],
         )
@@ -206,7 +199,6 @@ def load_from_pipeline(stage, data, typ="outs"):
             info={},
             plot=plt_d or plot,
             metric=metric,
-            live=live_d or live,
             **extra,
         )
 
@@ -263,10 +255,8 @@ class Output:
     PARAM_PLOT_TITLE = "title"
     PARAM_PLOT_HEADER = "header"
     PARAM_PERSIST = "persist"
-    PARAM_LIVE = "live"
-    PARAM_LIVE_SUMMARY = "summary"
-    PARAM_LIVE_HTML = "html"
     PARAM_REMOTE = "remote"
+    PARAM_PUSH = "push"
 
     METRIC_SCHEMA = Any(
         None,
@@ -292,7 +282,6 @@ class Output:
         plot=False,
         persist=False,
         checkpoint=False,
-        live=False,
         desc=None,
         type=None,  # pylint: disable=redefined-builtin
         labels=None,
@@ -301,6 +290,7 @@ class Output:
         repo=None,
         fs_config=None,
         files: List[Dict[str, Any]] = None,
+        push: bool = True,
     ):
         self.annot = Annotation(
             desc=desc, type=type, labels=labels or [], meta=meta or {}
@@ -355,7 +345,7 @@ class Output:
         self.plot = False if self.IS_DEPENDENCY else plot
         self.persist = persist
         self.checkpoint = checkpoint
-        self.live = live
+        self.can_push = push
 
         self.fs_path = self._parse_path(self.fs, fs_path)
         self.obj: Optional["HashFile"] = None
@@ -377,8 +367,12 @@ class Output:
             name=self.hash_name,
             value=getattr(self.meta, self.hash_name, None),
         )
-        if self.hash_info and self.hash_info.isdir:
+        if self.meta.nfiles or self.hash_info and self.hash_info.isdir:
             self.meta.isdir = True
+            if not self.hash_info and self.hash_name != "md5":
+                md5 = getattr(self.meta, "md5", None)
+                if md5:
+                    self.hash_info = HashInfo("md5", md5)
 
     def _parse_path(self, fs, fs_path):
         parsed = urlparse(self.def_path)
@@ -523,7 +517,10 @@ class Output:
         except NoRemoteError:
             remote = None
 
-        return DataIndexEntry(
+        if self.files and not self.obj:
+            self.obj = self.get_obj()
+
+        entry = DataIndexEntry(
             meta=self.meta,
             obj=self.obj,
             hash_info=self.hash_info,
@@ -531,6 +528,16 @@ class Output:
             cache=self.odb,
             remote=remote,
         )
+        if self.stage.is_import and not self.stage.is_repo_import:
+            dep = self.stage.deps[0]
+            entry.fs = dep.fs
+            entry.path = dep.fs_path
+            entry.meta = dep.meta
+            if not entry.obj:
+                if not dep.obj and dep.files:
+                    dep.obj = dep.get_obj()
+                entry.obj = dep.obj
+        return entry
 
     def changed_checksum(self):
         return self.hash_info != self.get_hash()
@@ -631,8 +638,16 @@ class Output:
         if self.metric:
             self.verify_metric()
 
-        if not self.use_cache:
-            _, self.meta, obj = build(
+        if self.use_cache:
+            _, self.meta, self.obj = build(
+                self.odb,
+                self.fs_path,
+                self.fs,
+                self.hash_name,
+                ignore=self.dvcignore,
+            )
+        else:
+            _, self.meta, self.obj = build(
                 self.repo.odb.local,
                 self.fs_path,
                 self.fs,
@@ -640,23 +655,11 @@ class Output:
                 ignore=self.dvcignore,
                 dry_run=True,
             )
-            self.hash_info = obj.hash_info
-            self.files = None
             if not self.IS_DEPENDENCY:
                 logger.debug(
                     "Output '%s' doesn't use cache. Skipping saving.", self
                 )
-            return
 
-        assert not self.IS_DEPENDENCY
-
-        _, self.meta, self.obj = build(
-            self.odb,
-            self.fs_path,
-            self.fs,
-            self.hash_name,
-            ignore=self.dvcignore,
-        )
         self.hash_info = self.obj.hash_info
         self.files = None
 
@@ -754,40 +757,40 @@ class Output:
 
         ret[self.PARAM_PATH] = path
 
-        if self.IS_DEPENDENCY:
-            return ret
+        if not self.IS_DEPENDENCY:
+            ret.update(self.annot.to_dict())
+            if not self.use_cache:
+                ret[self.PARAM_CACHE] = self.use_cache
 
-        ret.update(self.annot.to_dict())
-        if not self.use_cache:
-            ret[self.PARAM_CACHE] = self.use_cache
+            if isinstance(self.metric, dict):
+                if (
+                    self.PARAM_METRIC_XPATH in self.metric
+                    and not self.metric[self.PARAM_METRIC_XPATH]
+                ):
+                    del self.metric[self.PARAM_METRIC_XPATH]
 
-        if isinstance(self.metric, dict):
-            if (
-                self.PARAM_METRIC_XPATH in self.metric
-                and not self.metric[self.PARAM_METRIC_XPATH]
-            ):
-                del self.metric[self.PARAM_METRIC_XPATH]
+            if self.metric:
+                ret[self.PARAM_METRIC] = self.metric
 
-        if self.metric:
-            ret[self.PARAM_METRIC] = self.metric
+            if self.plot:
+                ret[self.PARAM_PLOT] = self.plot
 
-        if self.plot:
-            ret[self.PARAM_PLOT] = self.plot
+            if self.persist:
+                ret[self.PARAM_PERSIST] = self.persist
 
-        if self.persist:
-            ret[self.PARAM_PERSIST] = self.persist
+            if self.checkpoint:
+                ret[self.PARAM_CHECKPOINT] = self.checkpoint
 
-        if self.checkpoint:
-            ret[self.PARAM_CHECKPOINT] = self.checkpoint
+            if self.remote:
+                ret[self.PARAM_REMOTE] = self.remote
 
-        if self.live:
-            ret[self.PARAM_LIVE] = self.live
+            if not self.can_push:
+                ret[self.PARAM_PUSH] = self.can_push
 
-        if self.remote:
-            ret[self.PARAM_REMOTE] = self.remote
-
-        if self.hash_info.isdir and (
-            kwargs.get("with_files") or self.files is not None
+        if (
+            (not self.IS_DEPENDENCY or self.stage.is_import)
+            and self.hash_info.isdir
+            and (kwargs.get("with_files") or self.files is not None)
         ):
             if self.obj:
                 obj = self.obj
@@ -1040,9 +1043,15 @@ class Output:
         if not self.use_cache:
             return {}
 
+        push: bool = kwargs.pop("push", False)
         if self.stage.is_repo_import:
+            if push:
+                return {}
             return self.get_used_external(**kwargs)
 
+        if push and not self.can_push:
+            return {}
+
         if not self.hash_info:
             msg = (
                 "Output '{}'({}) is missing version info. "
@@ -1182,11 +1191,39 @@ class Output:
 
     @property
     def is_metric(self) -> bool:
-        return bool(self.metric) or bool(self.live)
+        return bool(self.metric)
 
     @property
     def is_plot(self) -> bool:
-        return bool(self.plot) or bool(self.live)
+        return bool(self.plot)
+
+    def restore_fields(self, other: "Output"):
+        """Restore attributes that need to be preserved when serialized."""
+        self.annot = other.annot
+        self.remote = other.remote
+        self.can_push = other.can_push
+
+    def merge_version_meta(self, other: "Output"):
+        """Merge version meta for files which are unchanged from other."""
+        if not self.hash_info:
+            return
+        if self.hash_info.isdir:
+            return self._merge_dir_version_meta(other)
+        if self.hash_info != other.hash_info:
+            return
+        self.meta = other.meta
+
+    def _merge_dir_version_meta(self, other: "Output"):
+        from dvc_data.hashfile.tree import update_meta
+
+        if not self.obj or not other.hash_info.isdir:
+            return
+        other_obj = other.obj if other.obj is not None else other.get_obj()
+        assert isinstance(self.obj, Tree) and isinstance(other_obj, Tree)
+        updated = update_meta(self.obj, other_obj)
+        assert updated.hash_info == self.obj.hash_info
+        self.obj = updated
+        self.files = updated.as_list(with_meta=True)
 
 
 META_SCHEMA = {
@@ -1217,5 +1254,6 @@ SCHEMA = {
     Output.PARAM_CACHE: bool,
     Output.PARAM_METRIC: Output.METRIC_SCHEMA,
     Output.PARAM_REMOTE: str,
+    Output.PARAM_PUSH: bool,
     Output.PARAM_FILES: [DIR_FILES_SCHEMA],
 }

dvc/parsing/__init__.py

@@ -36,7 +36,7 @@ from .interpolate import (
 )
 
 if TYPE_CHECKING:
-    from typing_extensions import NoReturn
+    from typing import NoReturn
 
     from dvc.repo import Repo
 

dvc/parsing/interpolate.py

@@ -1,3 +1,4 @@
+import os
 import re
 import typing
 from collections.abc import Iterable, Mapping
@@ -9,10 +10,9 @@ from dvc.exceptions import DvcException
 from dvc.utils.flatten import flatten
 
 if typing.TYPE_CHECKING:
-    from typing import List, Match
+    from typing import List, Match, NoReturn
 
     from pyparsing import ParseException
-    from typing_extensions import NoReturn
 
     from .context import Context
 
@@ -71,6 +71,16 @@ def embrace(s: str):
     return BRACE_OPEN + s + BRACE_CLOSE
 
 
+def escape_str(value):
+    if os.name == "nt":
+        from subprocess import list2cmdline
+
+        return list2cmdline([value])
+    from shlex import quote
+
+    return quote(value)
+
+
 @singledispatch
 def to_str(obj) -> str:
     return str(obj)
@@ -98,12 +108,12 @@ def _(obj: dict):
                     result += f"--no-{k} "
 
         elif isinstance(v, str):
-            result += f"--{k} '{v}' "
+            result += f"--{k} {escape_str(v)} "
 
         elif isinstance(v, Iterable):
             for n, i in enumerate(v):
                 if isinstance(i, str):
-                    i = f"'{i}'"
+                    i = escape_str(i)
                 elif isinstance(i, Iterable):
                     raise ParseError(
                         f"Cannot interpolate nested iterable in '{k}'"

dvc/render/convert.py

@@ -14,14 +14,14 @@ from dvc.render.converter.vega import VegaConverter
 
 
 def _get_converter(
-    renderer_class, props
+    renderer_class, renderer_id, props, data
 ) -> Union[VegaConverter, ImageConverter]:
     from dvc_render import ImageRenderer, VegaRenderer
 
     if renderer_class.TYPE == VegaRenderer.TYPE:
-        return VegaConverter(props)
+        return VegaConverter(renderer_id, data, props)
     if renderer_class.TYPE == ImageRenderer.TYPE:
-        return ImageConverter(props)
+        return ImageConverter(renderer_id, data, props)
 
     raise ValueError(f"Invalid renderer class {renderer_class}")
 

dvc/render/converter/__init__.py

@@ -1,9 +1,19 @@
-from typing import Dict, Optional
+from typing import Any, Dict, List, Optional, Tuple
 
 
 class Converter:
-    def __init__(self, plot_properties: Optional[Dict] = None):
-        self.plot_properties = plot_properties or {}
+    def __init__(
+        self,
+        plot_id: str,
+        data: Dict[str, Any] = None,
+        properties: Optional[Dict] = None,
+    ):
+        self.plot_id = plot_id
+        self.properties = properties or {}
+        self.data = data or {}
 
-    def convert(self, data, revision: str, filename: str, **kwargs):
+    def convert(self) -> Tuple[List[Tuple[str, str, Any]], Dict]:
+        raise NotImplementedError
+
+    def flat_datapoints(self, revision: str) -> Tuple[List[Dict], Dict]:
         raise NotImplementedError

dvc/render/converter/image.py

@@ -1,6 +1,6 @@
 import base64
 import os
-from typing import TYPE_CHECKING, Dict, List, Tuple
+from typing import TYPE_CHECKING, Any, Dict, List, Tuple
 
 from dvc.render import FILENAME_FIELD, REVISION_FIELD, SRC_FIELD
 
@@ -34,25 +34,33 @@ class ImageConverter(Converter):
         base64_str = base64.b64encode(image_data).decode()
         return f"data:image;base64,{base64_str}"
 
-    def convert(
-        self, data, revision: str, filename: str, **kwargs
-    ) -> Tuple[List[Dict], Dict]:
+    def convert(self) -> Tuple[List[Tuple[str, str, Any]], Dict]:
+        datas = []
+        for filename, image_data in self.data.items():
+            datas.append((filename, "", image_data))
+        return datas, self.properties
+
+    def flat_datapoints(self, revision: str) -> Tuple[List[Dict], Dict]:
         """
         Convert the DVC Plots content to DVC Render datapoints.
         Return both generated datapoints and updated properties.
         """
-        path = self.plot_properties.get("out")
-        if path:
-            if not os.path.isdir(path):
-                os.makedirs(path, exist_ok=True)
-            src = self._write_image(
-                os.path.abspath(path), revision, filename, data
-            )
-        else:
-            src = self._encode_image(data)
-        datapoint = {
-            REVISION_FIELD: revision,
-            FILENAME_FIELD: filename,
-            SRC_FIELD: src,
-        }
-        return [datapoint], self.plot_properties
+        path = self.properties.get("out")
+        datapoints = []
+        datas, properties = self.convert()
+        for filename, _, image_data in datas:
+            if path:
+                if not os.path.isdir(path):
+                    os.makedirs(path, exist_ok=True)
+                src = self._write_image(
+                    os.path.abspath(path), revision, filename, image_data
+                )
+            else:
+                src = self._encode_image(image_data)
+            datapoint = {
+                REVISION_FIELD: revision,
+                FILENAME_FIELD: filename,
+                SRC_FIELD: src,
+            }
+            datapoints.append(datapoint)
+        return datapoints, properties

dvc/render/converter/vega.py

@@ -1,90 +1,100 @@
 from copy import deepcopy
-from functools import partial
-from typing import Dict, Iterable, List, Optional, Set, Union
+from typing import Any, Dict, Iterable, List, Tuple, Union
 
-from funcy import first, project
+from funcy import first, last
 
 from dvc.exceptions import DvcException
-from dvc.render import (
-    FILENAME_FIELD,
-    INDEX_FIELD,
-    REVISION_FIELD,
-    VERSION_FIELD,
-)
+from dvc.render import FILENAME_FIELD, INDEX_FIELD, VERSION_FIELD
 
 from . import Converter
 
 
-class FieldsNotFoundError(DvcException):
-    def __init__(self, expected_fields, found_fields):
-        expected_str = ", ".join(expected_fields)
+class FieldNotFoundError(DvcException):
+    def __init__(self, expected_field, found_fields):
         found_str = ", ".join(found_fields)
         super().__init__(
-            f"Could not find all provided fields ('{expected_str}') "
+            f"Could not find provided field ('{expected_field}') "
             f"in data fields ('{found_str}')."
         )
 
 
-class PlotDataStructureError(DvcException):
-    def __init__(self):
-        super().__init__(
-            "Plot data extraction failed. Please see "
-            "https://man.dvc.org/plots for supported data formats."
-        )
+def _lists(blob: Union[Dict, List]) -> Iterable[List]:
+    if isinstance(blob, list):
+        yield blob
+    else:
+        for _, value in blob.items():
+            if isinstance(value, dict):
+                yield from _lists(value)
+            elif isinstance(value, list):
+                yield value
 
 
-def _filter_fields(
-    datapoints: List[Dict], fields: Set, **kwargs
-) -> List[Dict]:
-    if not fields:
-        return datapoints
-    assert isinstance(fields, set)
-
-    new_data = []
-    for data_point in datapoints:
-        keys = set(data_point.keys())
-        if not fields <= keys:
-            raise FieldsNotFoundError(fields, keys)
-
-        new_data.append(project(data_point, fields))
-
-    return new_data
+def _file_field(*args):
+    for axis_def in args:
+        if axis_def is not None:
+            for file, val in axis_def.items():
+                if isinstance(val, str):
+                    yield file, val
+                elif isinstance(val, list):
+                    for field in val:
+                        yield file, field
 
 
-def _lists(dictionary: Dict):
-    for _, value in dictionary.items():
-        if isinstance(value, dict):
-            yield from _lists(value)
-        elif isinstance(value, list):
-            yield value
+def _find(
+    filename: str,
+    field: str,
+    data_series: List[Tuple[str, str, Any]],
+):
+    for data_file, data_field, data in data_series:
+        if data_file == filename and data_field == field:
+            return data_file, data_field, data
+    return None
 
 
-def _find_first_list(
-    data: Union[Dict, List], fields: Set, **kwargs
-) -> List[Dict]:
-    fields = fields or set()
-
-    if not isinstance(data, dict):
-        return data
-
-    for lst in _lists(data):
-        if (
-            all(isinstance(dp, dict) for dp in lst)
-            # if fields is empty, it will match any set
-            and set(first(lst).keys()) & fields == fields
-        ):
-            return lst
-
-    raise PlotDataStructureError()
+def _verify_field(file2datapoints: Dict[str, List], filename: str, field: str):
+    if filename in file2datapoints:
+        datapoint = first(file2datapoints[filename])
+        if field not in datapoint:
+            raise FieldNotFoundError(field, datapoint.keys())
+    return
 
 
-def _append_index(datapoints: List[Dict], **kwargs) -> List[Dict]:
-    if INDEX_FIELD in first(datapoints).keys():
-        return datapoints
+def _get_xs(properties: Dict, file2datapoints: Dict[str, List[Dict]]):
+    x = properties.get("x", None)
+    if x is not None and isinstance(x, dict):
+        for filename, field in _file_field(x):
+            _verify_field(file2datapoints, filename, field)
+            yield filename, field
 
-    for index, data_point in enumerate(datapoints):
-        data_point[INDEX_FIELD] = index
-    return datapoints
+
+def _get_ys(properties, file2datapoints: Dict[str, List[Dict]]):
+    y = properties.get("y", None)
+    if y is not None:
+        for filename, field in _file_field(y):
+            _verify_field(file2datapoints, filename, field)
+            yield filename, field
+
+
+def _is_datapoints(lst: List[Dict]):
+    """
+    check if dict keys match, datapoints with different keys mgiht lead
+    to unexpected behavior
+    """
+
+    return all(isinstance(item, dict) for item in lst) and set(
+        first(lst).keys()
+    ) == {key for keys in lst for key in keys}
+
+
+def get_datapoints(file_content: Dict):
+    result: List[Dict[str, Any]] = []
+    for lst in _lists(file_content):
+        if _is_datapoints(lst):
+            for index, datapoint in enumerate(lst):
+                if len(result) <= index:
+                    result.append({})
+                result[index].update(datapoint)
+    return result
 
 
 class VegaConverter(Converter):
@@ -95,228 +105,230 @@ class VegaConverter(Converter):
     ('x', 'y') it will attempt to fill in the blanks.
     """
 
-    def __init__(self, plot_properties: Optional[Dict] = None):
-        super().__init__(plot_properties)
+    def __init__(
+        self, plot_id: str, data: Dict = None, properties: Dict = None
+    ):
+        super().__init__(plot_id, data, properties)
+        self.plot_id = plot_id
         self.inferred_properties: Dict = {}
 
-        self.steps = []
+        # TODO we should be handling that in `convert`,
+        #      to avoid stateful `self.inferred_properties`
+        self._infer_x_y()
 
-        self._infer_x()
-        self._infer_fields()
+    def _infer_y_from_data(self):
+        if self.plot_id in self.data:
+            for lst in _lists(self.data[self.plot_id]):
+                if all(isinstance(item, dict) for item in lst):
+                    datapoint = first(lst)
+                    field = last(datapoint.keys())
+                    self.inferred_properties["y"] = {self.plot_id: field}
+                    break
 
-        self.steps.append(
-            (
-                "find_data",
-                partial(
-                    _find_first_list,
-                    fields=self.inferred_properties.get("fields", set())
-                    - {INDEX_FIELD},
-                ),
-            )
-        )
+    def _infer_x_y(self):
+        x = self.properties.get("x", None)
+        y = self.properties.get("y", None)
 
-        if not self.plot_properties.get("x", None):
-            self.steps.append(("append_index", partial(_append_index)))
-
-        self.steps.append(
-            (
-                "filter_fields",
-                partial(
-                    _filter_fields,
-                    fields=self.inferred_properties.get("fields", set()),
-                ),
-            )
-        )
-        self.steps.append(
-            (
-                "infer_y",
-                partial(
-                    self._infer_y,
-                ),
-            )
-        )
-
-        self.steps.append(
-            (
-                "generate_y",
-                partial(
-                    self._generate_y_values,
-                ),
-            )
-        )
-
-    def _infer_x(self):
-        if not self.plot_properties.get("x", None):
-            self.inferred_properties["x"] = INDEX_FIELD
-
-    def skip_step(self, name: str):
-        self.steps = [(_name, fn) for _name, fn in self.steps if _name != name]
-
-    def _infer_fields(self):
-        fields = self.plot_properties.get("fields", set())
-        if fields:
-            fields = {
-                *fields,
-                self.plot_properties.get("x", None),
-                self.plot_properties.get("y", None),
-                self.inferred_properties.get("x", None),
-            } - {None}
-            self.inferred_properties["fields"] = fields
-
-    def _infer_y(self, datapoints: List[Dict], **kwargs):
-        if "y" not in self.plot_properties:
-            data_fields = list(first(datapoints))
-            skip = (
-                REVISION_FIELD,
-                self.plot_properties.get("x", None)
-                or self.inferred_properties.get("x"),
-                FILENAME_FIELD,
-                VERSION_FIELD,
-            )
-            inferred_y = first(
-                f for f in reversed(data_fields) if f not in skip
-            )
-            if "y" in self.inferred_properties:
-                previous_y = self.inferred_properties["y"]
-                if previous_y != inferred_y:
-                    raise DvcException(
-                        f"Inferred y ('{inferred_y}' value does not match"
-                        f"previously matched one ('f{previous_y}')."
-                    )
+        # Infer x.
+        if isinstance(x, str):
+            self.inferred_properties["x"] = {}
+            # If multiple y files, duplicate x for each file.
+            if isinstance(y, dict):
+                for file, fields in y.items():
+                    # Duplicate x for each y.
+                    if isinstance(fields, list):
+                        self.inferred_properties["x"][file] = [x] * len(fields)
+                    else:
+                        self.inferred_properties["x"][file] = x
+            # Otherwise use plot ID as file.
             else:
-                self.inferred_properties["y"] = inferred_y
-        return datapoints
+                self.inferred_properties["x"][self.plot_id] = x
+
+        # Infer y.
+        if y is None:
+            self._infer_y_from_data()
+        # If y files not provided, use plot ID as file.
+        elif not isinstance(y, dict):
+            self.inferred_properties["y"] = {self.plot_id: y}
+
+    def _find_datapoints(self):
+        result = {}
+        for file, content in self.data.items():
+            result[file] = get_datapoints(content)
+
+        return result
+
+    @staticmethod
+    def infer_y_label(properties):
+        y_label = properties.get("y_label", None)
+        if y_label is None:
+            y = properties.get("y", None)
+            if isinstance(y, dict):
+                fields = {field for _, field in _file_field(y)}
+                if len(fields) == 1:
+                    y_label = first(fields)
+                else:
+                    y_label = "y"
+            elif isinstance(y, list):
+                y_label = "y"
+            elif isinstance(y, str):
+                y_label = y
+
+        return y_label
+
+    @staticmethod
+    def infer_x_label(properties):
+        x_label = properties.get("x_label", None)
+        if x_label is None:
+            x = properties.get("x", None)
+            if isinstance(x, dict):
+                fields = {field for _, field in _file_field(x)}
+                if len(fields) == 1:
+                    x_label = first(fields)
+                else:
+                    x_label = "x"
+            else:
+                x_label = INDEX_FIELD
+        return x_label
+
+    def flat_datapoints(self, revision):
+
+        file2datapoints, properties = self.convert()
+
+        props_update = {}
+
+        xs = list(_get_xs(properties, file2datapoints))
+
+        # assign "step" if no x provided
+        if not xs:
+            x_file, x_field = (
+                None,
+                INDEX_FIELD,
+            )
+        else:
+            x_file, x_field = xs[0]
+        props_update["x"] = x_field
+
+        ys = list(_get_ys(properties, file2datapoints))
+
+        num_xs = len(xs)
+        num_ys = len(ys)
+        if num_xs > 1 and num_xs != num_ys:
+            raise DvcException(
+                f"Cannot have different number of x and y data sources. Found "
+                f"{num_xs} x and {num_ys} y data sources."
+            )
+
+        all_datapoints = []
+        all_y_fields = {y_field for _, y_field in ys}
+
+        # override to unified y field name if there are different y fields
+        if len(all_y_fields) > 1:
+            props_update["y"] = "dvc_inferred_y_value"
+        else:
+            props_update["y"] = first(all_y_fields)
+
+        for i, (y_file, y_field) in enumerate(ys):
+            if num_xs > 1:
+                x_file, x_field = xs[i]
+            datapoints = deepcopy(file2datapoints.get(y_file, []))
+
+            if props_update.get("y", None) == "dvc_inferred_y_value":
+                _update_from_field(
+                    datapoints,
+                    field="dvc_inferred_y_value",
+                    source_field=y_field,
+                )
+
+            if x_field == INDEX_FIELD and x_file is None:
+                _update_from_index(datapoints, INDEX_FIELD)
+            else:
+                x_datapoints = file2datapoints.get(x_file, [])
+                try:
+                    _update_from_field(
+                        datapoints,
+                        field=x_field,
+                        source_datapoints=x_datapoints,
+                    )
+                except IndexError:
+                    raise DvcException(
+                        f"Cannot join '{x_field}' from '{x_file}' and "
+                        f"'{y_field}' from '{y_file}'. "
+                        "They have to have same length."
+                    )
+
+            _update_all(
+                datapoints,
+                update_dict={
+                    VERSION_FIELD: {
+                        "revision": revision,
+                        FILENAME_FIELD: y_file,
+                        "field": y_field,
+                    }
+                },
+            )
+
+            all_datapoints.extend(datapoints)
+
+        if not all_datapoints:
+            return [], {}
+
+        properties = {**properties, **props_update}
+
+        return all_datapoints, properties
 
     def convert(
         self,
-        data,
-        revision: str,
-        filename: str,
-        skip: List = None,
-        **kwargs,
     ):
         """
         Convert the data. Fill necessary fields ('x', 'y') and return both
-        generated datapoints and updated properties.
+        generated datapoints and updated properties. If `x` is not provided,
+        leave it as None, fronteds should handle it.
+
+        NOTE: Studio uses this method.
+              The only thing studio FE handles is filling `x` and `y`.
+              `x/y_label` should be filled here.
+
+              Datapoints are not stripped according to config, because users
+              might be utilizing other fields in their custom plots.
         """
-        if not skip:
-            skip = []
+        datapoints = self._find_datapoints()
+        properties = {**self.properties, **self.inferred_properties}
 
-        processed = deepcopy(data)
+        properties["y_label"] = self.infer_y_label(properties)
+        properties["x_label"] = self.infer_x_label(properties)
 
-        for step_name, step in self.steps:
-            if step_name not in skip:
-                processed = step(  # type: ignore
-                    processed,
-                    revision=revision,
-                    filename=filename,
-                )
+        return datapoints, properties
 
-        return processed, {**self.plot_properties, **self.inferred_properties}
 
-    def _generate_y_values(  # noqa: C901
-        self,
-        datapoints: List[Dict],
-        revision: str,
-        filename: str,
-        **kwargs,
-    ) -> List[Dict]:
+def _update_from_field(
+    target_datapoints: List[Dict],
+    field: str,
+    source_datapoints: List[Dict] = None,
+    source_field: str = None,
+):
+    if source_datapoints is None:
+        source_datapoints = target_datapoints
+    if source_field is None:
+        source_field = field
 
-        y_values = self.plot_properties.get(
-            "y", None
-        ) or self.inferred_properties.get("y", None)
+    if len(source_datapoints) != len(target_datapoints):
+        raise IndexError(
+            "Source and target datapoints must have the same length"
+        )
 
-        assert y_values is not None
+    for index, datapoint in enumerate(target_datapoints):
+        source_datapoint = source_datapoints[index]
+        if source_field in source_datapoint:
+            datapoint[field] = source_datapoint[source_field]
 
-        result = []
-        properties_update = {}
 
-        def _add_version_info(datapoint, version_info):
-            tmp = datapoint.copy()
-            tmp[VERSION_FIELD] = version_info
-            return tmp
+def _update_from_index(datapoints: List[Dict], new_field: str):
+    for index, datapoint in enumerate(datapoints):
+        datapoint[new_field] = index
 
-        def _version_info(revision, filename=None, field=None):
-            res = {"revision": revision}
-            if filename is not None:
-                res["filename"] = filename
-            if field is not None:
-                res["field"] = field
-            return res
 
-        def _generate_y(datapoint, field):
-            tmp = datapoint.copy()
-            tmp["dvc_inferred_y_value"] = datapoint[field]
-            tmp = _add_version_info(
-                tmp, _version_info(revision, filename, field)
-            )
-            if (
-                "y_label" not in properties_update
-                and "y_label" not in self.plot_properties
-            ):
-                properties_update["y_label"] = "y"
+def _update_all(datapoints: List[Dict], update_dict: Dict):
 
-            properties_update["y"] = "dvc_inferred_y_value"
-
-            return tmp
-
-        if isinstance(y_values, str):
-            for datapoint in datapoints:
-                result.append(
-                    _add_version_info(
-                        datapoint, _version_info(revision, filename, y_values)
-                    )
-                )
-
-        if isinstance(y_values, list):
-            for datapoint in datapoints:
-                for y_val in y_values:
-                    if y_val in datapoint:
-                        result.append(_generate_y(datapoint, y_val))
-
-        if isinstance(y_values, dict):
-
-            def _to_set(values: Iterable):
-                result = set()
-                for val in values:
-                    if isinstance(val, list):
-                        for elem in val:
-                            result.add(elem)
-                    else:
-                        result.add(val)
-
-                return result
-
-            all_fields = _to_set(y_values.values())
-            if (
-                all([isinstance(field, str) for field in all_fields])
-                and len(all_fields) == 1
-            ):
-                # if we use the same field from all files,
-                # we dont have to generate it
-                y_field = all_fields.pop()
-                for datapoint in datapoints:
-                    result.append(
-                        _add_version_info(
-                            datapoint,
-                            _version_info(revision, filename, y_field),
-                        )
-                    )
-                properties_update.update({"y": y_field})
-            else:
-                for def_filename, val in y_values.items():
-                    if isinstance(val, str):
-                        fields = [val]
-                    if isinstance(val, list):
-                        fields = val
-                    for datapoint in datapoints:
-                        for field in fields:
-                            if field in datapoint and def_filename in filename:
-                                result.append(_generate_y(datapoint, field))
-
-        self.inferred_properties = {
-            **self.inferred_properties,
-            **properties_update,
-        }
-
-        return result
+    for datapoint in datapoints:
+        datapoint.update(update_dict)

dvc/render/match.py

@@ -42,7 +42,7 @@ class PlotsData:
         return dict(groups)
 
     def get_definition_data(self, target_files, rev):
-        result = []
+        result = {}
         for file in target_files:
             file_content = (
                 self.data.get(rev, {})
@@ -52,7 +52,7 @@ class PlotsData:
                 .get("data", {})
             )
             if file_content:
-                result.append((file, file_content))
+                result[file] = file_content
 
         return result
 
@@ -91,15 +91,14 @@ def match_defs_renderers(
                 renderer_cls = VegaRenderer
                 renderer_id = plot_id
 
-            converter = _get_converter(renderer_cls, props)
+            converter = _get_converter(
+                renderer_cls, inner_id, props, definitions_data
+            )
 
-            for filename, plot_data in definitions_data:
-                dps, final_props = converter.convert(
-                    revision=rev,
-                    filename=filename,
-                    data=plot_data,
-                )
-                plot_datapoints.extend(dps)
+            dps, rev_props = converter.flat_datapoints(rev)
+            if not final_props and rev_props:
+                final_props = rev_props
+            plot_datapoints.extend(dps)
 
         if "title" not in final_props:
             final_props["title"] = renderer_id

dvc/repo/__init__.py

@@ -269,7 +269,7 @@ class Repo:
         except SCMError:
             if self._uninitialized:
                 # might not be a git/dvc repo at all
-                # used in `params/metrics/plots/live` targets
+                # used in `params/metrics/plots` targets
                 return SCM(self.root_dir, no_scm=True)
             raise
 
@@ -401,6 +401,7 @@ class Repo:
         used_run_cache=None,
         revs=None,
         num=1,
+        push: bool = False,
     ):
         """Get the stages related to the given target and collect
         the `info` of its outputs.
@@ -435,6 +436,7 @@ class Repo:
                 jobs=jobs,
                 recursive=recursive,
                 with_deps=with_deps,
+                push=push,
             ).items():
                 used[odb].update(objs)
 
@@ -534,11 +536,16 @@ class Repo:
         self.scm.close()
         self.state.close()
 
+    def _reset_cached_indecies(self):
+        self.__dict__.pop("index", None)
+        self.__dict__.pop("dvcignore", None)
+        self.__dict__.pop("dvcfs", None)
+        self.__dict__.pop("datafs", None)
+
     def _reset(self):
         self.state.close()
         self.scm._reset()  # pylint: disable=protected-access
-        self.__dict__.pop("index", None)
-        self.__dict__.pop("dvcignore", None)
+        self._reset_cached_indecies()
 
     def __enter__(self):
         return self

dvc/repo/add.py

@@ -187,7 +187,7 @@ def add(  # noqa: C901
                 stage.transfer(source, to_remote=to_remote, odb=odb, **kwargs)
             else:
                 try:
-                    stage.save()
+                    stage.save(merge_versioned=True)
                     if not no_commit:
                         stage.commit()
                 except CacheLinkError:

dvc/repo/brancher.py

@@ -77,7 +77,7 @@ def brancher(  # noqa: E302
         from dvc.fs import GitFileSystem
 
         for sha, names in found_revs.items():
-            self.__dict__.pop("index", None)
+            self._reset_cached_indecies()  # pylint: disable=W0212
             self.fs = GitFileSystem(scm=scm, rev=sha)
             self.root_dir = self.fs.path.join("/", *repo_root_parts)
 

dvc/repo/commit.py

@@ -56,10 +56,13 @@ def commit(
     ]
     for stage_info in stages_info:
         stage = stage_info.stage
-        changes = stage.changed_entries()
-        if any(changes):
-            prompt_to_commit(stage, changes, force=force)
+        if force:
             stage.save(allow_missing=allow_missing)
+        else:
+            changes = stage.changed_entries()
+            if any(changes):
+                prompt_to_commit(stage, changes, force=force)
+                stage.save(allow_missing=allow_missing)
         stage.commit(
             filter_info=stage_info.filter_info, allow_missing=allow_missing
         )

dvc/repo/data.py

@@ -12,6 +12,7 @@ from typing import (
     Optional,
     Tuple,
     TypedDict,
+    Union,
     cast,
 )
 
@@ -19,10 +20,9 @@ from dvc.fs.git import GitFileSystem
 from dvc.ui import ui
 
 if TYPE_CHECKING:
-    from scmrepo.base import Base
-
     from dvc.output import Output
     from dvc.repo import Repo
+    from dvc.scm import Git, NoSCM
     from dvc_data.hashfile.db import HashFileDB
     from dvc_data.hashfile.hash_info import HashInfo
     from dvc_data.hashfile.obj import HashFile
@@ -140,7 +140,9 @@ class GitInfo(TypedDict, total=False):
     is_dirty: bool
 
 
-def _git_info(scm: "Base", untracked_files: str = "all") -> GitInfo:
+def _git_info(
+    scm: Union["Git", "NoSCM"], untracked_files: str = "all"
+) -> GitInfo:
     from scmrepo.exceptions import SCMError
 
     from dvc.scm import NoSCM

dvc/repo/experiments/__init__.py

@@ -2,9 +2,9 @@ import logging
 import os
 import re
 import time
-from typing import Any, Dict, Iterable, Optional
+from typing import Dict, Iterable, Optional
 
-from funcy import cached_property, first
+from funcy import cached_property, chain, first
 
 from dvc.exceptions import DvcException
 from dvc.ui import ui
@@ -16,7 +16,7 @@ from .exceptions import (
     InvalidExpRefError,
     MultipleBranchError,
 )
-from .executor.base import BaseExecutor, ExecutorInfo, TaskStatus
+from .executor.base import BaseExecutor
 from .queue.base import BaseStashQueue, QueueEntry
 from .queue.celery import LocalCeleryQueue
 from .queue.tempdir import TempDirQueue
@@ -25,7 +25,6 @@ from .refs import (
     CELERY_FAILED_STASH,
     CELERY_STASH,
     EXEC_APPLY,
-    EXEC_BRANCH,
     EXEC_CHECKPOINT,
     EXEC_NAMESPACE,
     EXPS_NAMESPACE,
@@ -33,7 +32,7 @@ from .refs import (
     ExpRefInfo,
 )
 from .stash import ExpStashEntry
-from .utils import exp_refs_by_rev, scm_locked, unlocked_repo
+from .utils import check_ref_format, exp_refs_by_rev, unlocked_repo
 
 logger = logging.getLogger(__name__)
 
@@ -51,18 +50,12 @@ class Experiments:
     )
 
     def __init__(self, repo):
-        from dvc.lock import make_lock
         from dvc.scm import NoSCMError
 
         if repo.config["core"].get("no_scm", False):
             raise NoSCMError
 
         self.repo = repo
-        self.scm_lock = make_lock(
-            os.path.join(self.repo.tmp_dir, "exp_scm_lock"),
-            tmp_dir=self.repo.tmp_dir,
-            hardlink_lock=repo.config["core"].get("hardlink_lock", False),
-        )
 
     @property
     def scm(self):
@@ -172,12 +165,22 @@ class Experiments:
     ) -> Dict[str, str]:
         results: Dict[str, str] = {}
         if entries is None:
-            entries = list(self.celery_queue.iter_queued())
+            entries = list(
+                chain(
+                    self.celery_queue.iter_active(),
+                    self.celery_queue.iter_queued(),
+                )
+            )
+
+        logger.debug(
+            "reproduce all these entries '%s'",
+            entries,
+        )
+
         if not entries:
             return results
 
-        # TODO: re-enable --jobs concurrency
-        self.celery_queue.spawn_worker()
+        self.celery_queue.start_workers(count=kwargs.get("jobs", 1))
         failed = []
         try:
             ui.write(
@@ -190,7 +193,10 @@ class Experiments:
                     time.sleep(1)
                 self.celery_queue.follow(entry)
                 # wait for task collection to complete
-                result = self.celery_queue.get_result(entry)
+                try:
+                    result = self.celery_queue.get_result(entry)
+                except FileNotFoundError:
+                    result = None
                 if result is None or result.exp_hash is None:
                     name = entry.name or entry.stash_rev[:7]
                     failed.append(name)
@@ -229,18 +235,6 @@ class Experiments:
             "\tdvc exp branch <exp> <branch>\n"
         )
 
-    def _validate_new_ref(self, exp_ref: ExpRefInfo):
-        from .utils import check_ref_format
-
-        if not exp_ref.name:
-            return
-
-        check_ref_format(self.scm, exp_ref)
-
-        if self.scm.get_ref(str(exp_ref)):
-            raise ExperimentExistsError(exp_ref.name)
-
-    @scm_locked
     def new(
         self,
         queue: BaseStashQueue,
@@ -259,13 +253,13 @@ class Experiments:
 
         name = kwargs.get("name", None)
         baseline_sha = kwargs.get("baseline_rev") or self.repo.scm.get_rev()
-        exp_ref = ExpRefInfo(baseline_sha=baseline_sha, name=name)
 
-        try:
-            self._validate_new_ref(exp_ref)
-        except ExperimentExistsError as err:
-            if not (kwargs.get("force", False) or kwargs.get("reset", False)):
-                raise err
+        if name:
+            exp_ref = ExpRefInfo(baseline_sha=baseline_sha, name=name)
+            check_ref_format(self.scm, exp_ref)
+            force = kwargs.get("force", False)
+            if self.scm.get_ref(str(exp_ref)) and not force:
+                raise ExperimentExistsError(exp_ref.name)
 
         return queue.put(*args, **kwargs)
 
@@ -438,70 +432,15 @@ class Experiments:
             result[rev] = name
         return result
 
-    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Any]:
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Dict]:
         """Return info for running experiments."""
         result = {}
-        infofile = self.workspace_queue.get_infofile_path("workspace")
-        result.update(
-            self._fetch_running_exp("workspace", infofile, fetch_refs)
-        )
-        for queue in (self.tempdir_queue, self.celery_queue):
-            for entry in queue.iter_active():
-                infofile = queue.get_infofile_path(entry.stash_rev)
-                result.update(
-                    self._fetch_running_exp(
-                        entry.stash_rev, infofile, fetch_refs
-                    )
-                )
-        return result
-
-    def _fetch_running_exp(
-        self, rev: str, infofile: str, fetch_refs: bool
-    ) -> Dict[str, Any]:
-        from dvc.scm import InvalidRemoteSCMRepo
-        from dvc.utils.serialize import load_json
-
-        from .executor.local import TempDirExecutor
-
-        result: Dict[str, Any] = {}
-        try:
-            info = ExecutorInfo.from_dict(load_json(infofile))
-        except OSError:
-            return result
-        if info.result is None:
-            if rev == "workspace":
-                # If we are appending to a checkpoint branch in a workspace
-                # run, show the latest checkpoint as running.
-                if info.status > TaskStatus.RUNNING:
-                    return result
-                last_rev = self.scm.get_ref(EXEC_BRANCH)
-                if last_rev:
-                    result[last_rev] = info.asdict()
-                else:
-                    result[rev] = info.asdict()
-            else:
-                result[rev] = info.asdict()
-                if info.git_url and fetch_refs:
-
-                    def on_diverged(_ref: str, _checkpoint: bool):
-                        return False
-
-                    executor = TempDirExecutor.from_info(info)
-                    try:
-                        for ref in executor.fetch_exps(
-                            self.scm,
-                            on_diverged=on_diverged,
-                        ):
-                            logger.debug(
-                                "Updated running experiment '%s'.", ref
-                            )
-                            last_rev = self.scm.get_ref(ref)
-                            result[rev]["last"] = last_rev
-                            if last_rev:
-                                result[last_rev] = info.asdict()
-                    except InvalidRemoteSCMRepo:
-                        # ignore stale info files
-                        del result[rev]
+        for queue in (
+            self.workspace_queue,
+            self.tempdir_queue,
+            self.celery_queue,
+        ):
+            result.update(queue.get_running_exps(fetch_refs))
         return result
 
     def apply(self, *args, **kwargs):
@@ -529,6 +468,11 @@ class Experiments:
 
         return run(self.repo, *args, **kwargs)
 
+    def save(self, *args, **kwargs):
+        from dvc.repo.experiments.save import save
+
+        return save(self.repo, *args, **kwargs)
+
     def gc(self, *args, **kwargs):
         from dvc.repo.experiments.gc import gc
 

dvc/repo/experiments/apply.py

@@ -16,9 +16,10 @@ from .executor.base import BaseExecutor
 from .refs import EXEC_APPLY
 
 if TYPE_CHECKING:
-    from scmrepo import Git
+    from scmrepo.git import Git
 
     from dvc.repo import Repo
+    from dvc.repo.experiments import Experiments
 
 logger = logging.getLogger(__name__)
 
@@ -31,20 +32,33 @@ def apply(repo: "Repo", rev: str, force: bool = True, **kwargs):
     from dvc.repo.checkout import checkout as dvc_checkout
     from dvc.scm import GitMergeError, RevError, resolve_rev
 
-    exps = repo.experiments
+    exps: "Experiments" = repo.experiments
+
+    is_stash: bool = False
 
     try:
         exp_rev = resolve_rev(repo.scm, rev)
         exps.check_baseline(exp_rev)
-    except (RevError, BaselineMismatchError) as exc:
+    except RevError as exc:
+        (
+            exp_ref_info,
+            queue_entry,
+        ) = exps.celery_queue.get_ref_and_entry_by_names(rev)[rev]
+        baseline_sha = repo.scm.get_rev()
+        if exp_ref_info:
+            if baseline_sha != exp_ref_info.baseline_sha:
+                raise InvalidExpRevError(rev)
+            exp_rev = repo.scm.get_ref(str(exp_ref_info))
+        elif queue_entry:
+            if queue_entry.baseline_rev != baseline_sha:
+                raise InvalidExpRevError(rev)
+            exp_rev = queue_entry.stash_rev
+            is_stash = True
+        else:
+            raise InvalidExpRevError(rev) from exc
+    except BaselineMismatchError as exc:
         raise InvalidExpRevError(rev) from exc
 
-    stash_rev = exp_rev in exps.stash_revs
-    if not stash_rev and not exps.get_branch_by_rev(
-        exp_rev, allow_multiple=True
-    ):
-        raise InvalidExpRevError(exp_rev)
-
     # NOTE: we don't use scmrepo's stash_workspace() here since we need
     # finer control over the merge behavior when we unstash everything
     with _apply_workspace(repo, rev, force):
@@ -55,7 +69,7 @@ def apply(repo: "Repo", rev: str, force: bool = True, **kwargs):
 
     repo.scm.reset()
 
-    if stash_rev:
+    if is_stash:
         args_path = os.path.join(repo.tmp_dir, BaseExecutor.PACKED_ARGS_FILE)
         if os.path.exists(args_path):
             remove(args_path)

dvc/repo/experiments/diff.py

@@ -7,7 +7,7 @@ logger = logging.getLogger(__name__)
 
 
 def diff(repo, *args, a_rev=None, b_rev=None, param_deps=False, **kwargs):
-    from dvc.repo.experiments.show import _collect_experiment_commit
+    from dvc.repo.experiments.show import collect_experiment_commit
     from dvc.scm import resolve_rev
 
     if repo.scm.no_commits:
@@ -15,15 +15,15 @@ def diff(repo, *args, a_rev=None, b_rev=None, param_deps=False, **kwargs):
 
     if a_rev:
         rev = resolve_rev(repo.scm, a_rev)
-        old = _collect_experiment_commit(repo, rev, param_deps=param_deps)
+        old = collect_experiment_commit(repo, rev, param_deps=param_deps)
     else:
-        old = _collect_experiment_commit(repo, "HEAD", param_deps=param_deps)
+        old = collect_experiment_commit(repo, "HEAD", param_deps=param_deps)
 
     if b_rev:
         rev = resolve_rev(repo.scm, b_rev)
-        new = _collect_experiment_commit(repo, rev, param_deps=param_deps)
+        new = collect_experiment_commit(repo, rev, param_deps=param_deps)
     else:
-        new = _collect_experiment_commit(
+        new = collect_experiment_commit(
             repo, "workspace", param_deps=param_deps
         )
 

dvc/repo/experiments/exceptions.py

@@ -6,12 +6,6 @@ if TYPE_CHECKING:
     from .refs import ExpRefInfo
 
 
-class UnchangedExperimentError(DvcException):
-    def __init__(self, rev):
-        super().__init__(f"Experiment unchanged from '{rev[:7]}'.")
-        self.rev = rev
-
-
 class BaselineMismatchError(DvcException):
     def __init__(self, rev, expected):
         if hasattr(rev, "hexsha"):
@@ -25,13 +19,13 @@ class BaselineMismatchError(DvcException):
 
 
 class ExperimentExistsError(DvcException):
-    def __init__(self, name: str):
+    def __init__(self, name: str, command: str = "run"):
         msg = (
             "Experiment conflicts with existing experiment "
             f"'{name}'. To overwrite the existing experiment run:\n\n"
-            "\tdvc exp run -f ...\n\n"
-            "To run this experiment with a different name run:\n\n"
-            f"\tdvc exp run -n <new_name> ...\n"
+            f"\tdvc exp {command} -f ...\n\n"
+            f"To {command} this experiment with a different name run:\n\n"
+            f"\tdvc exp {command} -n <new_name> ...\n"
         )
         super().__init__(msg)
         self.name = name

dvc/repo/experiments/executor/base.py

@@ -12,6 +12,7 @@ from typing import (
     Callable,
     Dict,
     Iterable,
+    List,
     NamedTuple,
     Optional,
     Tuple,
@@ -20,6 +21,8 @@ from typing import (
     Union,
 )
 
+from scmrepo.exceptions import SCMError
+
 from dvc.env import DVC_EXP_AUTO_PUSH, DVC_EXP_GIT_REMOTE
 from dvc.exceptions import DvcException
 from dvc.stage.serialize import to_lockfile
@@ -27,22 +30,8 @@ from dvc.ui import ui
 from dvc.utils import dict_sha256, env2bool, relpath
 from dvc.utils.fs import remove
 
-from ..exceptions import (
-    CheckpointExistsError,
-    ExperimentExistsError,
-    UnchangedExperimentError,
-)
-from ..refs import (
-    EXEC_BASELINE,
-    EXEC_BRANCH,
-    EXEC_CHECKPOINT,
-    EXEC_HEAD,
-    EXEC_MERGE,
-    EXEC_NAMESPACE,
-    EXPS_NAMESPACE,
-    EXPS_STASH,
-    ExpRefInfo,
-)
+from ..exceptions import CheckpointExistsError, ExperimentExistsError
+from ..refs import EXEC_BASELINE, EXEC_BRANCH, EXEC_CHECKPOINT, ExpRefInfo
 
 if TYPE_CHECKING:
     from queue import Queue
@@ -56,10 +45,6 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 
-EXEC_TMP_DIR = "exps"
-EXEC_PID_DIR = "run"
-
-
 class ExecutorResult(NamedTuple):
     exp_hash: Optional[str]
     ref_info: Optional["ExpRefInfo"]
@@ -170,6 +155,7 @@ class BaseExecutor(ABC):
     @abstractmethod
     def init_git(
         self,
+        repo: "Repo",
         scm: "Git",
         stash_rev: str,
         entry: "ExpStashEntry",
@@ -317,6 +303,7 @@ class BaseExecutor(ABC):
     def fetch_exps(
         self,
         dest_scm: "Git",
+        refs: List[str],
         force: bool = False,
         on_diverged: Callable[[str, bool], None] = None,
         **kwargs,
@@ -325,26 +312,19 @@ class BaseExecutor(ABC):
 
         Args:
             dest_scm: Destination Git instance.
+            refs: reference names to be fetched from the remotes.
             force: If True, diverged refs will be overwritten
             on_diverged: Callback in the form on_diverged(ref, is_checkpoint)
                 to be called when an experiment ref has diverged.
 
         Extra kwargs will be passed into the remote git client.
         """
-        from ..utils import iter_remote_refs
 
-        refs = []
-        has_checkpoint = False
-        for ref in iter_remote_refs(
-            dest_scm,
-            self.git_url,
-            base=EXPS_NAMESPACE,
-            **kwargs,
-        ):
-            if ref == EXEC_CHECKPOINT:
-                has_checkpoint = True
-            elif not ref.startswith(EXEC_NAMESPACE) and ref != EXPS_STASH:
-                refs.append(ref)
+        if EXEC_CHECKPOINT in refs:
+            refs.remove(EXEC_CHECKPOINT)
+            has_checkpoint = True
+        else:
+            has_checkpoint = False
 
         def on_diverged_ref(orig_ref: str, new_rev: str):
             if force:
@@ -361,21 +341,25 @@ class BaseExecutor(ABC):
             return False
 
         # fetch experiments
-        dest_scm.fetch_refspecs(
-            self.git_url,
-            [f"{ref}:{ref}" for ref in refs],
-            on_diverged=on_diverged_ref,
-            force=force,
-            **kwargs,
-        )
-        # update last run checkpoint (if it exists)
-        if has_checkpoint:
+        try:
             dest_scm.fetch_refspecs(
                 self.git_url,
-                [f"{EXEC_CHECKPOINT}:{EXEC_CHECKPOINT}"],
-                force=True,
+                [f"{ref}:{ref}" for ref in refs],
+                on_diverged=on_diverged_ref,
+                force=force,
                 **kwargs,
             )
+            # update last run checkpoint (if it exists)
+            if has_checkpoint:
+                dest_scm.fetch_refspecs(
+                    self.git_url,
+                    [f"{EXEC_CHECKPOINT}:{EXEC_CHECKPOINT}"],
+                    force=True,
+                    **kwargs,
+                )
+        except SCMError:
+            pass
+
         return refs
 
     @classmethod
@@ -508,7 +492,6 @@ class BaseExecutor(ABC):
                     info,
                     stages,
                     exp_hash,
-                    checkpoint_reset,
                     auto_push,
                     git_remote,
                     repro_force,
@@ -529,29 +512,20 @@ class BaseExecutor(ABC):
         info,
         stages,
         exp_hash,
-        checkpoint_reset,
         auto_push,
         git_remote,
         repro_force,
     ) -> Tuple[Optional[str], Optional["ExpRefInfo"], bool]:
-        try:
-            is_checkpoint = any(stage.is_checkpoint for stage in stages)
-            if is_checkpoint and checkpoint_reset:
-                # For reset checkpoint stages, we need to force
-                # overwriting existing checkpoint refs even though
-                # repro may not have actually been run with --force
-                repro_force = True
-            cls.commit(
-                dvc.scm,
-                exp_hash,
-                exp_name=info.name,
-                force=repro_force,
-                checkpoint=is_checkpoint,
-            )
-            if auto_push:
-                cls._auto_push(dvc, dvc.scm, git_remote)
-        except UnchangedExperimentError:
-            pass
+        is_checkpoint = any(stage.is_checkpoint for stage in stages)
+        cls.commit(
+            dvc.scm,
+            exp_hash,
+            exp_name=info.name,
+            force=repro_force,
+            checkpoint=is_checkpoint,
+        )
+        if auto_push:
+            cls._auto_push(dvc, dvc.scm, git_remote)
         ref: Optional[str] = dvc.scm.get_ref(EXEC_BRANCH, follow=False)
         exp_ref: Optional["ExpRefInfo"] = (
             ExpRefInfo.from_ref(ref) if ref else None
@@ -596,29 +570,22 @@ class BaseExecutor(ABC):
             logger.debug("Running repro in '%s'", os.getcwd())
             yield dvc
             info.status = TaskStatus.SUCCESS
-            if infofile is not None:
-                info.dump_json(infofile)
-
         except CheckpointKilledError:
             info.status = TaskStatus.FAILED
-            if infofile is not None:
-                info.dump_json(infofile)
             raise
         except DvcException:
             if log_errors:
                 logger.exception("")
             info.status = TaskStatus.FAILED
-            if infofile is not None:
-                info.dump_json(infofile)
             raise
         except Exception:
             if log_errors:
                 logger.exception("unexpected error")
             info.status = TaskStatus.FAILED
-            if infofile is not None:
-                info.dump_json(infofile)
             raise
         finally:
+            if infofile is not None:
+                info.dump_json(infofile)
             dvc.close()
             os.chdir(old_cwd)
 
@@ -669,18 +636,15 @@ class BaseExecutor(ABC):
         unchanged: Iterable["PipelineStage"],
         stages: Iterable["PipelineStage"],
     ):
-        try:
-            exp_hash = cls.hash_exp(list(stages) + list(unchanged))
-            exp_rev = cls.commit(
-                scm, exp_hash, exp_name=name, force=force, checkpoint=True
-            )
+        exp_hash = cls.hash_exp(list(stages) + list(unchanged))
+        exp_rev = cls.commit(
+            scm, exp_hash, exp_name=name, force=force, checkpoint=True
+        )
 
-            if env2bool(DVC_EXP_AUTO_PUSH):
-                git_remote = os.getenv(DVC_EXP_GIT_REMOTE)
-                cls._auto_push(dvc, scm, git_remote)
-            ui.write(f"Checkpoint experiment iteration '{exp_rev[:7]}'.")
-        except UnchangedExperimentError:
-            pass
+        if env2bool(DVC_EXP_AUTO_PUSH):
+            git_remote = os.getenv(DVC_EXP_GIT_REMOTE)
+            cls._auto_push(dvc, scm, git_remote)
+        ui.write(f"Checkpoint experiment iteration '{exp_rev[:7]}'.")
 
     @classmethod
     def commit(
@@ -695,7 +659,6 @@ class BaseExecutor(ABC):
         rev = scm.get_rev()
         if not scm.is_dirty(untracked_files=False):
             logger.debug("No changes to commit")
-            raise UnchangedExperimentError(rev)
 
         check_conflict = False
         branch = scm.get_ref(EXEC_BRANCH, follow=False)
@@ -756,21 +719,12 @@ class BaseExecutor(ABC):
             dvc_logger.setLevel(level)
 
     @contextmanager
-    def set_exec_refs(
-        self, scm: "Git", stash_rev: str, entry: "ExpStashEntry"
-    ):
+    def set_temp_refs(self, scm: "Git", temp_dict: Dict[str, str]):
         try:
-            # Executor will be initialized with an empty git repo that
-            # we populate by pushing:
-            #   EXEC_HEAD - the base commit for this experiment
-            #   EXEC_MERGE - the unmerged changes (from our stash)
-            #       to be reproduced
-            #   EXEC_BASELINE - the baseline commit for this experiment
-            scm.set_ref(EXEC_HEAD, entry.head_rev)
-            scm.set_ref(EXEC_MERGE, stash_rev)
-            scm.set_ref(EXEC_BASELINE, entry.baseline_rev)
+            for ref, rev in temp_dict.items():
+                scm.set_ref(ref, rev)
             yield
         finally:
-            for ref in (EXEC_HEAD, EXEC_MERGE, EXEC_BASELINE):
+            for ref in temp_dict:
                 if scm.get_ref(ref):
                     scm.remove_ref(ref)

dvc/repo/experiments/executor/local.py

@@ -4,9 +4,11 @@ from contextlib import ExitStack
 from tempfile import mkdtemp
 from typing import TYPE_CHECKING, Optional
 
-from funcy import cached_property
+from funcy import cached_property, retry
 from scmrepo.exceptions import SCMError as _SCMError
+from shortuuid import uuid
 
+from dvc.lock import LockError
 from dvc.scm import SCM, GitMergeError
 from dvc.utils.fs import makedirs, remove
 
@@ -18,15 +20,17 @@ from ..refs import (
     EXEC_HEAD,
     EXEC_MERGE,
     EXEC_NAMESPACE,
+    EXPS_TEMP,
+    ExpRefInfo,
 )
-from .base import EXEC_TMP_DIR, BaseExecutor, TaskStatus
+from ..utils import EXEC_TMP_DIR, get_exp_rwlock
+from .base import BaseExecutor, TaskStatus
 
 if TYPE_CHECKING:
     from scmrepo.git import Git
 
     from dvc.repo import Repo
 
-    from ..refs import ExpRefInfo
     from ..stash import ExpStashEntry
 
 logger = logging.getLogger(__name__)
@@ -47,9 +51,9 @@ class BaseLocalExecutor(BaseExecutor):
         return SCM(self.root_dir)
 
     def cleanup(self, infofile: str):
-        super().cleanup(infofile)
         self.scm.close()
         del self.scm
+        super().cleanup(infofile)
 
     def collect_cache(
         self, repo: "Repo", exp_ref: "ExpRefInfo", run_cache: bool = True
@@ -67,8 +71,10 @@ class TempDirExecutor(BaseLocalExecutor):
     QUIET = True
     DEFAULT_LOCATION = "tempdir"
 
+    @retry(180, errors=LockError, timeout=1)
     def init_git(
         self,
+        repo: "Repo",
         scm: "Git",
         stash_rev: str,
         entry: "ExpStashEntry",
@@ -85,18 +91,43 @@ class TempDirExecutor(BaseLocalExecutor):
         if infofile:
             self.info.dump_json(infofile)
 
-        with self.set_exec_refs(scm, stash_rev, entry):
-            refspec = f"{EXEC_NAMESPACE}/"
-            push_refspec(scm, self.git_url, refspec, refspec)
+        temp_head = f"{EXPS_TEMP}/head-{uuid()}"
+        temp_merge = f"{EXPS_TEMP}/merge-{uuid()}"
+        temp_baseline = f"{EXPS_TEMP}/baseline-{uuid()}"
 
-        if branch:
-            push_refspec(scm, self.git_url, branch, branch)
-            self.scm.set_ref(EXEC_BRANCH, branch, symbolic=True)
-        elif self.scm.get_ref(EXEC_BRANCH):
-            self.scm.remove_ref(EXEC_BRANCH)
+        temp_ref_dict = {
+            temp_head: entry.head_rev,
+            temp_merge: stash_rev,
+            temp_baseline: entry.baseline_rev,
+        }
+        with get_exp_rwlock(
+            repo, writes=[temp_head, temp_merge, temp_baseline]
+        ), self.set_temp_refs(scm, temp_ref_dict):
+            # Executor will be initialized with an empty git repo that
+            # we populate by pushing:
+            #   EXEC_HEAD - the base commit for this experiment
+            #   EXEC_MERGE - the unmerged changes (from our stash)
+            #       to be reproduced
+            #   EXEC_BASELINE - the baseline commit for this experiment
+            refspec = [
+                (temp_head, EXEC_HEAD),
+                (temp_merge, EXEC_MERGE),
+                (temp_baseline, EXEC_BASELINE),
+            ]
+
+            if branch:
+                refspec.append((branch, branch))
+                with get_exp_rwlock(repo, reads=[branch]):
+                    push_refspec(scm, self.git_url, refspec)
+                self.scm.set_ref(EXEC_BRANCH, branch, symbolic=True)
+            else:
+                push_refspec(scm, self.git_url, refspec)
+                if self.scm.get_ref(EXEC_BRANCH):
+                    self.scm.remove_ref(EXEC_BRANCH)
+
+            if self.scm.get_ref(EXEC_CHECKPOINT):
+                self.scm.remove_ref(EXEC_CHECKPOINT)
 
-        if self.scm.get_ref(EXEC_CHECKPOINT):
-            self.scm.remove_ref(EXEC_CHECKPOINT)
         # checkout EXEC_HEAD and apply EXEC_MERGE on top of it without
         # committing
         head = EXEC_BRANCH if branch else EXEC_HEAD
@@ -167,8 +198,10 @@ class WorkspaceExecutor(BaseLocalExecutor):
         logger.debug("Init workspace executor in '%s'", root_dir)
         return executor
 
+    @retry(180, errors=LockError, timeout=1)
     def init_git(
         self,
+        repo: "Repo",
         scm: "Git",
         stash_rev: str,
         entry: "ExpStashEntry",
@@ -179,25 +212,26 @@ class WorkspaceExecutor(BaseLocalExecutor):
         if infofile:
             self.info.dump_json(infofile)
 
-        scm.set_ref(EXEC_HEAD, entry.head_rev)
-        scm.set_ref(EXEC_MERGE, stash_rev)
-        scm.set_ref(EXEC_BASELINE, entry.baseline_rev)
-        self._detach_stack.enter_context(
-            self.scm.detach_head(
-                self.scm.get_ref(EXEC_HEAD),
-                force=True,
-                client="dvc",
+        with get_exp_rwlock(repo, writes=[EXEC_NAMESPACE]):
+            scm.set_ref(EXEC_HEAD, entry.head_rev)
+            scm.set_ref(EXEC_MERGE, stash_rev)
+            scm.set_ref(EXEC_BASELINE, entry.baseline_rev)
+            self._detach_stack.enter_context(
+                self.scm.detach_head(
+                    self.scm.get_ref(EXEC_HEAD),
+                    force=True,
+                    client="dvc",
+                )
             )
-        )
-        merge_rev = self.scm.get_ref(EXEC_MERGE)
-        try:
-            self.scm.merge(merge_rev, squash=True, commit=False)
-        except _SCMError as exc:
-            raise GitMergeError(str(exc), scm=self.scm)
-        if branch:
-            self.scm.set_ref(EXEC_BRANCH, branch, symbolic=True)
-        elif scm.get_ref(EXEC_BRANCH):
-            self.scm.remove_ref(EXEC_BRANCH)
+            merge_rev = self.scm.get_ref(EXEC_MERGE)
+            try:
+                self.scm.merge(merge_rev, squash=True, commit=False)
+            except _SCMError as exc:
+                raise GitMergeError(str(exc), scm=self.scm)
+            if branch:
+                self.scm.set_ref(EXEC_BRANCH, branch, symbolic=True)
+            elif scm.get_ref(EXEC_BRANCH):
+                self.scm.remove_ref(EXEC_BRANCH)
 
     def init_cache(self, repo: "Repo", rev: str, run_cache: bool = True):
         pass

dvc/repo/experiments/executor/ssh.py

@@ -9,6 +9,7 @@ from dvc_ssh import SSHFileSystem
 from funcy import first
 
 from ..refs import (
+    EXEC_BASELINE,
     EXEC_BRANCH,
     EXEC_CHECKPOINT,
     EXEC_HEAD,
@@ -119,6 +120,7 @@ class SSHExecutor(BaseExecutor):
 
     def init_git(
         self,
+        repo: "Repo",
         scm: "Git",
         stash_rev: str,
         entry: "ExpStashEntry",
@@ -147,12 +149,18 @@ class SSHExecutor(BaseExecutor):
             # (see https://github.com/iterative/dvc/issues/6508)
             kwargs = self._git_client_args(fs)
 
-            with self.set_exec_refs(scm, stash_rev, entry):
-                refspec = f"{EXEC_NAMESPACE}/"
-                push_refspec(scm, self.git_url, refspec, refspec, **kwargs)
+            ref_dict = {
+                EXEC_HEAD: entry.head_rev,
+                EXEC_MERGE: stash_rev,
+                EXEC_BASELINE: entry.baseline_rev,
+            }
+            with self.set_temp_refs(scm, ref_dict):
+                exec_namespace = f"{EXEC_NAMESPACE}/"
+                refspec = [(exec_namespace, exec_namespace)]
+                push_refspec(scm, self.git_url, refspec, **kwargs)
 
             if branch:
-                push_refspec(scm, self.git_url, branch, branch, **kwargs)
+                push_refspec(scm, self.git_url, [(branch, branch)], **kwargs)
                 self._ssh_cmd(fs, f"git symbolic-ref {EXEC_BRANCH} {branch}")
             else:
                 self._ssh_cmd(

dvc/repo/experiments/init.py

@@ -248,11 +248,6 @@ def init(
         except MissingParamsFile:
             pass
 
-    models = context.get("models")
-    live_path = context.pop("live", None)
-    live_metrics = f"{live_path}.json" if live_path else None
-    live_plots = os.path.join(live_path, "scalars") if live_path else None
-
     if type == "checkpoint":
         outs_key = "checkpoints"
         metrics_key = "metrics_persist_no_cache"
@@ -262,6 +257,13 @@ def init(
         metrics_key = "metrics_no_cache"
         plots_key = "plots_no_cache"
 
+    models = [context.get("models")]
+    metrics = [context.get("metrics")]
+    plots = [context.get("plots")]
+    if live_path := context.pop("live", None):
+        metrics.append(os.path.join(live_path, "metrics.json"))
+        plots.append(os.path.join(live_path, "plots"))
+
     stage = repo.stage.create(
         name=name,
         cmd=context["cmd"],
@@ -269,9 +271,9 @@ def init(
         params=[{params: None}] if params else None,
         force=force,
         **{
-            outs_key: compact([models]),
-            metrics_key: compact([context.get("metrics"), live_metrics]),
-            plots_key: compact([context.get("plots"), live_plots]),
+            outs_key: compact(models),
+            metrics_key: compact(metrics),
+            plots_key: compact(plots),
         },
     )
 

dvc/repo/experiments/ls.py

@@ -6,7 +6,7 @@ from dvc.repo.scm_context import scm_context
 from dvc.scm import iter_revs
 from dvc.types import Optional
 
-from .utils import exp_refs, exp_refs_by_baseline
+from .utils import exp_refs_by_baseline
 
 logger = logging.getLogger(__name__)
 
@@ -20,19 +20,22 @@ def ls(
     num: int = 1,
     git_remote: Optional[str] = None,
 ):
-    results = defaultdict(list)
-    if all_commits:
-        gen = exp_refs(repo.scm, git_remote)
-        for info in gen:
-            results[info.baseline_sha].append(info.name)
-        return results
-
-    rev = rev or "HEAD"
-
-    revs = iter_revs(repo.scm, [rev], num)
-    rev_set = set(revs.keys())
+    rev_set = None
+    if not all_commits:
+        revs = iter_revs(repo.scm, [rev or "HEAD"], num)
+        rev_set = set(revs.keys())
     ref_info_dict = exp_refs_by_baseline(repo.scm, rev_set, git_remote)
-    for rev, ref_info_list in ref_info_dict.items():
-        results[rev] = [ref_info.name for ref_info in ref_info_list]
+
+    tags = repo.scm.describe(ref_info_dict.keys())
+    remained = {baseline for baseline, tag in tags.items() if tag is None}
+    base = "refs/heads"
+    ref_heads = repo.scm.describe(remained, base=base)
+
+    results = defaultdict(list)
+    for baseline in ref_info_dict:
+        name = baseline[:7]
+        if tags[baseline] or ref_heads[baseline]:
+            name = tags[baseline] or ref_heads[baseline][len(base) + 1 :]
+        results[name] = [info.name for info in ref_info_dict[baseline]]
 
     return results

dvc/repo/experiments/queue/base.py

@@ -1,6 +1,5 @@
 import logging
 import os
-import sys
 from abc import ABC, abstractmethod
 from dataclasses import asdict, dataclass
 from typing import (
@@ -21,22 +20,24 @@ from typing import (
 from funcy import cached_property, retry
 
 from dvc.dependency import ParamsDependency
-from dvc.env import DVCLIVE_RESUME
+from dvc.env import DVC_EXP_BASELINE_REV, DVC_EXP_NAME, DVCLIVE_RESUME
 from dvc.exceptions import DvcException
 from dvc.lock import LockError
 from dvc.ui import ui
 
 from ..exceptions import CheckpointExistsError, ExperimentExistsError
-from ..executor.base import (
-    EXEC_PID_DIR,
-    EXEC_TMP_DIR,
-    BaseExecutor,
-    ExecutorResult,
-)
+from ..executor.base import BaseExecutor, ExecutorResult
 from ..executor.local import WorkspaceExecutor
 from ..refs import ExpRefInfo
 from ..stash import ExpStash, ExpStashEntry
-from ..utils import exp_refs_by_rev, scm_locked
+from ..utils import (
+    EXEC_PID_DIR,
+    EXEC_TMP_DIR,
+    exp_refs_by_rev,
+    get_exp_rwlock,
+    get_random_exp_name,
+)
+from .utils import get_remote_executor_refs
 
 if TYPE_CHECKING:
     from scmrepo.git import Git
@@ -85,6 +86,11 @@ class QueueDoneResult(NamedTuple):
     result: Optional[ExecutorResult]
 
 
+class ExpRefAndQueueEntry(NamedTuple):
+    exp_ref_info: Optional["ExpRefInfo"]
+    queue_entry: Optional["QueueEntry"]
+
+
 class BaseStashQueue(ABC):
     """Naive Git-stash based experiment queue.
 
@@ -269,6 +275,7 @@ class BaseStashQueue(ABC):
                 finish any active experiments before shutting down.
         """
 
+    @abstractmethod
     def logs(
         self,
         rev: str,
@@ -285,7 +292,7 @@ class BaseStashQueue(ABC):
                 output.
         """
 
-    def _stash_exp(
+    def _stash_exp(  # noqa: C901
         self,
         *args,
         params: Optional[Dict[str, List[str]]] = None,
@@ -400,7 +407,14 @@ class BaseStashQueue(ABC):
                             )
 
                     # save additional repro command line arguments
-                    run_env = {DVCLIVE_RESUME: "1"} if resume_rev else {}
+                    run_env = {
+                        DVC_EXP_BASELINE_REV: baseline_rev,
+                    }
+                    if not name:
+                        name = get_random_exp_name(self.scm, baseline_rev)
+                    run_env[DVC_EXP_NAME] = name
+                    if resume_rev:
+                        run_env[DVCLIVE_RESUME] = "1"
                     self._pack_args(*args, run_env=run_env, **kwargs)
 
                     # save experiment as a stash commit
@@ -525,16 +539,9 @@ class BaseStashQueue(ABC):
         .. _Hydra Override:
             https://hydra.cc/docs/advanced/override_grammar/basic/
         """
-        logger.debug("Using experiment params '%s'", params)
+        from dvc.utils.hydra import apply_overrides, compose_and_dump
 
-        try:
-            from dvc.utils.hydra import apply_overrides, compose_and_dump
-        except ValueError:
-            if sys.version_info >= (3, 11):
-                raise DvcException(
-                    "--set-param is not supported in Python >= 3.11"
-                )
-            raise
+        logger.debug("Using experiment params '%s'", params)
 
         hydra_config = self.repo.config.get("hydra", {})
         hydra_enabled = hydra_config.get("enabled", False)
@@ -562,28 +569,39 @@ class BaseStashQueue(ABC):
 
     @staticmethod
     @retry(180, errors=LockError, timeout=1)
-    @scm_locked
+    def get_stash_entry(
+        exp: "Experiments",
+        queue_entry: QueueEntry,
+    ) -> "ExpStashEntry":
+        stash = ExpStash(exp.scm, queue_entry.stash_ref)
+        stash_rev = queue_entry.stash_rev
+        with get_exp_rwlock(exp.repo, writes=[queue_entry.stash_ref]):
+            stash_entry = stash.stash_revs.get(
+                stash_rev,
+                ExpStashEntry(None, stash_rev, stash_rev, None, None),
+            )
+            if stash_entry.stash_index is not None:
+                stash.drop(stash_entry.stash_index)
+        return stash_entry
+
+    @classmethod
     def init_executor(
+        cls,
         exp: "Experiments",
         queue_entry: QueueEntry,
         executor_cls: Type[BaseExecutor] = WorkspaceExecutor,
         **kwargs,
     ) -> BaseExecutor:
-        scm = exp.scm
-        stash = ExpStash(scm, queue_entry.stash_ref)
-        stash_rev = queue_entry.stash_rev
-        stash_entry = stash.stash_revs.get(
-            stash_rev,
-            ExpStashEntry(None, stash_rev, stash_rev, None, None),
-        )
-        if stash_entry.stash_index is not None:
-            stash.drop(stash_entry.stash_index)
+        stash_entry = cls.get_stash_entry(exp, queue_entry)
+
         executor = executor_cls.from_stash_entry(
             exp.repo, stash_entry, **kwargs
         )
 
+        stash_rev = queue_entry.stash_rev
         infofile = exp.celery_queue.get_infofile_path(stash_rev)
         executor.init_git(
+            exp.repo,
             exp.repo.scm,
             stash_rev,
             stash_entry,
@@ -604,7 +622,6 @@ class BaseStashQueue(ABC):
 
     @staticmethod
     @retry(180, errors=LockError, timeout=1)
-    @scm_locked
     def collect_git(
         exp: "Experiments",
         executor: BaseExecutor,
@@ -618,16 +635,20 @@ class BaseStashQueue(ABC):
                 raise CheckpointExistsError(ref_info.name)
             raise ExperimentExistsError(ref_info.name)
 
-        for ref in executor.fetch_exps(
-            exp.scm,
-            force=exec_result.force,
-            on_diverged=on_diverged,
-        ):
-            exp_rev = exp.scm.get_ref(ref)
-            if exp_rev:
-                assert exec_result.exp_hash
-                logger.debug("Collected experiment '%s'.", exp_rev[:7])
-                results[exp_rev] = exec_result.exp_hash
+        refs = get_remote_executor_refs(exp.scm, executor.git_url)
+
+        with get_exp_rwlock(exp.repo, writes=refs):
+            for ref in executor.fetch_exps(
+                exp.scm,
+                refs,
+                force=exec_result.force,
+                on_diverged=on_diverged,
+            ):
+                exp_rev = exp.scm.get_ref(ref)
+                if exp_rev:
+                    assert exec_result.exp_hash
+                    logger.debug("Collected experiment '%s'.", exp_rev[:7])
+                    results[exp_rev] = exec_result.exp_hash
 
         return results
 
@@ -706,3 +727,11 @@ class BaseStashQueue(ABC):
                 entry.stash_rev,
                 message=f"commit: {msg}",
             )
+
+    @abstractmethod
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Dict]:
+        """Get the execution info of the currently running experiments
+
+        Args:
+            fetch_ref (bool): fetch completed checkpoints or not.
+        """

dvc/repo/experiments/queue/celery.py

@@ -12,10 +12,11 @@ from typing import (
     NamedTuple,
     Optional,
     Set,
+    Union,
 )
 
 from celery.result import AsyncResult
-from funcy import cached_property
+from funcy import cached_property, first
 from kombu.message import Message
 
 from dvc.daemon import daemonize
@@ -23,20 +24,27 @@ from dvc.exceptions import DvcException
 from dvc.ui import ui
 
 from ..exceptions import UnresolvedQueueExpNamesError
-from ..executor.base import (
-    EXEC_TMP_DIR,
-    ExecutorInfo,
-    ExecutorResult,
-    TaskStatus,
+from ..executor.base import ExecutorInfo, ExecutorResult
+from ..refs import CELERY_STASH
+from ..utils import EXEC_TMP_DIR, get_exp_rwlock
+from .base import (
+    BaseStashQueue,
+    ExpRefAndQueueEntry,
+    QueueDoneResult,
+    QueueEntry,
+    QueueGetResult,
 )
-from .base import BaseStashQueue, QueueDoneResult, QueueEntry, QueueGetResult
+from .exceptions import CannotKillTasksError
 from .tasks import run_exp
+from .utils import fetch_running_exp_from_temp_dir
 
 if TYPE_CHECKING:
     from dvc_task.app import FSApp
     from dvc_task.proc.manager import ProcessManager
     from dvc_task.worker import TemporaryWorker
 
+    from ..refs import ExpRefInfo
+
 logger = logging.getLogger(__name__)
 
 
@@ -58,8 +66,6 @@ class LocalCeleryQueue(BaseStashQueue):
 
     CELERY_DIR = "celery"
 
-    _shutdown_task_ids: Set[str] = set()
-
     @cached_property
     def wdir(self) -> str:
         return os.path.join(self.repo.tmp_dir, EXEC_TMP_DIR, self.CELERY_DIR)
@@ -117,7 +123,7 @@ class LocalCeleryQueue(BaseStashQueue):
             timeout=10,
         )
 
-    def spawn_worker(self, num: int = 1):
+    def _spawn_worker(self, num: int = 1):
         """spawn one single worker to process to queued tasks.
 
         Argument:
@@ -160,14 +166,15 @@ class LocalCeleryQueue(BaseStashQueue):
             if node_name in active_worker:
                 logger.debug(f"Exp queue worker {node_name} already exist")
                 continue
-            self.spawn_worker(num)
+            self._spawn_worker(num)
             started += 1
 
         return started
 
     def put(self, *args, **kwargs) -> QueueEntry:
         """Stash an experiment and add it to the queue."""
-        entry = self._stash_exp(*args, **kwargs)
+        with get_exp_rwlock(self.repo, writes=["workspace", CELERY_STASH]):
+            entry = self._stash_exp(*args, **kwargs)
         self.celery.signature(run_exp.s(entry.asdict())).delay()
         return entry
 
@@ -186,6 +193,7 @@ class LocalCeleryQueue(BaseStashQueue):
                 continue
             args, kwargs, _embed = msg.decode()
             entry_dict = kwargs.get("entry_dict", args[0])
+            logger.debug("Found queued task %s", entry_dict["stash_rev"])
             yield _MessageEntry(msg, QueueEntry.from_dict(entry_dict))
 
     def _iter_processed(self) -> Generator[_MessageEntry, None, None]:
@@ -202,6 +210,7 @@ class LocalCeleryQueue(BaseStashQueue):
             task_id = msg.headers["id"]
             result: AsyncResult = AsyncResult(task_id)
             if not result.ready():
+                logger.debug("Found active task %s", entry.stash_rev)
                 yield _TaskEntry(result, entry)
 
     def _iter_done_tasks(self) -> Generator[_TaskEntry, None, None]:
@@ -210,6 +219,7 @@ class LocalCeleryQueue(BaseStashQueue):
             task_id = msg.headers["id"]
             result: AsyncResult = AsyncResult(task_id)
             if result.ready():
+                logger.debug("Found done task %s", entry.stash_rev)
                 yield _TaskEntry(result, entry)
 
     def iter_active(self) -> Generator[QueueEntry, None, None]:
@@ -242,76 +252,136 @@ class LocalCeleryQueue(BaseStashQueue):
     def reproduce(self) -> Mapping[str, Mapping[str, str]]:
         raise NotImplementedError
 
-    def get_result(
+    def _load_info(self, rev: str) -> ExecutorInfo:
+        infofile = self.get_infofile_path(rev)
+        return ExecutorInfo.load_json(infofile)
+
+    def _get_done_result(
         self, entry: QueueEntry, timeout: Optional[float] = None
-    ) -> Optional[ExecutorResult]:
+    ) -> Optional["ExecutorResult"]:
         from celery.exceptions import TimeoutError as _CeleryTimeout
 
-        def _load_info(rev: str) -> ExecutorInfo:
-            infofile = self.get_infofile_path(rev)
-            return ExecutorInfo.load_json(infofile)
-
-        def _load_collected(rev: str) -> Optional[ExecutorResult]:
-            executor_info = _load_info(rev)
-            if executor_info.status > TaskStatus.SUCCESS:
+        for msg, processed_entry in self._iter_processed():
+            if entry.stash_rev == processed_entry.stash_rev:
+                task_id = msg.headers["id"]
+                result: AsyncResult = AsyncResult(task_id)
+                if not result.ready():
+                    logger.debug(
+                        "Waiting for exp task '%s' to complete", result.id
+                    )
+                    try:
+                        result.get(timeout=timeout)
+                    except _CeleryTimeout as exc:
+                        raise DvcException(
+                            "Timed out waiting for exp to finish."
+                        ) from exc
+                executor_info = self._load_info(entry.stash_rev)
                 return executor_info.result
-            raise FileNotFoundError
+        raise FileNotFoundError
+
+    def get_result(
+        self, entry: QueueEntry, timeout: Optional[float] = None
+    ) -> Optional["ExecutorResult"]:
 
         try:
-            return _load_collected(entry.stash_rev)
+            return self._get_done_result(entry, timeout)
         except FileNotFoundError:
-            # Infofile will not be created until execution begins
             pass
 
         for queue_entry in self.iter_queued():
             if entry.stash_rev == queue_entry.stash_rev:
                 raise DvcException("Experiment has not been started.")
-        for result, active_entry in self._iter_active_tasks():
-            if entry.stash_rev == active_entry.stash_rev:
-                logger.debug(
-                    "Waiting for exp task '%s' to complete", result.id
-                )
-                try:
-                    result.get(timeout=timeout)
-                except _CeleryTimeout as exc:
-                    raise DvcException(
-                        "Timed out waiting for exp to finish."
-                    ) from exc
-                executor_info = _load_info(entry.stash_rev)
-                return executor_info.result
 
         # NOTE: It's possible for an exp to complete while iterating through
         # other queued and active tasks, in which case the exp will get moved
         # out of the active task list, and needs to be loaded here.
-        return _load_collected(entry.stash_rev)
+        return self._get_done_result(entry, timeout)
 
-    def kill(self, revs: Collection[str]) -> None:
-        to_kill: Set[QueueEntry] = set()
+    def _get_running_task_ids(self) -> Set[str]:
+        running_task_ids: Set[str] = set()
+        active_workers = self.worker_status()
+        for _, tasks in active_workers.items():
+            task = first(tasks)
+            if task:
+                running_task_ids.add(task["id"])
+        return running_task_ids
+
+    def _try_to_kill_tasks(
+        self, to_kill: Dict[QueueEntry, str], force: bool
+    ) -> Dict[QueueEntry, str]:
+        fail_to_kill_entries: Dict[QueueEntry, str] = {}
+        for queue_entry, rev in to_kill.items():
+            try:
+                if force:
+                    self.proc.kill(queue_entry.stash_rev)
+                else:
+                    self.proc.interrupt(queue_entry.stash_rev)
+                logger.debug(f"Task {rev} had been killed.")
+            except ProcessLookupError:
+                fail_to_kill_entries[queue_entry] = rev
+        return fail_to_kill_entries
+
+    def _mark_inactive_tasks_failure(self, remained_entries):
+        remained_revs: List[str] = []
+        running_ids = self._get_running_task_ids()
+        logger.debug(f"Current running tasks ids: {running_ids}.")
+        for msg, entry in self._iter_processed():
+            if entry not in remained_entries:
+                continue
+            task_id = msg.headers["id"]
+            if task_id in running_ids:
+                remained_revs.append(remained_entries[entry])
+            else:
+                result: AsyncResult = AsyncResult(task_id)
+                if not result.ready():
+                    logger.debug(
+                        f"Task id {task_id} rev {remained_entries[entry]} "
+                        "marked as failure."
+                    )
+                    self.celery.backend.mark_as_failure(task_id, None)
+
+        if remained_revs:
+            raise CannotKillTasksError(remained_revs)
+
+    def _kill_entries(self, entries: Dict[QueueEntry, str], force: bool):
+        logger.debug(
+            "Found active tasks: '%s' to kill",
+            list(entries.values()),
+        )
+        inactive_entries: Dict[QueueEntry, str] = self._try_to_kill_tasks(
+            entries, force
+        )
+
+        if inactive_entries:
+            self._mark_inactive_tasks_failure(inactive_entries)
+
+    def kill(self, revs: Collection[str], force: bool = False) -> None:
         name_dict: Dict[
             str, Optional[QueueEntry]
         ] = self.match_queue_entry_by_name(set(revs), self.iter_active())
 
-        missing_rev: List[str] = []
+        missing_revs: List[str] = []
+        to_kill: Dict[QueueEntry, str] = {}
         for rev, queue_entry in name_dict.items():
             if queue_entry is None:
-                missing_rev.append(rev)
+                missing_revs.append(rev)
             else:
-                to_kill.add(queue_entry)
+                to_kill[queue_entry] = rev
 
-        if missing_rev:
-            raise UnresolvedQueueExpNamesError(missing_rev)
+        if missing_revs:
+            raise UnresolvedQueueExpNamesError(missing_revs)
 
-        for queue_entry in to_kill:
-            self.proc.kill(queue_entry.stash_rev)
+        if to_kill:
+            self._kill_entries(to_kill, force)
 
     def shutdown(self, kill: bool = False):
         self.celery.control.shutdown()
         if kill:
-            for _, task_entry in self._iter_active_tasks():
-                try:
-                    self.proc.kill(task_entry.stash_rev)
-                except ProcessLookupError:
-                    continue
+            to_kill: Dict[QueueEntry, str] = {}
+            for entry in self.iter_active():
+                to_kill[entry] = entry.name or entry.stash_rev
+            if to_kill:
+                self._kill_entries(to_kill, True)
 
     def follow(
         self,
@@ -358,10 +428,11 @@ class LocalCeleryQueue(BaseStashQueue):
         ) as fobj:
             ui.write(fobj.read())
 
-    def worker_status(self) -> Dict:
+    def worker_status(self) -> Dict[str, List[Dict]]:
         """Return the current active celery worker"""
         status = self.celery.control.inspect().active() or {}
         logger.debug(f"Worker status: {status}")
+
         return status
 
     def clear(self, *args, **kwargs):
@@ -373,3 +444,46 @@ class LocalCeleryQueue(BaseStashQueue):
         from .remove import celery_remove
 
         return celery_remove(self, *args, **kwargs)
+
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Dict]:
+        """Get the execution info of the currently running experiments
+
+        Args:
+            fetch_ref (bool): fetch completed checkpoints or not.
+        """
+        result: Dict[str, Dict] = {}
+        for entry in self.iter_active():
+            result.update(
+                fetch_running_exp_from_temp_dir(
+                    self, entry.stash_rev, fetch_refs
+                )
+            )
+        return result
+
+    def get_ref_and_entry_by_names(
+        self,
+        exp_names: Union[str, List[str]],
+        git_remote: Optional[str] = None,
+    ) -> Dict[str, ExpRefAndQueueEntry]:
+        """Find finished ExpRefInfo or queued or failed QueueEntry by name"""
+        from ..utils import resolve_name
+
+        if isinstance(exp_names, str):
+            exp_names = [exp_names]
+        results: Dict[str, ExpRefAndQueueEntry] = {}
+
+        exp_ref_match: Dict[str, Optional["ExpRefInfo"]] = resolve_name(
+            self.scm, exp_names, git_remote
+        )
+        if not git_remote:
+            queue_entry_match: Dict[
+                str, Optional["QueueEntry"]
+            ] = self.match_queue_entry_by_name(
+                exp_names, self.iter_queued(), self.iter_done()
+            )
+
+        for exp_name in exp_names:
+            exp_ref = exp_ref_match[exp_name]
+            queue_entry = None if git_remote else queue_entry_match[exp_name]
+            results[exp_name] = ExpRefAndQueueEntry(exp_ref, queue_entry)
+        return results

dvc/repo/experiments/queue/exceptions.py

@@ -0,0 +1,12 @@
+from typing import Collection
+
+from dvc.exceptions import DvcException
+
+
+class CannotKillTasksError(DvcException):
+    def __init__(self, revs: Collection[str]):
+        rev_str = ",".join(revs)
+        super().__init__(
+            f"Task '{rev_str}' is initializing, please wait a few seconds "
+            "until the experiments start running to retry the kill operation."
+        )

dvc/repo/experiments/queue/tasks.py

@@ -1,4 +1,4 @@
-from typing import Any, Dict
+from typing import TYPE_CHECKING, Any, Dict
 
 from celery import shared_task
 from celery.utils.log import get_task_logger
@@ -7,11 +7,15 @@ from ..executor.base import ExecutorInfo
 from ..executor.local import TempDirExecutor
 from .base import BaseStashQueue, QueueEntry
 
+if TYPE_CHECKING:
+    from ..executor.base import BaseExecutor
+
+
 logger = get_task_logger(__name__)
 
 
 @shared_task
-def setup_exp(entry_dict: Dict[str, Any]) -> TempDirExecutor:
+def setup_exp(entry_dict: Dict[str, Any]) -> "BaseExecutor":
     """Setup an experiment.
 
     Arguments:

dvc/repo/experiments/queue/tempdir.py

@@ -1,20 +1,23 @@
 import logging
 import os
+from collections import defaultdict
 from typing import TYPE_CHECKING, Dict, Generator, Optional
 
 from funcy import cached_property, first
 
+from dvc.exceptions import DvcException
+
 from ..exceptions import ExpQueueEmptyError
 from ..executor.base import (
-    EXEC_PID_DIR,
-    EXEC_TMP_DIR,
     BaseExecutor,
     ExecutorInfo,
     ExecutorResult,
     TaskStatus,
 )
 from ..executor.local import TempDirExecutor
+from ..utils import EXEC_PID_DIR, EXEC_TMP_DIR
 from .base import BaseStashQueue, QueueEntry, QueueGetResult
+from .utils import fetch_running_exp_from_temp_dir
 from .workspace import WorkspaceQueue
 
 if TYPE_CHECKING:
@@ -76,9 +79,8 @@ class TempDirQueue(WorkspaceQueue):
         for stash_rev in self.proc:
             infofile = self.get_infofile_path(stash_rev)
             executor_info = ExecutorInfo.load_json(infofile)
-            if (
-                not executor_info.status <= TaskStatus.SUCCESS
-                and os.path.exists(executor_info.root_dir)
+            if executor_info.status <= TaskStatus.SUCCESS and os.path.exists(
+                executor_info.root_dir
             ):
                 yield QueueEntry(
                     self.repo.root_dir,
@@ -91,6 +93,51 @@ class TempDirQueue(WorkspaceQueue):
                     None,
                 )
 
+    def _reproduce_entry(
+        self, entry: QueueEntry, executor: BaseExecutor
+    ) -> Dict[str, Dict[str, str]]:
+        from dvc.stage.monitor import CheckpointKilledError
+
+        results: Dict[str, Dict[str, str]] = defaultdict(dict)
+        exec_name = self._EXEC_NAME or entry.stash_rev
+        infofile = self.get_infofile_path(exec_name)
+        try:
+            rev = entry.stash_rev
+            exec_result = executor.reproduce(
+                info=executor.info,
+                rev=rev,
+                infofile=infofile,
+                log_level=logger.getEffectiveLevel(),
+                log_errors=True,
+            )
+            if not exec_result.exp_hash:
+                raise DvcException(
+                    f"Failed to reproduce experiment '{rev[:7]}'"
+                )
+            if exec_result.ref_info:
+                results[rev].update(
+                    self.collect_executor(
+                        self.repo.experiments, executor, exec_result
+                    )
+                )
+        except CheckpointKilledError:
+            results[rev].update(
+                self.collect_executor(
+                    self.repo.experiments, executor, exec_result
+                )
+            )
+
+            return results
+        except DvcException:
+            raise
+        except Exception as exc:
+            raise DvcException(
+                f"Failed to reproduce experiment '{rev[:7]}'"
+            ) from exc
+        finally:
+            executor.cleanup(infofile)
+        return results
+
     @staticmethod
     def collect_executor(
         exp: "Experiments",
@@ -98,3 +145,13 @@ class TempDirQueue(WorkspaceQueue):
         exec_result: ExecutorResult,
     ) -> Dict[str, str]:
         return BaseStashQueue.collect_executor(exp, executor, exec_result)
+
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Dict]:
+        result: Dict[str, Dict] = {}
+        for entry in self.iter_active():
+            result.update(
+                fetch_running_exp_from_temp_dir(
+                    self, entry.stash_rev, fetch_refs
+                )
+            )
+        return result

dvc/repo/experiments/queue/utils.py

@@ -0,0 +1,86 @@
+import logging
+from typing import TYPE_CHECKING, Dict, List
+
+from scmrepo.exceptions import SCMError
+
+from ..executor.base import ExecutorInfo, TaskStatus
+from ..refs import EXEC_CHECKPOINT, EXEC_NAMESPACE, EXPS_NAMESPACE, EXPS_STASH
+from ..utils import get_exp_rwlock, iter_remote_refs
+
+logger = logging.getLogger(__name__)
+
+
+if TYPE_CHECKING:
+    from scmrepo.git import Git
+
+    from .base import BaseStashQueue
+
+
+def get_remote_executor_refs(scm: "Git", remote_url: str) -> List[str]:
+    """Get result list refs from a remote repository
+
+    Args:
+        remote_url : remote executor's url
+    """
+    refs = []
+    for ref in iter_remote_refs(
+        scm,
+        remote_url,
+        base=EXPS_NAMESPACE,
+    ):
+        if ref == EXEC_CHECKPOINT or (
+            not ref.startswith(EXEC_NAMESPACE) and ref != EXPS_STASH
+        ):
+            refs.append(ref)
+    return refs
+
+
+def fetch_running_exp_from_temp_dir(
+    queue: "BaseStashQueue", rev: str, fetch_refs: bool
+) -> Dict[str, Dict]:
+    """Fetch status of running exps out of current working directory
+
+    Args:
+        queue (BaseStashQueue):
+        rev (str): stash revision of the experiment
+        fetch_refs (bool): fetch running checkpoint results to local or not.
+
+    Returns:
+        Dict[str, Dict]: _description_
+    """
+    from dvc.scm import InvalidRemoteSCMRepo
+    from dvc.utils.serialize import load_json
+
+    from ..executor.local import TempDirExecutor
+
+    result: Dict[str, Dict] = {}
+    infofile = queue.get_infofile_path(rev)
+    try:
+        info = ExecutorInfo.from_dict(load_json(infofile))
+    except OSError:
+        return result
+    if info.status <= TaskStatus.RUNNING:
+        result[rev] = info.asdict()
+        if info.git_url and fetch_refs and info.status > TaskStatus.PREPARING:
+
+            def on_diverged(_ref: str, _checkpoint: bool):
+                return True
+
+            executor = TempDirExecutor.from_info(info)
+            try:
+                refs = get_remote_executor_refs(queue.scm, executor.git_url)
+                with get_exp_rwlock(queue.repo, writes=refs):
+                    for ref in executor.fetch_exps(
+                        queue.scm,
+                        refs,
+                        on_diverged=on_diverged,
+                    ):
+                        logger.debug("Updated running experiment '%s'.", ref)
+                        last_rev = queue.scm.get_ref(ref)
+                        result[rev]["last"] = last_rev
+                        if last_rev:
+                            result[last_rev] = info.asdict()
+            except (InvalidRemoteSCMRepo, SCMError):
+                # ignore stale info files
+                del result[rev]
+    return result

dvc/repo/experiments/queue/workspace.py

@@ -1,15 +1,28 @@
+import json
 import logging
 from collections import defaultdict
-from typing import TYPE_CHECKING, Collection, Dict, Generator, Optional
+from typing import TYPE_CHECKING, Collection, Dict, Generator, List, Optional
 
+import psutil
 from funcy import first
+from voluptuous import Invalid
 
 from dvc.exceptions import DvcException
+from dvc.lock import make_lock
+from dvc.rwlock import RWLOCK_FILE, RWLOCK_LOCK, SCHEMA
+from dvc.utils import relpath
+from dvc.utils.fs import remove
 
 from ..exceptions import ExpQueueEmptyError
-from ..executor.base import BaseExecutor, ExecutorResult
+from ..executor.base import (
+    BaseExecutor,
+    ExecutorInfo,
+    ExecutorResult,
+    TaskStatus,
+)
 from ..executor.local import WorkspaceExecutor
-from ..refs import EXEC_BRANCH
+from ..refs import EXEC_BRANCH, WORKSPACE_STASH
+from ..utils import get_exp_rwlock, to_studio_params
 from .base import BaseStashQueue, QueueDoneResult, QueueEntry, QueueGetResult
 
 if TYPE_CHECKING:
@@ -22,7 +35,8 @@ class WorkspaceQueue(BaseStashQueue):
     _EXEC_NAME: Optional[str] = "workspace"
 
     def put(self, *args, **kwargs) -> QueueEntry:
-        return self._stash_exp(*args, **kwargs)
+        with get_exp_rwlock(self.repo, writes=["workspace", WORKSPACE_STASH]):
+            return self._stash_exp(*args, **kwargs)
 
     def get(self) -> QueueGetResult:
         revs = self.stash.stash_revs
@@ -82,6 +96,8 @@ class WorkspaceQueue(BaseStashQueue):
     def _reproduce_entry(
         self, entry: QueueEntry, executor: BaseExecutor
     ) -> Dict[str, Dict[str, str]]:
+        from dvc_studio_client.post_live_metrics import post_live_metrics
+
         from dvc.stage.monitor import CheckpointKilledError
 
         results: Dict[str, Dict[str, str]] = defaultdict(dict)
@@ -89,6 +105,13 @@ class WorkspaceQueue(BaseStashQueue):
         infofile = self.get_infofile_path(exec_name)
         try:
             rev = entry.stash_rev
+            post_live_metrics(
+                "start",
+                executor.info.baseline_rev,
+                executor.info.name,
+                "dvc",
+                params=to_studio_params(self.repo.params.show()),
+            )
             exec_result = executor.reproduce(
                 info=executor.info,
                 rev=rev,
@@ -116,6 +139,13 @@ class WorkspaceQueue(BaseStashQueue):
                 f"Failed to reproduce experiment '{rev[:7]}'"
             ) from exc
         finally:
+            post_live_metrics(
+                "done",
+                executor.info.baseline_rev,
+                executor.info.name,
+                "dvc",
+                experiment_rev=first(results.get(rev, {})),
+            )
             executor.cleanup(infofile)
         return results
 
@@ -150,3 +180,89 @@ class WorkspaceQueue(BaseStashQueue):
         follow: bool = False,
     ):
         raise NotImplementedError
+
+    def check_rwlock(
+        self,
+        hardlink: bool = False,
+        autocorrect: bool = False,
+    ) -> bool:
+        """Check and autocorrect the RWLock status for file paths.
+
+        Args:
+            hardlink (bool): use hardlink lock to guard rwlock file when on
+                            edit.
+            autocorrect (bool): autocorrect corrupted rwlock file.
+
+        Return:
+            (bool): if the pid alive.
+        """
+        path = self.repo.fs.path.join(self.repo.tmp_dir, RWLOCK_FILE)
+
+        rwlock_guard = make_lock(
+            self.repo.fs.path.join(self.repo.tmp_dir, RWLOCK_LOCK),
+            tmp_dir=self.repo.tmp_dir,
+            hardlink_lock=hardlink,
+        )
+        with rwlock_guard:
+            try:
+                with self.repo.fs.open(path, encoding="utf-8") as fobj:
+                    lock: Dict[str, List[Dict]] = SCHEMA(json.load(fobj))
+                file_path = first(lock["read"])
+                if not file_path:
+                    return False
+                lock_info = first(lock["read"][file_path])
+                pid = int(lock_info["pid"])
+                if psutil.pid_exists(pid):
+                    return True
+                cmd = lock_info["cmd"]
+                logger.warning(
+                    "Process '%s' with (Pid %s), in RWLock-file '%s'"
+                    " had been killed.",
+                    cmd,
+                    pid,
+                    relpath(path),
+                )
+            except FileNotFoundError:
+                return False
+            except json.JSONDecodeError:
+                logger.warning(
+                    "Unable to read RWLock-file '%s'. JSON structure is"
+                    " corrupted",
+                    relpath(path),
+                )
+            except Invalid:
+                logger.warning("RWLock-file '%s' format error.", relpath(path))
+            if autocorrect:
+                logger.warning(
+                    "Delete corrupted RWLock-file '%s'", relpath(path)
+                )
+                remove(path)
+            return False
+
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Dict]:
+        from dvc.utils.serialize import load_json
+
+        assert self._EXEC_NAME
+        result: Dict[str, Dict] = {}
+
+        if not self.check_rwlock(autocorrect=True):
+            return result
+
+        infofile = self.get_infofile_path(self._EXEC_NAME)
+
+        try:
+            info = ExecutorInfo.from_dict(load_json(infofile))
+        except OSError:
+            return result
+
+        if info.status < TaskStatus.FAILED:
+            # If we are appending to a checkpoint branch in a workspace
+            # run, show the latest checkpoint as running.
+            if info.status == TaskStatus.SUCCESS:
+                return result
+            last_rev = self.scm.get_ref(EXEC_BRANCH)
+            if last_rev:
+                result[last_rev] = info.asdict()
+            else:
+                result[self._EXEC_NAME] = info.asdict()
+        return result

dvc/repo/experiments/refs.py

@@ -16,6 +16,7 @@ EXEC_BRANCH = f"{EXEC_NAMESPACE}/EXEC_BRANCH"
 EXEC_BASELINE = f"{EXEC_NAMESPACE}/EXEC_BASELINE"
 EXEC_HEAD = f"{EXEC_NAMESPACE}/EXEC_HEAD"
 EXEC_MERGE = f"{EXEC_NAMESPACE}/EXEC_MERGE"
+EXPS_TEMP = f"{EXPS_NAMESPACE}/temp"
 STASHES = {WORKSPACE_STASH, CELERY_STASH}
 
 
@@ -23,9 +24,7 @@ class ExpRefInfo:
 
     namespace = EXPS_NAMESPACE
 
-    def __init__(
-        self, baseline_sha: Optional[str] = None, name: Optional[str] = None
-    ):
+    def __init__(self, baseline_sha: str, name: Optional[str] = None):
         self.baseline_sha = baseline_sha
         self.name: str = name if name else ""
 
@@ -33,7 +32,7 @@ class ExpRefInfo:
         return "/".join(self.parts)
 
     def __repr__(self):
-        baseline = f"'{self.baseline_sha}'" if self.baseline_sha else "None"
+        baseline = f"'{self.baseline_sha}'"
         name = f"'{self.name}'" if self.name else "None"
         return f"ExpRefInfo(baseline_sha={baseline}, name={name})"
 
@@ -41,11 +40,7 @@ class ExpRefInfo:
     def parts(self):
         return (
             (self.namespace,)
-            + (
-                (self.baseline_sha[:2], self.baseline_sha[2:])
-                if self.baseline_sha
-                else ()
-            )
+            + ((self.baseline_sha[:2], self.baseline_sha[2:]))
             + ((self.name,) if self.name else ())
         )
 
@@ -54,14 +49,13 @@ class ExpRefInfo:
         try:
             parts = ref.split("/")
             if (
-                len(parts) < 2
-                or len(parts) == 3
+                len(parts) < 4
                 or len(parts) > 5
                 or "/".join(parts[:2]) != EXPS_NAMESPACE
             ):
                 raise InvalidExpRefError(ref)
         except ValueError:
             raise InvalidExpRefError(ref)
-        baseline_sha = parts[2] + parts[3] if len(parts) >= 4 else None
+        baseline_sha = parts[2] + parts[3]
         name = parts[4] if len(parts) == 5 else None
         return cls(baseline_sha, name)

dvc/repo/experiments/remove.py

@@ -1,19 +1,12 @@
 import logging
-from typing import (
-    TYPE_CHECKING,
-    Dict,
-    Iterable,
-    List,
-    NamedTuple,
-    Optional,
-    Union,
-)
+from typing import TYPE_CHECKING, Dict, Iterable, List, Optional, Union
 
 from dvc.repo import locked
 from dvc.repo.scm_context import scm_context
 from dvc.scm import iter_revs
 
 from .exceptions import UnresolvedExpNamesError
+from .queue.base import QueueEntry
 from .refs import ExpRefInfo
 from .utils import exp_refs, exp_refs_by_baseline, push_refspec
 
@@ -22,58 +15,12 @@ if TYPE_CHECKING:
     from dvc.repo.experiments.queue.celery import LocalCeleryQueue
     from dvc.scm import Git
 
-    from .queue.base import QueueEntry
+    from .queue.base import ExpRefAndQueueEntry
 
 
 logger = logging.getLogger(__name__)
 
 
-class ExpRefAndQueueEntry(NamedTuple):
-    exp_ref_list: List["ExpRefInfo"]
-    queue_entry_list: List["QueueEntry"]
-    removed: List["str"]
-
-
-def _get_ref_and_entry_by_names(
-    exp_names: Union[str, List[str]],
-    scm: "Git",
-    celery_queue: "LocalCeleryQueue",
-    git_remote: Optional[str],
-) -> ExpRefAndQueueEntry:
-    from .utils import resolve_name
-
-    exp_ref_list: List["ExpRefInfo"] = []
-    queue_entry_list: List["QueueEntry"] = []
-    removed: List[str] = []
-    if isinstance(exp_names, str):
-        exp_names = [exp_names]
-    exp_ref_match: Dict[str, Optional["ExpRefInfo"]] = resolve_name(
-        scm, exp_names, git_remote
-    )
-    if not git_remote:
-        queue_entry_match: Dict[
-            str, Optional["QueueEntry"]
-        ] = celery_queue.match_queue_entry_by_name(
-            exp_names, celery_queue.iter_queued(), celery_queue.iter_done()
-        )
-
-    remained = []
-    for exp_name in exp_names:
-        exp_ref = exp_ref_match[exp_name]
-        queue_entry = None if git_remote else queue_entry_match[exp_name]
-        if exp_ref or queue_entry:
-            if exp_ref:
-                exp_ref_list.append(exp_ref)
-            if queue_entry:
-                queue_entry_list.append(queue_entry)
-            removed.append(exp_name)
-        else:
-            remained.append(exp_name)
-    if remained:
-        raise UnresolvedExpNamesError(remained)
-    return ExpRefAndQueueEntry(exp_ref_list, queue_entry_list, removed)
-
-
 @locked
 @scm_context
 def remove(
@@ -103,12 +50,22 @@ def remove(
     exp_ref_list: List["ExpRefInfo"] = []
     queue_entry_list: List["QueueEntry"] = []
     if exp_names:
-        result = _get_ref_and_entry_by_names(
-            exp_names, repo.scm, celery_queue, git_remote
-        )
-        removed.extend(result.removed)
-        exp_ref_list.extend(result.exp_ref_list)
-        queue_entry_list.extend(result.queue_entry_list)
+        results: Dict[
+            str, "ExpRefAndQueueEntry"
+        ] = celery_queue.get_ref_and_entry_by_names(exp_names, git_remote)
+        remained: List[str] = []
+        for name, result in results.items():
+            if not result.exp_ref_info and not result.queue_entry:
+                remained.append(name)
+                continue
+            removed.append(name)
+            if result.exp_ref_info:
+                exp_ref_list.append(result.exp_ref_info)
+            if result.queue_entry:
+                queue_entry_list.append(result.queue_entry)
+
+        if remained:
+            raise UnresolvedExpNamesError(remained)
     elif rev:
         exp_ref_dict = _resolve_exp_by_baseline(repo, rev, num, git_remote)
         removed.extend(exp_ref_dict.keys())
@@ -152,8 +109,7 @@ def _remove_commited_exps(
                 push_refspec(
                     scm,
                     remote,
-                    None,
-                    str(ref_info),
+                    [(None, str(ref_info))],
                     progress=pbar.update_git,
                 )
     else:

dvc/repo/experiments/run.py

@@ -28,14 +28,24 @@ def run(
     Returns a dict mapping new experiment SHAs to the results
     of `repro` for that experiment.
     """
-    from dvc.utils.hydra import to_hydra_overrides
-
     if run_all:
-        entries = list(repo.experiments.celery_queue.iter_queued())
-        return repo.experiments.reproduce_celery(entries, jobs=jobs)
+        return repo.experiments.reproduce_celery(jobs=jobs)
 
+    hydra_sweep = None
     if params:
+        from dvc.utils.hydra import to_hydra_overrides
+
         path_overrides = to_path_overrides(params)
+        hydra_sweep = any(
+            x.is_sweep_override()
+            for param_file in path_overrides
+            for x in to_hydra_overrides(path_overrides[param_file])
+        )
+
+        if hydra_sweep and not queue:
+            raise InvalidArgumentError(
+                "Sweep overrides can't be used without `--queue`"
+            )
     else:
         path_overrides = {}
 
@@ -45,23 +55,16 @@ def run(
         # Force `_update_params` even if `--set-param` was not used
         path_overrides[hydra_output_file] = []
 
-    hydra_sweep = any(
-        x.is_sweep_override()
-        for param_file in path_overrides
-        for x in to_hydra_overrides(path_overrides[param_file])
-    )
-
-    if hydra_sweep and not queue:
-        raise InvalidArgumentError(
-            "Sweep overrides can't be used without `--queue`"
-        )
-
     if not queue:
         return repo.experiments.reproduce_one(
             targets=targets, params=path_overrides, tmp_dir=tmp_dir, **kwargs
         )
 
     if hydra_sweep:
+        if kwargs.get("name"):
+            raise InvalidArgumentError(
+                "Sweep overrides can't be used alongside `--name`"
+            )
         from dvc.utils.hydra import get_hydra_sweeps
 
         sweeps = get_hydra_sweeps(path_overrides)

dvc/repo/experiments/save.py

@@ -0,0 +1,80 @@
+import logging
+import os
+from typing import TYPE_CHECKING, List, Optional
+
+from .exceptions import ExperimentExistsError
+from .refs import ExpRefInfo
+from .utils import check_ref_format, get_random_exp_name
+
+if TYPE_CHECKING:
+    from dvc.repo import Repo
+
+
+logger = logging.getLogger(__name__)
+
+
+def _save_experiment(
+    repo: "Repo",
+    baseline_rev: str,
+    force: bool,
+    name: Optional[str],
+    include_untracked: Optional[List[str]],
+) -> str:
+    repo.commit([], force=True)
+
+    name = name or get_random_exp_name(repo.scm, baseline_rev)
+    ref_info = ExpRefInfo(baseline_rev, name)
+    check_ref_format(repo.scm.dulwich, ref_info)
+    ref = str(ref_info)
+    if repo.scm.get_ref(ref) and not force:
+        raise ExperimentExistsError(ref_info.name, command="save")
+
+    repo.scm.add([], update=True)
+    if include_untracked:
+        repo.scm.add(include_untracked)
+    repo.scm.commit(f"dvc: commit experiment {name}", no_verify=True)
+    exp_rev = repo.scm.get_rev()
+    repo.scm.set_ref(ref, exp_rev, old_ref=None)
+
+    return exp_rev
+
+
+def save(
+    repo: "Repo",
+    name: Optional[str] = None,
+    force: bool = False,
+    include_untracked: Optional[List[str]] = None,
+) -> Optional[str]:
+    """Save the current workspace status as an experiment.
+
+    Returns the saved experiment's SHAs.
+    """
+    logger.debug("Saving workspace in %s", os.getcwd())
+
+    _, _, untracked = repo.scm.status()
+    if include_untracked:
+        untracked = [
+            file for file in untracked if file not in include_untracked
+        ]
+    if untracked:
+        logger.warning(
+            "The following untracked files were present in "
+            "the workspace before saving but "
+            "will not be included in the experiment commit:\n"
+            "\t%s",
+            ", ".join(untracked),
+        )
+
+    with repo.scm.detach_head(client="dvc") as orig_head:
+        with repo.scm.stash_workspace() as workspace:
+            try:
+                if workspace is not None:
+                    repo.scm.stash.apply(workspace)
+
+                exp_rev = _save_experiment(
+                    repo, orig_head, force, name, include_untracked
+                )
+            finally:
+                repo.scm.reset(hard=True)
+
+    return exp_rev

dvc/repo/experiments/show.py

@@ -3,17 +3,32 @@ from collections import OrderedDict, defaultdict
 from datetime import datetime
 from enum import Enum
 from itertools import chain
-from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, Union
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Callable,
+    Dict,
+    Iterable,
+    List,
+    Optional,
+    Set,
+    Tuple,
+    Union,
+)
+
+from scmrepo.exceptions import SCMError as InnerScmError
 
-from dvc.repo.experiments.queue.base import QueueDoneResult
 from dvc.repo.metrics.show import _gather_metrics
 from dvc.repo.params.show import _gather_params
-from dvc.scm import iter_revs
+from dvc.scm import SCMError, iter_revs, resolve_rev
 from dvc.utils import error_handler, onerror_collect, relpath
 
 from .refs import ExpRefInfo
 
 if TYPE_CHECKING:
+    from scmrepo.git import Git
+    from scmrepo.git.objects import GitCommit
+
     from dvc.repo import Repo
 
 logger = logging.getLogger(__name__)
@@ -26,34 +41,49 @@ class ExpStatus(Enum):
     Failed = 3
 
 
+def _is_scm_error(collected_exp: Dict[str, Any]) -> bool:
+    if "error" in collected_exp and (
+        isinstance(collected_exp["error"], SCMError)
+        or isinstance(collected_exp["error"], InnerScmError)
+    ):
+        return True
+    return False
+
+
+def _show_onerror_collect(result: Dict, exception: Exception, *args, **kwargs):
+    onerror_collect(result, exception, *args, **kwargs)
+    result["data"] = {}
+
+
 @error_handler
-def _collect_experiment_commit(
+def collect_experiment_commit(
     repo: "Repo",
     exp_rev: str,
     status: ExpStatus = ExpStatus.Success,
     param_deps=False,
-    running=None,
+    running: Optional[Dict[str, Any]] = None,
     onerror: Optional[Callable] = None,
-):
+) -> Dict[str, Any]:
     from dvc.dependency import ParamsDependency, RepoDependency
 
-    res: Dict[str, Optional[Any]] = defaultdict(dict)
+    result: Dict[str, Any] = defaultdict(dict)
+    running = running or {}
     for rev in repo.brancher(revs=[exp_rev]):
         if rev == "workspace":
             if exp_rev != "workspace":
                 continue
-            res["timestamp"] = None
+            result["timestamp"] = None
         else:
             commit = repo.scm.resolve_commit(rev)
-            res["timestamp"] = datetime.fromtimestamp(commit.commit_time)
+            result["timestamp"] = datetime.fromtimestamp(commit.commit_time)
 
         params = _gather_params(
             repo, rev=rev, targets=None, deps=param_deps, onerror=onerror
         )
         if params:
-            res["params"] = params
+            result["params"] = params
 
-        res["deps"] = {
+        result["deps"] = {
             relpath(dep.fs_path, repo.root_dir): {
                 "hash": dep.hash_info.value,
                 "size": dep.meta.size,
@@ -62,7 +92,7 @@ def _collect_experiment_commit(
             for dep in repo.index.deps
             if not isinstance(dep, (ParamsDependency, RepoDependency))
         }
-        res["outs"] = {
+        result["outs"] = {
             relpath(out.fs_path, repo.root_dir): {
                 "hash": out.hash_info.value,
                 "size": out.meta.size,
@@ -74,14 +104,14 @@ def _collect_experiment_commit(
             if not (out.is_metric or out.is_plot)
         }
 
-        res["status"] = status.name
+        result["status"] = status.name
         if status == ExpStatus.Running:
-            res["executor"] = running[exp_rev].get("location")
+            result["executor"] = running.get(exp_rev, {}).get("location", None)
         else:
-            res["executor"] = None
+            result["executor"] = None
 
         if status == ExpStatus.Failed:
-            res["error"] = {
+            result["error"] = {
                 "msg": "Experiment run failed.",
                 "type": "",
             }
@@ -90,96 +120,274 @@ def _collect_experiment_commit(
             vals = _gather_metrics(
                 repo, targets=None, rev=rev, recursive=False, onerror=onerror
             )
-            res["metrics"] = vals
+            result["metrics"] = vals
 
-    return res
+    return result
 
 
-def _collect_experiment_branch(
-    res,
-    repo,
-    branch,
-    baseline,
-    onerror: Optional[Callable] = None,
-    running=None,
+def _collect_complete_experiment(
+    repo: "Repo",
+    baseline: str,
+    exp_rev: str,
+    running: Dict[str, Any],
+    revs: List[str],
     **kwargs,
-):
-    from dvc.scm import resolve_rev
+) -> Dict[str, Dict[str, Any]]:
+    results: "OrderedDict[str, Dict[str, Any]]" = OrderedDict()
+
+    checkpoint: bool = True if len(revs) > 1 else False
+    prev = ""
 
-    exp_rev = resolve_rev(repo.scm, branch)
-    prev = None
-    revs = list(repo.scm.branch_revs(exp_rev, baseline))
     for rev in revs:
         status = ExpStatus.Running if rev in running else ExpStatus.Success
-        collected_exp = _collect_experiment_commit(
+        collected_exp = collect_experiment_commit(
             repo,
             rev,
-            onerror=onerror,
             status=status,
             running=running,
             **kwargs,
         )
-        if len(revs) > 1:
+        if _is_scm_error(collected_exp):
+            return {}
+        if checkpoint:
             exp = {"checkpoint_tip": exp_rev}
             if prev:
-                res[prev]["data"][  # type: ignore[unreachable]
+                results[prev]["data"][  # type: ignore[unreachable]
                     "checkpoint_parent"
                 ] = rev
-            if rev in res:
-                res[rev]["data"].update(exp)
-                res.move_to_end(rev)
+            if rev in results:
+                results[rev]["data"].update(exp)
+                results.move_to_end(rev)
             else:
                 exp.update(collected_exp["data"])
         else:
             exp = collected_exp["data"]
-        if rev not in res:
-            res[rev] = {"data": exp}
+        if rev not in results:
+            results[rev] = {"data": exp}
         prev = rev
-    if len(revs) > 1:
-        res[prev]["data"]["checkpoint_parent"] = baseline
-    return res
+    if checkpoint and prev:
+        results[prev]["data"]["checkpoint_parent"] = baseline
+    return results
 
 
-def get_names(repo: "Repo", result: Dict[str, Dict[str, Any]]):
+def _collect_branch(
+    repo: "Repo",
+    baseline: str,
+    running: Dict[str, Any],
+    **kwargs,
+) -> Dict[str, Dict[str, Any]]:
+    results: Dict[str, Dict[str, Any]] = OrderedDict()
+    status = ExpStatus.Running if baseline in running else ExpStatus.Success
+    results["baseline"] = collect_experiment_commit(
+        repo,
+        baseline,
+        status=status,
+        running=running,
+        **kwargs,
+    )
+    if baseline == "workspace" or _is_scm_error(results["baseline"]):
+        return results
+
+    ref_info = ExpRefInfo(baseline_sha=baseline)
+    commits: List[Tuple[str, "GitCommit", str, List[str]]] = []
+
+    for ref in repo.scm.iter_refs(base=str(ref_info)):
+        try:
+            commit = repo.scm.resolve_commit(ref)
+            exp_rev = resolve_rev(repo.scm, ref)
+            revs = list(repo.scm.branch_revs(exp_rev, baseline))
+        except (SCMError, InnerScmError):
+            continue
+        commits.append((ref, commit, exp_rev, revs))
+
+    for exp_ref, _, exp_rev, revs in sorted(
+        commits, key=lambda x: x[1].commit_time, reverse=True
+    ):
+        ref_info = ExpRefInfo.from_ref(exp_ref)
+        assert ref_info.baseline_sha == baseline
+        collected_exp = _collect_complete_experiment(
+            repo,
+            baseline=baseline,
+            exp_rev=exp_rev,
+            running=running,
+            revs=revs,
+            **kwargs,
+        )
+        if _is_scm_error(collected_exp):
+            continue
+        results.update(collected_exp)
+    return results
+
+
+def get_branch_names(
+    scm: "Git", baseline_set: Iterable[str]
+) -> Dict[str, Optional[str]]:
+    names: Dict[str, Optional[str]] = {}
+    for base in [
+        f"refs/exps/{baseline[:2]}/{baseline[2:]}/"
+        for baseline in baseline_set
+    ] + ["refs/heads/", "refs/tags/"]:
+        for ref in scm.iter_refs(base=base):
+            if ref:
+                try:
+                    rev = scm.get_ref(ref)
+                    names[rev] = ref[len(base) :]
+                except KeyError:
+                    logger.debug("unreosolved ref %s", ref)
+    logger.debug("found refs for revs: %s", names)
+    return names
+
+
+def update_names(
+    repo: "Repo",
+    branch_names: Dict[str, Optional[str]],
+    result: Dict[str, Dict[str, Any]],
+):
 
     rev_set = set()
-    baseline_set = set()
     for baseline in result:
         for rev in result[baseline]:
             if rev == "baseline":
                 rev = baseline
-                baseline_set.add(rev)
             if rev != "workspace":
                 rev_set.add(rev)
 
-    names: Dict[str, Optional[str]] = {}
-    for base in ("refs/tags/", "refs/heads/"):
-        if rev_set:
-            names.update(
-                (rev, ref[len(base) :])
-                for rev, ref in repo.scm.describe(
-                    baseline_set, base=base
-                ).items()
-                if ref is not None
-            )
-            rev_set.difference_update(names.keys())
+    if rev_set:
+        rev_set.difference_update(branch_names.keys())
 
     exact_name = repo.experiments.get_exact_name(rev_set)
 
     for baseline, baseline_results in result.items():
+        name_set: Set[str] = set()
         for rev, rev_result in baseline_results.items():
             name: Optional[str] = None
             if rev == "baseline":
                 rev = baseline
                 if rev == "workspace":
                     continue
-                name = names.get(rev, None)
-            name = name or exact_name[rev]
-            if name:
+            name = branch_names.get(rev, None) or exact_name[rev]
+            if name and name not in name_set:
+                name_set.add(name)
                 rev_result["data"]["name"] = name
 
 
-# flake8: noqa: C901
+def _collect_active_experiment(
+    repo: "Repo",
+    found_revs: Dict[str, List[str]],
+    running: Dict[str, Any],
+    **kwargs,
+) -> Dict[str, Dict[str, Any]]:
+    result: Dict[str, Dict] = defaultdict(OrderedDict)
+    for entry in chain(
+        repo.experiments.tempdir_queue.iter_active(),
+        repo.experiments.celery_queue.iter_active(),
+    ):
+        stash_rev = entry.stash_rev
+        if entry.baseline_rev in found_revs and (
+            stash_rev not in running or not running[stash_rev].get("last")
+        ):
+            collected_exp = collect_experiment_commit(
+                repo,
+                stash_rev,
+                status=ExpStatus.Running,
+                running=running,
+                **kwargs,
+            )
+            if _is_scm_error(collected_exp):
+                continue
+            result[entry.baseline_rev][stash_rev] = collected_exp
+    return result
+
+
+def _collect_queued_experiment(
+    repo: "Repo",
+    found_revs: Dict[str, List[str]],
+    running: Dict[str, Any],
+    **kwargs,
+) -> Dict[str, Dict[str, Any]]:
+    result: Dict[str, Dict] = defaultdict(OrderedDict)
+    for entry in repo.experiments.celery_queue.iter_queued():
+        stash_rev = entry.stash_rev
+        if entry.baseline_rev in found_revs:
+            collected_exp = collect_experiment_commit(
+                repo,
+                stash_rev,
+                status=ExpStatus.Queued,
+                running=running,
+                **kwargs,
+            )
+            if _is_scm_error(collected_exp):
+                continue
+            result[entry.baseline_rev][stash_rev] = collected_exp
+    return result
+
+
+def _collect_failed_experiment(
+    repo: "Repo",
+    found_revs: Dict[str, List[str]],
+    running: Dict[str, Any],
+    **kwargs,
+) -> Dict[str, Dict[str, Any]]:
+    result: Dict[str, Dict] = defaultdict(OrderedDict)
+    for queue_done_result in repo.experiments.celery_queue.iter_failed():
+        entry = queue_done_result.entry
+        stash_rev = entry.stash_rev
+        if entry.baseline_rev in found_revs:
+            collected_exp = collect_experiment_commit(
+                repo,
+                stash_rev,
+                status=ExpStatus.Failed,
+                running=running,
+                **kwargs,
+            )
+            if _is_scm_error(collected_exp):
+                continue
+            result[entry.baseline_rev][stash_rev] = collected_exp
+    return result
+
+
+def update_new(
+    to_dict: Dict[str, Dict[str, Any]], from_dict: Dict[str, Dict[str, Any]]
+):
+    for baseline, experiments in from_dict.items():
+        for rev, experiment in experiments.items():
+            to_dict[baseline][rev] = to_dict[baseline].get(rev, experiment)
+
+
+def move_properties_to_head(result: Dict[str, Dict[str, Dict[str, Any]]]):
+    for _, baseline_results in result.items():
+        checkpoint: bool = False
+        head: Dict[str, Any] = {}
+        for rev, rev_data in baseline_results.items():
+            if (
+                "data" not in rev_data
+                or rev_data["data"].get("checkpoint_tip", None) is None
+            ):
+                checkpoint = False
+                head = {}
+                continue
+
+            rev_result: Dict[str, Any] = rev_data["data"]
+            if (
+                checkpoint is True
+                and rev_result["checkpoint_tip"] == head["checkpoint_tip"]
+            ):
+                if "name" in rev_result and "name" not in head:
+                    head["name"] = rev_result["name"]
+                    del rev_result["name"]
+                if rev_result["executor"]:
+                    if not head["executor"]:
+                        head["executor"] = rev_result["executor"]
+                    rev_result["executor"] = None
+                if rev_result["status"] == ExpStatus.Running.name:
+                    head["status"] = ExpStatus.Running.name
+                    rev_result["status"] = ExpStatus.Success.name
+            else:
+                if rev_result["checkpoint_tip"] == rev:
+                    head = rev_result
+                    checkpoint = True
+
+
 def show(
     repo: "Repo",
     all_branches=False,
@@ -198,8 +406,7 @@ def show(
     if repo.scm.no_commits:
         return {}
 
-    if onerror is None:
-        onerror = onerror_collect
+    onerror = onerror or _show_onerror_collect
 
     res: Dict[str, Dict] = defaultdict(OrderedDict)
 
@@ -212,77 +419,62 @@ def show(
     found_revs.update(
         iter_revs(repo.scm, revs, num, all_branches, all_tags, all_commits)
     )
+    branch_names = get_branch_names(repo.scm, found_revs)
 
-    running = repo.experiments.get_running_exps(fetch_refs=fetch_running)
+    running: Dict[str, Dict] = repo.experiments.get_running_exps(
+        fetch_refs=fetch_running
+    )
 
-    for rev in found_revs:
-        status = ExpStatus.Running if rev in running else ExpStatus.Success
-        res[rev]["baseline"] = _collect_experiment_commit(
+    queued_experiment = (
+        _collect_queued_experiment(
             repo,
-            rev,
-            status=status,
+            found_revs,
+            running,
             param_deps=param_deps,
+            onerror=onerror,
+        )
+        if not hide_queued
+        else {}
+    )
+
+    active_experiment = _collect_active_experiment(
+        repo,
+        found_revs,
+        running,
+        param_deps=param_deps,
+        onerror=onerror,
+    )
+
+    failed_experiments = (
+        _collect_failed_experiment(
+            repo,
+            found_revs,
+            running,
+            param_deps=param_deps,
+            onerror=onerror,
+        )
+        if not hide_failed
+        else {}
+    )
+
+    for baseline in found_revs:
+        res[baseline] = _collect_branch(
+            repo,
+            baseline,
             running=running,
+            param_deps=param_deps,
             onerror=onerror,
         )
 
-        if rev == "workspace":
-            continue
+    update_new(res, failed_experiments)
 
-        ref_info = ExpRefInfo(baseline_sha=rev)
-        commits = [
-            (ref, repo.scm.resolve_commit(ref))
-            for ref in repo.scm.iter_refs(base=str(ref_info))
-        ]
-        for exp_ref, _ in sorted(
-            commits, key=lambda x: x[1].commit_time, reverse=True
-        ):
-            ref_info = ExpRefInfo.from_ref(exp_ref)
-            assert ref_info.baseline_sha == rev
-            _collect_experiment_branch(
-                res[rev],
-                repo,
-                exp_ref,
-                rev,
-                param_deps=param_deps,
-                running=running,
-                onerror=onerror,
-            )
+    update_new(res, active_experiment)
 
-        # collect standalone & celery experiments
-        for entry in chain(
-            repo.experiments.tempdir_queue.iter_active(),
-            repo.experiments.celery_queue.iter_active(),
-            repo.experiments.celery_queue.iter_queued(),
-            repo.experiments.celery_queue.iter_failed(),
-        ):
-            if isinstance(entry, QueueDoneResult):
-                entry = entry.entry
-                if hide_failed:
-                    continue
-                status = ExpStatus.Failed
-            elif entry.stash_rev in running:
-                status = ExpStatus.Running
-            else:
-                if hide_queued:
-                    continue
-                status = ExpStatus.Queued
-            stash_rev = entry.stash_rev
-            if entry.baseline_rev in found_revs:
-                if stash_rev not in running or not running[stash_rev].get(
-                    "last"
-                ):
-                    experiment = _collect_experiment_commit(
-                        repo,
-                        stash_rev,
-                        status=status,
-                        param_deps=param_deps,
-                        running=running,
-                        onerror=onerror,
-                    )
-                    res[entry.baseline_rev][stash_rev] = experiment
+    update_new(res, queued_experiment)
 
     if not sha_only:
-        get_names(repo, res)
+        update_names(repo, branch_names, res)
+
+    move_properties_to_head(res)
 
     return res

dvc/repo/fetch.py

@@ -1,6 +1,6 @@
 import logging
 from contextlib import suppress
-from typing import TYPE_CHECKING, Optional
+from typing import TYPE_CHECKING, Optional, Sequence
 
 from dvc.config import NoRemoteError
 from dvc.exceptions import DownloadError
@@ -9,6 +9,7 @@ from dvc.fs import Schemes
 from . import locked
 
 if TYPE_CHECKING:
+    from dvc.cloud import Remote
     from dvc.repo import Repo
     from dvc.types import TargetType
     from dvc_data.hashfile.transfer import TransferResult
@@ -17,20 +18,6 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 
-def _fetch_worktree(repo, remote):
-    from dvc_data.index import save
-
-    index = repo.index.data["repo"]
-    for key, entry in index.iteritems():
-        entry.fs = remote.fs
-        entry.path = remote.fs.path.join(
-            remote.path,
-            *key,
-        )
-    save(index)
-    return len(index)
-
-
 @locked
 def fetch(
     self,
@@ -64,12 +51,14 @@ def fetch(
     if isinstance(targets, str):
         targets = [targets]
 
+    worktree_remote: Optional["Remote"] = None
     with suppress(NoRemoteError):
         _remote = self.cloud.get_remote(name=remote)
-        if _remote.worktree:
-            return _fetch_worktree(self, _remote)
+        if _remote.worktree or _remote.fs.version_aware:
+            worktree_remote = _remote
 
     failed_count = 0
+    transferred_count = 0
 
     try:
         if run_cache:
@@ -80,22 +69,36 @@ def fetch(
     no_remote_msg: Optional[str] = None
     result = TransferResult(set(), set())
     try:
-        d, f = _fetch(
-            self,
-            targets,
-            all_branches=all_branches,
-            all_tags=all_tags,
-            all_commits=all_commits,
-            with_deps=with_deps,
-            force=True,
-            remote=remote,
-            jobs=jobs,
-            recursive=recursive,
-            revs=revs,
-            odb=odb,
-        )
-        result.transferred.update(d)
-        result.failed.update(f)
+        if worktree_remote is not None:
+            transferred_count += _fetch_worktree(
+                self,
+                worktree_remote,
+                revs=revs,
+                all_branches=all_branches,
+                all_tags=all_tags,
+                all_commits=all_commits,
+                targets=targets,
+                jobs=jobs,
+                with_deps=with_deps,
+                recursive=recursive,
+            )
+        else:
+            d, f = _fetch(
+                self,
+                targets,
+                all_branches=all_branches,
+                all_tags=all_tags,
+                all_commits=all_commits,
+                with_deps=with_deps,
+                force=True,
+                remote=remote,
+                jobs=jobs,
+                recursive=recursive,
+                revs=revs,
+                odb=odb,
+            )
+            result.transferred.update(d)
+            result.failed.update(f)
     except NoRemoteError as exc:
         no_remote_msg = str(exc)
 
@@ -121,7 +124,8 @@ def fetch(
             logger.error(no_remote_msg)
         raise DownloadError(failed_count)
 
-    return len(result.transferred)
+    transferred_count += len(result.transferred)
+    return transferred_count
 
 
 def _fetch(
@@ -168,3 +172,29 @@ def _fetch(
             result.transferred.update(d)
             result.failed.update(f)
     return result
+
+
+def _fetch_worktree(
+    repo: "Repo",
+    remote: "Remote",
+    revs: Optional[Sequence[str]] = None,
+    all_branches: bool = False,
+    all_tags: bool = False,
+    all_commits: bool = False,
+    targets: Optional["TargetType"] = None,
+    jobs: Optional[int] = None,
+    **kwargs,
+) -> int:
+    from dvc.repo.worktree import fetch_worktree
+
+    downloaded = 0
+    for _ in repo.brancher(
+        revs=revs,
+        all_branches=all_branches,
+        all_tags=all_tags,
+        all_commits=all_commits,
+    ):
+        downloaded += fetch_worktree(
+            repo, remote, targets=targets, jobs=jobs, **kwargs
+        )
+    return downloaded

dvc/repo/imp_url.py

@@ -97,6 +97,9 @@ def imp_url(
     else:
         stage.run(jobs=jobs, no_download=no_download)
 
+    if not no_exec and stage.deps[0].fs.version_aware:
+        stage.outs[0].can_push = False
+
     stage.frozen = frozen
 
     dvcfile.dump(stage)

dvc/repo/imports.py

@@ -44,7 +44,7 @@ def unfetched_view(
 
         return True
 
-    unfetched = index.targets_view(targets, filter_fn=need_fetch, **kwargs)
+    unfetched = index.targets_view(targets, stage_filter=need_fetch, **kwargs)
     return unfetched, changed_deps
 
 
@@ -53,7 +53,7 @@ def partial_view(
 ) -> "IndexView":
     return index.targets_view(
         targets,
-        filter_fn=lambda s: s.is_partial_import,
+        stage_filter=lambda s: s.is_partial_import,
         **kwargs,
     )
 

dvc/repo/index.py

@@ -1,4 +1,3 @@
-from contextlib import suppress
 from typing import (
     TYPE_CHECKING,
     Any,
@@ -7,13 +6,14 @@ from typing import (
     Iterable,
     Iterator,
     List,
+    NamedTuple,
     Optional,
     Set,
+    Tuple,
+    Union,
 )
 
-from funcy import cached_property, nullcontext
-
-from dvc.utils import dict_md5
+from funcy import cached_property
 
 if TYPE_CHECKING:
     from networkx import DiGraph
@@ -22,9 +22,10 @@ if TYPE_CHECKING:
     from dvc.dependency import Dependency, ParamsDependency
     from dvc.fs import FileSystem
     from dvc.output import Output
+    from dvc.repo import Repo
     from dvc.repo.stage import StageInfo, StageLoad
     from dvc.stage import Stage
-    from dvc.types import StrPath, TargetType
+    from dvc.types import TargetType
     from dvc_data.hashfile.hash_info import HashInfo
     from dvc_data.index import DataIndex, DataIndexKey, DataIndexView
     from dvc_objects.db import ObjectDB
@@ -65,14 +66,65 @@ class Index:
         if stages is not None:
             self.stages: List["Stage"] = stages
         self._collected_targets: Dict[int, List["StageInfo"]] = {}
+        self._metrics: Dict[str, List[str]] = {}
+        self._plots: Dict[str, Any] = {}
+        self._params: Dict[str, List[str]] = {}
 
     @cached_property
     def stages(self) -> List["Stage"]:  # pylint: disable=method-hidden
         # note that ideally we should be keeping this in a set as it is unique,
         # hashable and has no concept of orderliness on its own. But we depend
         # on this to be somewhat ordered for status/metrics/plots, etc.
+        return self._collect()
+
+    @cached_property
+    def _top_metrics(self):
+        self._collect()
+        return self._metrics
+
+    @cached_property
+    def _top_plots(self):
+        self._collect()
+        return self._plots
+
+    @cached_property
+    def _plot_sources(self):
+        from dvc.repo.plots import _collect_pipeline_files
+
+        sources = []
+        for data in _collect_pipeline_files(self.repo, [], {}).values():
+            for plot_id, props in data.get("data", {}).items():
+                if isinstance(props.get("y"), dict):
+                    sources.extend(props["y"])
+                    if isinstance(props.get("x"), dict):
+                        sources.extend(props["x"])
+                else:
+                    sources.append(plot_id)
+        return sources
+
+    @cached_property
+    def _top_params(self):
+        self._collect()
+        return self._params
+
+    def _collect(self):
+        if "stages" in self.__dict__:
+            return self.stages
+
         onerror = self.repo.stage_collection_error_handler
-        return self.stage_collector.collect_repo(onerror=onerror)
+
+        # pylint: disable=protected-access
+        (
+            stages,
+            metrics,
+            plots,
+            params,
+        ) = self.stage_collector._collect_all_from_repo(onerror=onerror)
+        self.stages = stages
+        self._metrics = metrics
+        self._plots = plots
+        self._params = params
+        return stages
 
     def __repr__(self) -> str:
         from dvc.fs import LocalFileSystem
@@ -82,38 +134,6 @@ class Index:
             rev = self.repo.get_rev()[:7]
         return f"Index({self.repo}, fs@{rev})"
 
-    def __len__(self) -> int:
-        return len(self.stages)
-
-    def __contains__(self, stage: "Stage") -> bool:
-        # as we are keeping stages inside a list, it might be slower.
-        return stage in self.stages
-
-    def __iter__(self) -> Iterator["Stage"]:
-        yield from self.stages
-
-    def __getitem__(self, item: str) -> "Stage":
-        """Get a stage by its addressing attribute."""
-        for stage in self:
-            if stage.addressing == item:
-                return stage
-        raise KeyError(f"{item} - available stages are {self.stages}")
-
-    def filter(self, filter_fn: Callable[["Stage"], bool]) -> "Index":
-        stages_it = filter(filter_fn, self)
-        return Index(self.repo, self.fs, stages=list(stages_it))
-
-    def slice(self, path: "StrPath") -> "Index":
-        from dvc.utils import relpath
-        from dvc.utils.fs import path_isin
-
-        target_path = relpath(path, self.repo.root_dir)
-
-        def is_stage_inside_path(stage: "Stage") -> bool:
-            return path_isin(stage.path_in_repo, target_path)
-
-        return self.filter(is_stage_inside_path)
-
     @staticmethod
     def _hash_targets(
         targets: Iterable[Optional[str]],
@@ -127,7 +147,7 @@ class Index:
             )
         )
 
-    def _collect_targets(
+    def collect_targets(
         self, targets: Optional["TargetType"], **kwargs: Any
     ) -> List["StageInfo"]:
         from itertools import chain
@@ -150,29 +170,39 @@ class Index:
 
     def targets_view(
         self,
-        targets: "TargetType",
-        filter_fn: Optional[Callable[["Stage"], bool]] = None,
+        targets: Optional["TargetType"],
+        stage_filter: Optional[Callable[["Stage"], bool]] = None,
+        outs_filter: Optional[Callable[["Output"], bool]] = None,
         **kwargs: Any,
     ) -> "IndexView":
         """Return read-only view of index for the specified targets.
 
         Args:
             targets: Targets to collect
-            filter_fn: Optional stage filter to be applied after collecting
+            stage_filter: Optional stage filter to be applied after collecting
+                targets.
+            outs_filter: Optional output filter to be applied after collecting
                 targets.
 
         Additional kwargs will be passed into the stage collector.
+
+        Note:
+            If both stage_filter and outs_filter are provided, stage_filter
+            will be applied first, and the resulting view will only contain
+            outputs from stages that matched stage_filter. Outputs from stages
+            that did not match will be excluded from the view (whether or not
+            the output would have matched outs_filter).
         """
         stage_infos = [
             stage_info
-            for stage_info in self._collect_targets(targets, **kwargs)
-            if not filter_fn or filter_fn(stage_info.stage)
+            for stage_info in self.collect_targets(targets, **kwargs)
+            if not stage_filter or stage_filter(stage_info.stage)
         ]
-        return IndexView(self, stage_infos)
+        return IndexView(self, stage_infos, outs_filter=outs_filter)
 
     @property
     def outs(self) -> Iterator["Output"]:
-        for stage in self:
+        for stage in self.stages:
             yield from stage.outs
 
     @property
@@ -195,7 +225,7 @@ class Index:
 
     @property
     def deps(self) -> Iterator["Dependency"]:
-        for stage in self:
+        for stage in self.stages:
             yield from stage.deps
 
     @property
@@ -242,15 +272,8 @@ class Index:
             workspace, key = out.index_key
             data_index = by_workspace[workspace]
 
-            if out.files:
-                out.obj = out.get_obj()
-
             entry = out.get_entry()
             entry.key = key
-            if out.stage.is_import and not out.stage.is_repo_import:
-                entry.fs = out.stage.deps[0].fs
-                entry.path = out.stage.deps[0].fs_path
-                entry.meta = out.stage.deps[0].meta
             data_index.add(entry)
 
         return dict(by_workspace)
@@ -263,11 +286,12 @@ class Index:
         force: bool = False,
         recursive: bool = False,
         jobs: int = None,
+        push: bool = False,
     ) -> "ObjectContainer":
         from collections import defaultdict
 
         used: "ObjectContainer" = defaultdict(set)
-        pairs = self._collect_targets(
+        pairs = self.collect_targets(
             targets, recursive=recursive, with_deps=with_deps
         )
         for stage, filter_info in pairs:
@@ -276,6 +300,7 @@ class Index:
                 force=force,
                 jobs=jobs,
                 filter_info=filter_info,
+                push=push,
             ).items():
                 used[odb].update(objs)
         return used
@@ -294,147 +319,92 @@ class Index:
     def add(self, stage: "Stage") -> "Index":
         return self.update([stage])
 
-    def remove(
-        self, stage: "Stage", ignore_not_existing: bool = False
-    ) -> "Index":
-        stages = self._discard_stage(
-            stage, ignore_not_existing=ignore_not_existing
-        )
-        return Index(self.repo, self.fs, stages=stages)
-
-    def discard(self, stage: "Stage") -> "Index":
-        return self.remove(stage, ignore_not_existing=True)
-
-    def difference(self, stages: Iterable["Stage"]) -> "Index":
-        stages_set = set(self.stages) - set(stages)
-        return Index(self.repo, self.fs, stages=list(stages_set))
-
-    def _discard_stage(
-        self, stage: "Stage", ignore_not_existing: bool = False
-    ) -> List["Stage"]:
-        stages = self.stages[:]
-        ctx = suppress(ValueError) if ignore_not_existing else nullcontext()
-        with ctx:
-            stages.remove(stage)
-        return stages
-
     def check_graph(self) -> None:
         if not getattr(self.repo, "_skip_graph_checks", False):
             self.graph  # pylint: disable=pointless-statement
 
-    def dumpd(self) -> Dict[str, Dict]:
-        def dump(stage: "Stage"):
-            key = stage.path_in_repo
-            try:
-                key += ":" + stage.name  # type: ignore[attr-defined]
-            except AttributeError:
-                pass
-            return key, stage.dumpd()
 
-        return dict(dump(stage) for stage in self)
-
-    @cached_property
-    def identifier(self) -> str:
-        """Unique identifier for the index.
-
-        We can use this to optimize and skip opening some indices
-        eg: on push/pull/fetch/gc --all-commits.
-
-        Currently, it is unique to the platform (windows vs posix).
-        """
-        return dict_md5(self.dumpd())
+class _DataPrefixes(NamedTuple):
+    explicit: Set["DataIndexKey"]
+    recursive: Set["DataIndexKey"]
 
 
 class IndexView:
     """Read-only view of Index.data using filtered stages."""
 
     def __init__(  # pylint: disable=redefined-outer-name
-        self, index: Index, stage_infos: Iterable["StageInfo"]
+        self,
+        index: Index,
+        stage_infos: Iterable["StageInfo"],
+        outs_filter: Optional[Callable[["Output"], bool]],
     ):
         self._index = index
-        self._stages: Dict["Stage", Optional[str]] = dict(stage_infos)
-
-    def __len__(self) -> int:
-        return len(self._stages)
-
-    def __contains__(self, stage: "Stage") -> bool:
-        # as we are keeping stages inside a list, it might be slower.
-        return stage in self._stages
-
-    def __iter__(self) -> Iterator["Stage"]:
-        yield from self._stages
-
-    def __getitem__(self, item: str) -> "Stage":
-        """Get a stage by its addressing attribute."""
-        for stage in self:
-            if stage.addressing == item:
-                return stage
-        raise KeyError(f"{item} - available stages are {self.stages}")
+        self._stage_infos = stage_infos
+        # NOTE: stage_infos might have the same stage multiple times but with
+        # different filter_info
+        self._stages = list({stage for stage, _ in stage_infos})
+        self._outs_filter = outs_filter
 
     @property
     def stages(self):
-        return list(self._stages)
+        return self._stages
 
     @property
     def deps(self) -> Iterator["Dependency"]:
-        for stage in self:
+        for stage in self.stages:
             yield from stage.deps
 
+    @property
+    def _filtered_outs(self) -> Iterator[Tuple["Output", Optional[str]]]:
+        for stage, filter_info in self._stage_infos:
+            for out in stage.filter_outs(filter_info):
+                if not self._outs_filter or self._outs_filter(out):
+                    yield out, filter_info
+
     @property
     def outs(self) -> Iterator["Output"]:
-        for stage, filter_info in self._stages.items():
-            yield from stage.filter_outs(filter_info)
+        yield from {out for (out, _) in self._filtered_outs}
 
     @cached_property
-    def _data_prefixes(self) -> Dict[str, Set["DataIndexKey"]]:
+    def _data_prefixes(self) -> Dict[str, "_DataPrefixes"]:
         from collections import defaultdict
 
-        prefixes: Dict[str, Set["DataIndexKey"]] = defaultdict(set)
-        for out in self.outs:
+        prefixes: Dict[str, "_DataPrefixes"] = defaultdict(
+            lambda: _DataPrefixes(set(), set())
+        )
+        for out, filter_info in self._filtered_outs:
             workspace, key = out.index_key
-            prefixes[workspace].add(key)
-            prefixes[workspace].update(key[:i] for i in range(len(key), 0, -1))
+            if filter_info and out.fs.path.isin(filter_info, out.fs_path):
+                key = key + out.fs.path.relparts(filter_info, out.fs_path)
+            if out.meta.isdir:
+                prefixes[workspace].recursive.add(key)
+            prefixes[workspace].explicit.update(
+                key[:i] for i in range(len(key), 0, -1)
+            )
         return prefixes
 
     @cached_property
-    def data(self) -> "Dict[str, DataIndexView]":
+    def data(self) -> Dict[str, Union["DataIndex", "DataIndexView"]]:
         from functools import partial
 
-        from dvc_data.index import view
+        from dvc_data.index import DataIndex, view
 
         def key_filter(workspace: str, key: "DataIndexKey"):
             try:
-                return key in self._data_prefixes[workspace]
+                prefixes = self._data_prefixes[workspace]
+                return key in prefixes.explicit or any(
+                    key[: len(prefix)] == prefix
+                    for prefix in prefixes.recursive
+                )
             except KeyError:
                 return False
 
-        return {
-            workspace: view(index, partial(key_filter, workspace))
-            for workspace, index in self._index.data.items()
-        }
-
-
-if __name__ == "__main__":
-    import logging
-
-    from funcy import log_durations
-
-    from dvc.logger import setup
-    from dvc.repo import Repo
-
-    setup(level=logging.TRACE)  # type: ignore[attr-defined]
-
-    repo = Repo()
-    index = Index(repo, repo.fs)
-    print(index)
-    with log_durations(print, "collecting stages"):
-        # pylint: disable=pointless-statement
-        print("no of stages", len(index.stages))
-    with log_durations(print, "building graph"):
-        index.graph  # pylint: disable=pointless-statement
-    with log_durations(print, "calculating hash"):
-        print(index.identifier)
-    with log_durations(print, "updating"):
-        index2 = index.update(index.stages)
-    with log_durations(print, "calculating hash"):
-        print(index2.identifier)
+        data: Dict[str, Union["DataIndex", "DataIndexView"]] = {}
+        for workspace, data_index in self._index.data.items():
+            if self._stages:
+                data[workspace] = view(
+                    data_index, partial(key_filter, workspace)
+                )
+            else:
+                data[workspace] = DataIndex()
+        return data

dvc/repo/live.py

@@ -1,19 +0,0 @@
-import logging
-from typing import TYPE_CHECKING, Optional
-
-logger = logging.getLogger(__name__)
-
-if TYPE_CHECKING:
-    from dvc.output import Output
-
-
-def summary_fs_path(out: "Output") -> Optional[str]:
-    from dvc.output import Output
-
-    assert out.live
-    has_summary = True
-    if isinstance(out.live, dict):
-        has_summary = out.live.get(Output.PARAM_LIVE_SUMMARY, True)
-    if has_summary:
-        return out.fs.path.with_suffix(out.fs_path, ".json")
-    return None

dvc/repo/ls.py

@@ -70,11 +70,12 @@ def _ls(
         if not recursive:
             files.update(dirs)
 
+        parts = fs.path.relparts(root, fs_path)
+        if parts == (".",):
+            parts = ()
+
         for name, entry in files.items():
-            entry_fs_path = fs.path.join(root, name)
-            relparts = fs.path.relparts(entry_fs_path, fs_path)
-            name = os.path.join(*relparts)
-            infos[name] = entry
+            infos[os.path.join(*parts, name)] = entry
 
         if not recursive:
             break

dvc/repo/metrics/diff.py

@@ -11,7 +11,10 @@ def diff(repo, *args, a_rev=None, b_rev=None, **kwargs):
     a_rev = a_rev or "HEAD"
     b_rev = b_rev or "workspace"
 
-    metrics = repo.metrics.show(*args, **kwargs, revs=[a_rev, b_rev])
+    metrics = repo.metrics.show(
+        *args, **kwargs, revs=[a_rev, b_rev], hide_workspace=False
+    )
+
     old = metrics.get(a_rev, {}).get("data", {})
     new = metrics.get(b_rev, {}).get("data", {})
 

dvc/repo/metrics/show.py

@@ -8,9 +8,13 @@ from dvc.fs.dvc import DVCFileSystem
 from dvc.output import Output
 from dvc.repo import locked
 from dvc.repo.collect import StrPaths, collect
-from dvc.repo.live import summary_fs_path
 from dvc.scm import NoSCMError
-from dvc.utils import error_handler, errored_revisions, onerror_collect
+from dvc.utils import (
+    as_posix,
+    error_handler,
+    errored_revisions,
+    onerror_collect,
+)
 from dvc.utils.collections import ensure_list
 from dvc.utils.serialize import load_path
 
@@ -18,7 +22,7 @@ logger = logging.getLogger(__name__)
 
 
 def _is_metric(out: Output) -> bool:
-    return bool(out.metric) or bool(out.live)
+    return bool(out.metric)
 
 
 def _to_fs_paths(metrics: List[Output]) -> StrPaths:
@@ -26,13 +30,20 @@ def _to_fs_paths(metrics: List[Output]) -> StrPaths:
     for out in metrics:
         if out.metric:
             result.append(out.repo.dvcfs.from_os_path(out.fs_path))
-        elif out.live:
-            fs_path = summary_fs_path(out)
-            if fs_path:
-                result.append(out.repo.dvcfs.from_os_path(fs_path))
     return result
 
 
+def _collect_top_level_metrics(repo):
+    top_metrics = repo.index._top_metrics  # pylint: disable=protected-access
+    for dvcfile, metrics in top_metrics.items():
+        wdir = repo.fs.path.relpath(
+            repo.fs.path.parent(dvcfile), repo.root_dir
+        )
+        for file in metrics:
+            path = repo.fs.path.join(wdir, as_posix(file))
+            yield repo.fs.path.normpath(path)
+
+
 def _collect_metrics(repo, targets, revision, recursive):
     metrics, fs_paths = collect(
         repo,
@@ -99,6 +110,7 @@ def _read_metrics(repo, metrics, rev, onerror=None):
 
 def _gather_metrics(repo, targets, rev, recursive, onerror=None):
     metrics = _collect_metrics(repo, targets, rev, recursive)
+    metrics.extend(_collect_top_level_metrics(repo))
     return _read_metrics(repo, metrics, rev, onerror=onerror)
 
 
@@ -112,6 +124,7 @@ def show(
     revs=None,
     all_commits=False,
     onerror=None,
+    hide_workspace=True,
 ):
     if onerror is None:
         onerror = onerror_collect
@@ -130,16 +143,17 @@ def show(
             repo, targets, rev, recursive, onerror=onerror
         )
 
-    # Hide workspace metrics if they are the same as in the active branch
-    try:
-        active_branch = repo.scm.active_branch()
-    except (SCMError, NoSCMError):
-        # SCMError - detached head
-        # NoSCMError - no repo case
-        pass
-    else:
-        if res.get("workspace") == res.get(active_branch):
-            res.pop("workspace", None)
+    if hide_workspace:
+        # Hide workspace metrics if they are the same as in the active branch
+        try:
+            active_branch = repo.scm.active_branch()
+        except (SCMError, NoSCMError):
+            # SCMError - detached head
+            # NoSCMError - no repo case
+            pass
+        else:
+            if res.get("workspace") == res.get(active_branch):
+                res.pop("workspace", None)
 
     errored = errored_revisions(res)
     if errored:

dvc/repo/params/diff.py

@@ -11,7 +11,9 @@ def diff(repo, *args, a_rev=None, b_rev=None, **kwargs):
     a_rev = a_rev or "HEAD"
     b_rev = b_rev or "workspace"
 
-    params = repo.params.show(*args, **kwargs, revs=[a_rev, b_rev])
+    params = repo.params.show(
+        *args, **kwargs, revs=[a_rev, b_rev], hide_workspace=False
+    )
 
     old = params.get(a_rev, {}).get("data", {})
     new = params.get(b_rev, {}).get("data", {})

dvc/repo/params/show.py

@@ -20,7 +20,12 @@ from dvc.repo.collect import collect
 from dvc.scm import NoSCMError
 from dvc.stage import PipelineStage
 from dvc.ui import ui
-from dvc.utils import error_handler, errored_revisions, onerror_collect
+from dvc.utils import (
+    as_posix,
+    error_handler,
+    errored_revisions,
+    onerror_collect,
+)
 from dvc.utils.collections import ensure_list
 from dvc.utils.serialize import load_path
 
@@ -35,6 +40,17 @@ def _is_params(dep: "Output"):
     return isinstance(dep, ParamsDependency)
 
 
+def _collect_top_level_params(repo):
+    top_params = repo.index._top_params  # pylint: disable=protected-access
+    for dvcfile, params in top_params.items():
+        wdir = repo.fs.path.relpath(
+            repo.fs.path.parent(dvcfile), repo.root_dir
+        )
+        for file in params:
+            path = repo.fs.path.join(wdir, as_posix(file))
+            yield repo.fs.path.normpath(path)
+
+
 def _collect_configs(
     repo: "Repo", rev, targets=None, deps=False, stages=None
 ) -> Tuple[List["Output"], List[str]]:
@@ -128,6 +144,7 @@ def show(
     deps=False,
     onerror: Callable = None,
     stages=None,
+    hide_workspace=True,
 ):
     if onerror is None:
         onerror = onerror_collect
@@ -149,16 +166,17 @@ def show(
         if params:
             res[branch] = params
 
-    # Hide workspace params if they are the same as in the active branch
-    try:
-        active_branch = repo.scm.active_branch()
-    except (SCMError, NoSCMError):
-        # SCMError - detached head
-        # NoSCMError - no repo case
-        pass
-    else:
-        if res.get("workspace") == res.get(active_branch):
-            res.pop("workspace", None)
+    if hide_workspace:
+        # Hide workspace params if they are the same as in the active branch
+        try:
+            active_branch = repo.scm.active_branch()
+        except (SCMError, NoSCMError):
+            # SCMError - detached head
+            # NoSCMError - no repo case
+            pass
+        else:
+            if res.get("workspace") == res.get(active_branch):
+                res.pop("workspace", None)
 
     errored = errored_revisions(res)
     if errored:
@@ -176,6 +194,7 @@ def _gather_params(
     param_outs, params_fs_paths = _collect_configs(
         repo, rev, targets=targets, deps=deps, stages=stages
     )
+    params_fs_paths.extend(_collect_top_level_params(repo=repo))
     params = _read_params(
         repo,
         params=param_outs,

dvc/repo/plots/__init__.py

@@ -14,11 +14,12 @@ from typing import (
     List,
     Optional,
     Set,
+    Union,
 )
 
 import dpath.options
 import dpath.util
-from funcy import cached_property, first, project
+from funcy import cached_property, distinct, first, project
 
 from dvc.exceptions import DvcException
 from dvc.utils import error_handler, errored_revisions, onerror_collect
@@ -265,7 +266,7 @@ class Plots:
 
 
 def _is_plot(out: "Output") -> bool:
-    return bool(out.plot) or bool(out.live)
+    return bool(out.plot)
 
 
 def _resolve_data_sources(plots_data: Dict):
@@ -313,16 +314,18 @@ def _get_data_targets(definitions: Dict):
 
 
 def infer_data_sources(plot_id, config=None):
-    def _deduplicate(lst: List):
-        return list({elem: None for elem in lst}.keys())
-
     y = config.get("y", None)
+
     if isinstance(y, dict):
         sources = list(y.keys())
     else:
         sources = [plot_id]
 
-    return _deduplicate(source for source in sources)
+    x = config.get("x", None)
+    if isinstance(x, dict):
+        sources.append(first(x.keys()))
+
+    return distinct(source for source in sources)
 
 
 def _matches(targets, config_file, plot_id):
@@ -436,30 +439,33 @@ def _resolve_definitions(
 
 
 def _collect_pipeline_files(repo, targets: List[str], props, onerror=None):
-    from dvc.dvcfile import PipelineFile
-
     result: Dict[str, Dict] = {}
-    dvcfiles = {stage.dvcfile for stage in repo.index.stages}
-    for dvcfile in dvcfiles:
-        if isinstance(dvcfile, PipelineFile):
-            dvcfile_path = _relpath(repo.dvcfs, dvcfile.path)
-            dvcfile_defs = dvcfile.load().get("plots", {})
-            if isinstance(dvcfile_defs, list):
-                dvcfile_defs = {
-                    k: v for elem in dvcfile_defs for k, v in elem.items()
-                }
-            resolved = _resolve_definitions(
-                repo.dvcfs,
-                targets,
-                props,
-                dvcfile_path,
-                dvcfile_defs,
-                onerror=onerror,
-            )
-            dpath.util.merge(
-                result,
-                {dvcfile_path: resolved},
-            )
+    top_plots = repo.index._top_plots  # pylint: disable=protected-access
+    for dvcfile, plots_def in top_plots.items():
+        dvcfile_path = _relpath(repo.dvcfs, dvcfile)
+        dvcfile_defs_dict: Dict[str, Union[Dict, None]] = {}
+        if isinstance(plots_def, list):
+            for elem in plots_def:
+                if isinstance(elem, str):
+                    dvcfile_defs_dict[elem] = None
+                else:
+                    k, v = list(elem.items())[0]
+                    dvcfile_defs_dict[k] = v
+        else:
+            dvcfile_defs_dict = plots_def
+
+        resolved = _resolve_definitions(
+            repo.dvcfs,
+            targets,
+            props,
+            dvcfile_path,
+            dvcfile_defs_dict,
+            onerror=onerror,
+        )
+        dpath.util.merge(
+            result,
+            {dvcfile_path: resolved},
+        )
     return result
 
 
@@ -546,7 +552,7 @@ def parse(fs, path, props=None, **kwargs):
         return _load_sv(path=path, fs=fs, delimiter="\t", header=header)
     if extension in LOADERS or extension in (".yml", ".yaml"):
         return LOADERS[extension](path=path, fs=fs)
-    if extension in (".jpeg", ".jpg", ".gif", ".png"):
+    if extension in (".jpeg", ".jpg", ".gif", ".png", ".svg"):
         with fs.open(path, "rb") as fd:
             return fd.read()
     raise PlotMetricTypeError(path)
@@ -555,7 +561,7 @@ def parse(fs, path, props=None, **kwargs):
 def _plot_props(out: "Output") -> Dict:
     from dvc.schema import PLOT_PROPS
 
-    if not (out.plot or out.live):
+    if not (out.plot):
         raise NotAPlotError(out)
     if isinstance(out.plot, list):
         raise DvcException("Multiple plots per data file not supported.")

dvc/repo/push.py

@@ -1,56 +1,17 @@
-from typing import TYPE_CHECKING, Optional
+from typing import TYPE_CHECKING, Optional, Sequence
 
-from dvc.exceptions import UploadError
+from dvc.exceptions import InvalidArgumentError, UploadError
 
 from ..utils import glob_targets
 from . import locked
 
 if TYPE_CHECKING:
+    from dvc.cloud import Remote
+    from dvc.repo import Repo
+    from dvc.types import TargetType
     from dvc_objects.db.base import ObjectDB
 
 
-def _push_worktree(repo, remote):
-    from dvc_data.index import checkout
-    from dvc_data.index.save import build_tree
-
-    index = repo.index.data["repo"]
-    checkout(index, remote.path, remote.fs)
-
-    for stage in repo.index.stages:
-        for out in stage.outs:
-            if not out.use_cache:
-                continue
-
-            if not out.is_in_repo:
-                continue
-
-            workspace, key = out.index_key
-            index = repo.index.data[workspace]
-            entry = index[key]
-            if out.isdir():
-                old_tree = out.get_obj()
-                entry.hash_info = old_tree.hash_info
-                entry.meta = out.meta
-                for subkey, entry in index.iteritems(key):
-                    if entry.meta.isdir:
-                        continue
-                    fs_path = repo.fs.path.join(repo.root_dir, *subkey)
-                    _, hash_info = old_tree.get(
-                        repo.fs.path.relparts(fs_path, out.fs_path)
-                    )
-                    entry.hash_info = hash_info
-                tree_meta, new_tree = build_tree(index, key)
-                out.obj = new_tree
-                out.hash_info = new_tree.hash_info
-                out.meta = tree_meta
-            else:
-                out.hash_info = entry.hash_info
-                out.meta = entry.meta
-        stage.dvcfile.dump(stage, with_files=True, update_pipeline=False)
-
-    return len(index)
-
-
 @locked
 def push(
     self,
@@ -68,52 +29,91 @@ def push(
     odb: Optional["ObjectDB"] = None,
     include_imports=False,
 ):
-    _remote = self.cloud.get_remote(name=remote)
-    if _remote.worktree:
-        return _push_worktree(self, _remote)
 
+    worktree_remote: Optional["Remote"] = None
+    _remote = self.cloud.get_remote(name=remote)
+    if _remote.worktree or _remote.fs.version_aware:
+        worktree_remote = _remote
+
+    pushed = 0
     used_run_cache = (
         self.stage_cache.push(remote, odb=odb) if run_cache else []
     )
+    pushed += len(used_run_cache)
 
     if isinstance(targets, str):
         targets = [targets]
 
     expanded_targets = glob_targets(targets, glob=glob)
 
-    used = self.used_objs(
-        expanded_targets,
-        all_branches=all_branches,
-        all_tags=all_tags,
-        all_commits=all_commits,
-        with_deps=with_deps,
-        force=True,
-        remote=remote,
-        jobs=jobs,
-        recursive=recursive,
-        used_run_cache=used_run_cache,
-        revs=revs,
-    )
-
-    pushed = len(used_run_cache)
-    if odb:
-        all_ids = set()
-        for dest_odb, obj_ids in used.items():
-            if not include_imports and dest_odb and dest_odb.read_only:
-                continue
-            all_ids.update(obj_ids)
-        result = self.cloud.push(all_ids, jobs, remote=remote, odb=odb)
-        if result.failed:
-            raise UploadError(len(result.failed))
-        pushed += len(result.transferred)
+    if worktree_remote is not None:
+        pushed += _push_worktree(
+            self,
+            worktree_remote,
+            revs=revs,
+            all_branches=all_branches,
+            all_tags=all_tags,
+            all_commits=all_commits,
+            targets=expanded_targets,
+            jobs=jobs,
+            with_deps=with_deps,
+            recursive=recursive,
+        )
     else:
-        for dest_odb, obj_ids in used.items():
-            if dest_odb and dest_odb.read_only:
-                continue
-            result = self.cloud.push(
-                obj_ids, jobs, remote=remote, odb=odb or dest_odb
-            )
+        used = self.used_objs(
+            expanded_targets,
+            all_branches=all_branches,
+            all_tags=all_tags,
+            all_commits=all_commits,
+            with_deps=with_deps,
+            force=True,
+            remote=remote,
+            jobs=jobs,
+            recursive=recursive,
+            used_run_cache=used_run_cache,
+            revs=revs,
+            push=True,
+        )
+
+        if odb:
+            all_ids = set()
+            for dest_odb, obj_ids in used.items():
+                if not include_imports and dest_odb and dest_odb.read_only:
+                    continue
+                all_ids.update(obj_ids)
+            result = self.cloud.push(all_ids, jobs, remote=remote, odb=odb)
             if result.failed:
                 raise UploadError(len(result.failed))
             pushed += len(result.transferred)
+        else:
+            for dest_odb, obj_ids in used.items():
+                if dest_odb and dest_odb.read_only:
+                    continue
+                result = self.cloud.push(
+                    obj_ids, jobs, remote=remote, odb=odb or dest_odb
+                )
+                if result.failed:
+                    raise UploadError(len(result.failed))
+                pushed += len(result.transferred)
     return pushed
+
+
+def _push_worktree(
+    repo: "Repo",
+    remote: "Remote",
+    revs: Optional[Sequence[str]] = None,
+    all_branches: bool = False,
+    all_tags: bool = False,
+    all_commits: bool = False,
+    targets: Optional["TargetType"] = None,
+    jobs: Optional[int] = None,
+    **kwargs,
+) -> int:
+    from dvc.repo.worktree import push_worktree
+
+    if revs or all_branches or all_tags or all_commits:
+        raise InvalidArgumentError(
+            "Multiple rev push is unsupported for cloud versioned remotes"
+        )
+
+    return push_worktree(repo, remote, targets=targets, jobs=jobs, **kwargs)

dvc/repo/reproduce.py

@@ -1,13 +1,14 @@
 import logging
-import typing
 from functools import partial
+from typing import TYPE_CHECKING, Iterator
 
 from dvc.exceptions import DvcException, ReproductionError
 from dvc.repo.scm_context import scm_context
+from dvc.stage.exceptions import CheckpointKilledError
 
 from . import locked
 
-if typing.TYPE_CHECKING:
+if TYPE_CHECKING:
     from dvc.stage import Stage
 
     from . import Repo
@@ -58,7 +59,7 @@ def _dump_stage(stage):
     dvcfile.dump(stage, update_pipeline=False)
 
 
-def _get_stage_files(stage: "Stage") -> typing.Iterator[str]:
+def _get_stage_files(stage: "Stage") -> Iterator[str]:
     yield stage.dvcfile.relpath
     for dep in stage.deps:
         if (
@@ -72,12 +73,6 @@ def _get_stage_files(stage: "Stage") -> typing.Iterator[str]:
     for out in stage.outs:
         if not out.use_scm_ignore and out.is_in_repo:
             yield out.fs_path
-        if out.live:
-            from dvc.repo.live import summary_fs_path
-
-            summary = summary_fs_path(out)
-            if summary:
-                yield summary
 
 
 def _track_stage(stage: "Stage") -> None:
@@ -189,7 +184,8 @@ def _reproduce_stages(
     # `ret` is used to add a cosmetic newline.
     ret = []
     checkpoint_func = kwargs.pop("checkpoint_func", None)
-    for stage in steps:
+
+    for i, stage in enumerate(steps):
         if ret:
             logger.info("")
 
@@ -198,8 +194,6 @@ def _reproduce_stages(
                 _repro_callback, checkpoint_func, unchanged
             )
 
-        from dvc.stage.monitor import CheckpointKilledError
-
         try:
             ret = _reproduce_stage(stage, **kwargs)
 
@@ -216,7 +210,18 @@ def _reproduce_stages(
             if ret:
                 result.extend(ret)
         except CheckpointKilledError:
-            raise
+            result.append(stage)
+            logger.warning(
+                "Checkpoint stage '%s' was interrupted remaining stages in "
+                "pipeline will not be reproduced.",
+                stage.addressing,
+            )
+            logger.warning(
+                "skipped stages '%s'",
+                ", ".join(s.addressing for s in steps[i + 1 :]),
+            )
+
+            break
         except Exception as exc:
             raise ReproductionError(stage.addressing) from exc
 

dvc/repo/stage.py

@@ -23,6 +23,7 @@ from dvc.exceptions import (
     OutputNotFoundError,
 )
 from dvc.repo import lock_repo
+from dvc.ui import ui
 from dvc.utils import as_posix, parse_target, relpath
 
 logger = logging.getLogger(__name__)
@@ -139,7 +140,14 @@ class StageLoad:
         )
         with self.repo.scm_context:
             stage.dump(update_lock=update_lock)
-            stage.ignore_outs()
+            try:
+                stage.ignore_outs()
+            except FileNotFoundError as exc:
+                ui.warn(
+                    f"Could not create .gitignore entry in {exc.filename}."
+                    " DVC will attempt to create .gitignore entry again when"
+                    " the stage is run."
+                )
 
         return stage
 
@@ -361,7 +369,7 @@ class StageLoad:
             glob: Use `target` as a pattern to match stages in a file.
         """
         if not target:
-            return list(graph) if graph else list(self.repo.index)
+            return list(graph) if graph else self.repo.index.stages
 
         if recursive and self.fs.isdir(target):
             from dvc.repo.graph import collect_inside_path
@@ -402,7 +410,7 @@ class StageLoad:
             (see `collect()` for other arguments)
         """
         if not target:
-            return [StageInfo(stage) for stage in self.repo.index]
+            return [StageInfo(stage) for stage in self.repo.index.stages]
 
         target = as_posix(target)
 
@@ -502,3 +510,92 @@ class StageLoad:
 
     def collect_repo(self, onerror: Callable[[str, Exception], None] = None):
         return list(self._collect_repo(onerror))
+
+    def _load_file(self, path):
+        from dvc.dvcfile import Dvcfile
+        from dvc.stage.loader import SingleStageLoader, StageLoader
+
+        path = self._get_filepath(path)
+        dvcfile = Dvcfile(self.repo, path)
+        stages = dvcfile.stages  # type: ignore
+
+        if isinstance(stages, SingleStageLoader):
+            stages_ = [stages[None]]
+        else:
+            assert isinstance(stages, StageLoader)
+            keys = self._get_keys(stages)
+            stages_ = [stages[key] for key in keys]
+        return stages_, dvcfile
+
+    def _collect_all_from_repo(
+        self, onerror: Callable[[str, Exception], None] = None
+    ):
+        """Collects all of the stages present in the DVC repo.
+
+        Args:
+            onerror (optional): callable that will be called with two args:
+                the filepath whose collection failed and the exc instance.
+                It can report the error to continue with the collection
+                (and, skip failed ones), or raise the exception to abort
+                the collection.
+        """
+        from dvc.dvcfile import is_valid_filename
+        from dvc.fs import LocalFileSystem
+
+        scm = self.repo.scm
+        sep = self.fs.sep
+        outs: Set[str] = set()
+
+        is_local_fs = isinstance(self.fs, LocalFileSystem)
+
+        def is_ignored(path):
+            # apply only for the local fs
+            return is_local_fs and scm.is_ignored(path)
+
+        def is_dvcfile_and_not_ignored(root, file):
+            return is_valid_filename(file) and not is_ignored(
+                f"{root}{sep}{file}"
+            )
+
+        def is_out_or_ignored(root, directory):
+            dir_path = f"{root}{sep}{directory}"
+            # trailing slash needed to check if a directory is gitignored
+            return dir_path in outs or is_ignored(f"{dir_path}{sep}")
+
+        walk_iter = self.repo.dvcignore.walk(self.fs, self.repo.root_dir)
+        if logger.isEnabledFor(logging.TRACE):  # type: ignore[attr-defined]
+            walk_iter = log_walk(walk_iter)
+
+        stages = []
+        metrics = {}
+        plots = {}
+        params = {}
+
+        for root, dirs, files in walk_iter:
+            dvcfile_filter = partial(is_dvcfile_and_not_ignored, root)
+            for file in filter(dvcfile_filter, files):
+                file_path = self.fs.path.join(root, file)
+                try:
+                    new_stages, dvcfile = self._load_file(file_path)
+                except DvcException as exc:
+                    if onerror:
+                        onerror(relpath(file_path), exc)
+                        continue
+                    raise
+
+                stages.extend(new_stages)
+                if new_metrics := dvcfile.metrics:
+                    metrics[file_path] = new_metrics
+                if new_plots := dvcfile.plots:
+                    plots[file_path] = new_plots
+                if new_params := dvcfile.params:
+                    params[file_path] = new_params
+
+                outs.update(
+                    out.fspath
+                    for stage in new_stages
+                    for out in stage.outs
+                    if out.protocol == "local"
+                )
+            dirs[:] = [d for d in dirs if not is_out_or_ignored(root, d)]
+        return stages, metrics, plots, params

dvc/repo/status.py

@@ -89,6 +89,7 @@ def _cloud_status(
         remote=remote,
         jobs=jobs,
         recursive=recursive,
+        push=True,
     )
 
     ret = {}

dvc/repo/update.py

@@ -1,7 +1,13 @@
+from typing import TYPE_CHECKING, List
+
 from dvc.exceptions import InvalidArgumentError
+from dvc.stage.exceptions import StageUpdateError
 
 from . import locked
 
+if TYPE_CHECKING:
+    from dvc.repo.stage import StageInfo
+
 
 @locked
 def update(
@@ -15,6 +21,7 @@ def update(
     jobs=None,
 ):
     from ..dvcfile import Dvcfile
+    from .worktree import update_worktree_stages
 
     if not targets:
         targets = [None]
@@ -28,15 +35,21 @@ def update(
         )
 
     if not to_remote and remote:
-        raise InvalidArgumentError(
-            "--remote can't be used without --to-remote"
-        )
+        if not self.cloud.get_remote(name=remote).worktree:
+            raise InvalidArgumentError(
+                "--remote can't be used without --to-remote"
+            )
 
-    stages = set()
-    for target in targets:
-        stages.update(self.stage.collect(target, recursive=recursive))
+    import_stages = set()
+    other_stage_infos: List["StageInfo"] = []
 
-    for stage in stages:
+    for stage_info in self.index.collect_targets(targets, recursive=recursive):
+        if stage_info.stage.is_import:
+            import_stages.add(stage_info.stage)
+        else:
+            other_stage_infos.append(stage_info)
+
+    for stage in import_stages:
         stage.update(
             rev,
             to_remote=to_remote,
@@ -47,4 +60,29 @@ def update(
         dvcfile = Dvcfile(self, stage.path)
         dvcfile.dump(stage)
 
+    if other_stage_infos:
+        remote_obj = self.cloud.get_remote(name=remote)
+        if not remote_obj.worktree:
+            raise StageUpdateError(other_stage_infos[0].stage.relpath)
+        if rev:
+            raise InvalidArgumentError(
+                "--rev can't be used with worktree update"
+            )
+        if no_download:
+            raise InvalidArgumentError(
+                "--no-download can't be used with worktree update"
+            )
+        if to_remote:
+            raise InvalidArgumentError(
+                "--to-remote can't be used with worktree update"
+            )
+        update_worktree_stages(
+            self,
+            other_stage_infos,
+            remote_obj,
+        )
+
+    stages = import_stages | {
+        stage_info.stage for stage_info in other_stage_infos
+    }
     return list(stages)

dvc/repo/worktree.py

@@ -0,0 +1,283 @@
+import logging
+from functools import partial
+from typing import TYPE_CHECKING, Any, Dict, Iterable, Optional, Union, cast
+
+from dvc.fs.callbacks import Callback
+
+if TYPE_CHECKING:
+    from dvc.cloud import Remote
+    from dvc.index import Index, IndexView
+    from dvc.output import Output
+    from dvc.repo import Repo
+    from dvc.repo.stage import StageInfo
+    from dvc.stage import Stage
+    from dvc.types import TargetType
+    from dvc_data.hashfile.meta import Meta
+    from dvc_data.hashfile.tree import Tree
+    from dvc_data.index import DataIndex, DataIndexView
+    from dvc_objects.fs.base import FileSystem
+
+logger = logging.getLogger(__name__)
+
+
+# for files, if our version's checksum (etag) matches the latest remote
+# checksum, we do not need to push, even if the version IDs don't match
+def _meta_checksum(fs: "FileSystem", meta: "Meta") -> Any:
+    if not meta or meta.isdir:
+        return meta
+    return getattr(meta, fs.PARAM_CHECKSUM)
+
+
+def worktree_view(
+    index: "Index",
+    targets: Optional["TargetType"] = None,
+    push: bool = False,
+    **kwargs,
+) -> "IndexView":
+    """Return view of data that can be stored in worktree remotes.
+
+    Args:
+        targets: Optional targets.
+        push: Whether the view should be restricted to pushable data only.
+
+    Additional kwargs will be passed into target collection.
+    """
+
+    def stage_filter(stage: "Stage") -> bool:
+        if push and stage.is_repo_import:
+            return False
+        return True
+
+    def outs_filter(out: "Output") -> bool:
+        if (
+            not out.is_in_repo
+            or not out.use_cache
+            or (push and not out.can_push)
+        ):
+            return False
+        return True
+
+    return index.targets_view(
+        targets,
+        stage_filter=stage_filter,
+        outs_filter=outs_filter,
+        **kwargs,
+    )
+
+
+def fetch_worktree(
+    repo: "Repo",
+    remote: "Remote",
+    targets: Optional["TargetType"] = None,
+    jobs: Optional[int] = None,
+    **kwargs,
+) -> int:
+    from dvc_data.index import save
+
+    view = worktree_view(repo.index, targets=targets, **kwargs)
+    index = view.data["repo"]
+    for key, entry in index.iteritems():
+        entry.fs = remote.fs
+        entry.path = remote.fs.path.join(
+            remote.path,
+            *key,
+        )
+    total = len(index)
+    with Callback.as_tqdm_callback(
+        unit="file", desc="Fetch", disable=total == 0
+    ) as cb:
+        cb.set_size(total)
+        return save(index, callback=cb, jobs=jobs)
+
+
+def push_worktree(
+    repo: "Repo",
+    remote: "Remote",
+    targets: Optional["TargetType"] = None,
+    jobs: Optional[int] = None,
+    **kwargs,
+) -> int:
+    from dvc_data.index import checkout
+
+    view = worktree_view(
+        repo.index,
+        push=True,
+        targets=targets,
+        **kwargs,
+    )
+    new_index = view.data["repo"]
+    if remote.worktree:
+        logger.debug("indexing latest worktree for '%s'", remote.path)
+        old_index = _build_worktree_index(repo, remote, view)
+        logger.debug("Pushing worktree changes to '%s'", remote.path)
+    else:
+        old_index = None
+        logger.debug("Pushing version-aware files to '%s'", remote.path)
+
+    if remote.worktree:
+        diff_kwargs: Dict[str, Any] = {
+            "meta_only": True,
+            "meta_cmp_key": partial(_meta_checksum, remote.fs),
+        }
+    else:
+        diff_kwargs = {}
+
+    total = len(new_index)
+    with Callback.as_tqdm_callback(
+        unit="file", desc="Push", disable=total == 0
+    ) as cb:
+        cb.set_size(total)
+        pushed = checkout(
+            new_index,
+            remote.path,
+            remote.fs,
+            old=old_index,
+            delete=remote.worktree,
+            callback=cb,
+            latest_only=remote.worktree,
+            jobs=jobs,
+            **diff_kwargs,
+        )
+    if pushed:
+        for stage in view.stages:
+            for out in stage.outs:
+                workspace, _key = out.index_key
+                _update_out_meta(out, repo.index.data[workspace])
+            stage.dvcfile.dump(stage, with_files=True, update_pipeline=False)
+    return pushed
+
+
+def _build_worktree_index(
+    repo: "Repo", remote: "Remote", view: "IndexView"
+) -> "DataIndex":
+    from dvc_data.index import DataIndex
+    from dvc_data.index.build import build_entries
+
+    index = DataIndex()
+    for out in view.outs:
+        _workspace, key = out.index_key
+        parts = out.fs.path.relparts(out.fs_path, repo.root_dir)
+        path = remote.fs.path.join(remote.path, *parts)
+        for entry in build_entries(path, remote.fs):
+            if not entry.key or entry.key == ("",):
+                entry.key = key
+            else:
+                entry.key = key + entry.key
+            index.add(entry)
+    return index
+
+
+def _update_out_meta(
+    out: "Output", index: Union["DataIndex", "DataIndexView"]
+):
+    from dvc_data.index.save import build_tree
+
+    _, key = out.index_key
+    entry = index[key]
+    repo = out.stage.repo
+    if out.isdir():
+        old_tree = cast("Tree", out.get_obj())
+        entry.hash_info = old_tree.hash_info
+        entry.meta = out.meta
+        for subkey, entry in index.iteritems(key):
+            if entry.meta.isdir:
+                continue
+            fs_path = repo.fs.path.join(repo.root_dir, *subkey)
+            meta, hash_info = old_tree.get(
+                repo.fs.path.relparts(fs_path, out.fs_path)
+            )
+            entry.hash_info = hash_info
+            if meta.version_id is not None:
+                # preserve existing version IDs for unchanged files in
+                # this dir (entry will have the latest remote version
+                # ID after checkout)
+                entry.meta = meta
+        tree_meta, new_tree = build_tree(index, key)
+        out.obj = new_tree
+        out.hash_info = new_tree.hash_info
+        out.meta = tree_meta
+    else:
+        if entry.hash_info:
+            out.hash_info = entry.hash_info
+        if out.meta.version_id is None:
+            out.meta = entry.meta
+
+
+def update_worktree_stages(
+    repo: "Repo",
+    stage_infos: Iterable["StageInfo"],
+    remote: "Remote",
+):
+    from dvc.repo.index import IndexView
+    from dvc_data.index import build
+
+    def outs_filter(out: "Output") -> bool:
+        return out.is_in_repo and out.use_cache and out.can_push
+
+    view = IndexView(
+        repo.index,
+        stage_infos,
+        outs_filter=outs_filter,
+    )
+    local_index = view.data["repo"]
+    logger.debug("indexing latest worktree for '%s'", remote.path)
+    remote_index = build(remote.path, remote.fs)
+    for stage in view.stages:
+        for out in stage.outs:
+            _workspace, key = out.index_key
+            if key not in remote_index:
+                logger.warning(
+                    "Could not update '%s', it does not exist in the remote",
+                    out,
+                )
+                continue
+            entry = remote_index[key]
+            if entry.meta and entry.meta.isdir:
+                if not any(
+                    subkey != key and subentry.meta and not subentry.meta.isdir
+                    for subkey, subentry in remote_index.iteritems(key)
+                ):
+                    logger.warning(
+                        "Could not update '%s', directory is empty in the "
+                        "remote",
+                        out,
+                    )
+                    continue
+            _fetch_out_changes(out, local_index, remote_index, remote)
+        stage.save(merge_versioned=True)
+        for out in stage.outs:
+            _update_out_meta(out, remote_index)
+        stage.dvcfile.dump(stage, with_files=True, update_pipeline=False)
+
+
+def _fetch_out_changes(
+    out: "Output",
+    local_index: "DataIndex",
+    remote_index: "DataIndex",
+    remote: "Remote",
+):
+    from dvc_data.index import DataIndex, checkout
+
+    _, key = out.index_key
+    old = DataIndex()
+    new = DataIndex()
+    for _, entry in local_index.iteritems(key):
+        old.add(entry)
+    for _, entry in remote_index.iteritems(key):
+        new.add(entry)
+    total = len(new)
+    with Callback.as_tqdm_callback(
+        unit="file", desc=f"Updating '{out}'", disable=total == 0
+    ) as cb:
+        cb.set_size(total)
+        checkout(
+            new,
+            out.repo.root_dir,
+            out.fs,
+            old=old,
+            delete=True,
+            update_meta=False,
+            meta_only=True,
+            meta_cmp_key=partial(_meta_checksum, remote.fs),
+            callback=cb,
+        )

dvc/rwlock.py

@@ -1,8 +1,10 @@
 import json
+import logging
 import os
 from collections import defaultdict
 from contextlib import contextmanager
 
+import psutil
 from voluptuous import Invalid, Optional, Required, Schema
 
 from .exceptions import DvcException
@@ -10,6 +12,9 @@ from .fs import localfs
 from .lock import make_lock
 from .utils import relpath
 
+logger = logging.getLogger(__name__)
+
+
 INFO_SCHEMA = {Required("pid"): int, Required("cmd"): str}
 
 SCHEMA = Schema(
@@ -19,6 +24,9 @@ SCHEMA = Schema(
     }
 )
 
+RWLOCK_FILE = "rwlock"
+RWLOCK_LOCK = "rwlock.lock"
+
 
 class RWLockFileCorruptedError(DvcException):
     def __init__(self, path):
@@ -35,10 +43,10 @@ class RWLockFileFormatError(DvcException):
 
 @contextmanager
 def _edit_rwlock(lock_dir, fs, hardlink):
-    path = fs.path.join(lock_dir, "rwlock")
+    path = fs.path.join(lock_dir, RWLOCK_FILE)
 
     rwlock_guard = make_lock(
-        fs.path.join(lock_dir, "rwlock.lock"),
+        fs.path.join(lock_dir, RWLOCK_LOCK),
         tmp_dir=lock_dir,
         hardlink_lock=hardlink,
     )
@@ -65,35 +73,67 @@ def _infos_to_str(infos):
     )
 
 
-def _check_blockers(lock, info, *, mode, waiters):
+def _check_blockers(tmp_dir, lock, info, *, mode, waiters):
     from .lock import LockError
 
-    for waiter_path in waiters:
-        blockers = [
-            blocker
-            for path, infos in lock[mode].items()
-            if localfs.path.overlaps(waiter_path, path)
-            for blocker in (infos if isinstance(infos, list) else [infos])
-            if blocker != info
-        ]
+    non_existing_pid = set()
 
-        if not blockers:
+    blockers = []
+    to_release = defaultdict(list)
+    for path, infos in lock[mode].items():
+        for waiter_path in waiters:
+            if localfs.path.overlaps(waiter_path, path):
+                break
+        else:
             continue
 
+        infos = infos if isinstance(infos, list) else [infos]
+        for blocker in infos:
+            if blocker == info:
+                continue
+
+            pid = int(blocker["pid"])
+
+            if pid in non_existing_pid:
+                pass
+            elif psutil.pid_exists(pid):
+                blockers.append(blocker)
+                continue
+            else:
+                non_existing_pid.add(pid)
+                cmd = blocker["cmd"]
+                logger.warning(
+                    "Process '%s' with (Pid %s), in RWLock-file '%s'"
+                    " had been killed. Auto remove it from the lock file.",
+                    cmd,
+                    pid,
+                    relpath(path),
+                )
+            to_release[json.dumps(blocker, sort_keys=True)].append(path)
+
+    if to_release:
+        for info_json, path_list in to_release.items():
+            info = json.loads(info_json)
+            if mode == "read":
+                _release_read(lock, info, path_list)
+            elif mode == "write":
+                _release_write(lock, info, path_list)
+
+    if blockers:
         raise LockError(
-            "'{path}' is busy, it is being blocked by:\n"
-            "{blockers}\n"
+            f"'{waiter_path}' is busy, it is being blocked by:\n"
+            f"{_infos_to_str(blockers)}\n"
             "\n"
             "If there are no processes with such PIDs, you can manually "
-            "remove '.dvc/tmp/rwlock' and try again.".format(
-                path=waiter_path, blockers=_infos_to_str(blockers)
-            )
+            f"remove '{tmp_dir}/rwlock' and try again."
         )
 
 
 def _acquire_read(lock, info, paths):
     changes = []
 
+    lock["read"] = lock.get("read", defaultdict(list))
+
     for path in paths:
         readers = lock["read"][path]
         if info in readers:
@@ -108,6 +148,8 @@ def _acquire_read(lock, info, paths):
 def _acquire_write(lock, info, paths):
     changes = []
 
+    lock["write"] = lock.get("write", defaultdict(dict))
+
     for path in paths:
         if lock["write"][path] == info:
             continue
@@ -164,8 +206,10 @@ def rwlock(tmp_dir, fs, cmd, read, write, hardlink):
 
     with _edit_rwlock(tmp_dir, fs, hardlink) as lock:
 
-        _check_blockers(lock, info, mode="write", waiters=read + write)
-        _check_blockers(lock, info, mode="read", waiters=write)
+        _check_blockers(
+            tmp_dir, lock, info, mode="write", waiters=read + write
+        )
+        _check_blockers(tmp_dir, lock, info, mode="read", waiters=write)
 
         rchanges = _acquire_read(lock, info, read)
         wchanges = _acquire_write(lock, info, write)

dvc/schema.py

@@ -51,6 +51,7 @@ OUT_PSTAGE_DETAILED_SCHEMA = {
         Output.PARAM_PERSIST: bool,
         Output.PARAM_CHECKPOINT: bool,
         Output.PARAM_REMOTE: str,
+        Output.PARAM_PUSH: bool,
     }
 }
 
@@ -67,13 +68,6 @@ PLOT_PROPS = {
 PLOT_PROPS_SCHEMA = {**OUT_PSTAGE_DETAILED_SCHEMA[str], **PLOT_PROPS}
 PLOT_PSTAGE_SCHEMA = {str: Any(PLOT_PROPS_SCHEMA, [PLOT_PROPS_SCHEMA])}
 
-LIVE_PROPS = {
-    Output.PARAM_LIVE_SUMMARY: bool,
-    Output.PARAM_LIVE_HTML: bool,
-}
-LIVE_PROPS_SCHEMA = {**PLOT_PROPS_SCHEMA, **LIVE_PROPS}
-LIVE_PSTAGE_SCHEMA = {str: LIVE_PROPS_SCHEMA}
-
 PARAM_PSTAGE_NON_DEFAULT_SCHEMA = {str: [str]}
 
 VARS_SCHEMA = [str, dict]
@@ -93,7 +87,6 @@ STAGE_DEFINITION = {
         Any(str, OUT_PSTAGE_DETAILED_SCHEMA)
     ],
     Optional(StageParams.PARAM_PLOTS): [Any(str, PLOT_PSTAGE_SCHEMA)],
-    Optional(StageParams.PARAM_LIVE): Any(str, LIVE_PSTAGE_SCHEMA),
 }
 
 
@@ -111,7 +104,7 @@ def either_or(primary, fallback, fallback_includes=None):
 
 
 PLOT_DEFINITION = {
-    Output.PARAM_PLOT_X: str,
+    Output.PARAM_PLOT_X: Any(str, {str: str}),
     Output.PARAM_PLOT_Y: Any(str, [str], {str: Any(str, [str])}),
     Output.PARAM_PLOT_X_LABEL: str,
     Output.PARAM_PLOT_Y_LABEL: str,
@@ -127,9 +120,11 @@ SINGLE_PIPELINE_STAGE_SCHEMA = {
     str: either_or(STAGE_DEFINITION, FOREACH_IN, [FOREACH_KWD, DO_KWD])
 }
 MULTI_STAGE_SCHEMA = {
-    PLOTS: Any(SINGLE_PLOT_SCHEMA, [SINGLE_PLOT_SCHEMA]),
+    PLOTS: Any(SINGLE_PLOT_SCHEMA, [Any(str, SINGLE_PLOT_SCHEMA)]),
     STAGES: SINGLE_PIPELINE_STAGE_SCHEMA,
     VARS_KWD: VARS_SCHEMA,
+    StageParams.PARAM_PARAMS: [str],
+    StageParams.PARAM_METRICS: [str],
 }
 
 COMPILED_SINGLE_STAGE_SCHEMA = Schema(SINGLE_STAGE_SCHEMA)

dvc/scm.py

@@ -147,7 +147,7 @@ def clone(url: str, to_path: str, **kwargs):
                 fetch_all_exps(git, url, progress=pbar.update_git)
             return git
         except InternalCloneError as exc:
-            raise CloneError(str(exc))
+            raise CloneError("SCM error") from exc
 
 
 def resolve_rev(scm: "Git", rev: str) -> str:

dvc/stage/__init__.py

@@ -113,10 +113,11 @@ def restore_fields(stage):
     stage.meta = old.meta
     stage.desc = old.desc
 
-    old_fields = {out.def_path: (out.annot, out.remote) for out in old.outs}
+    old_outs = {out.def_path: out for out in old.outs}
     for out in stage.outs:
-        if out_fields := old_fields.get(out.def_path, None):
-            out.annot, out.remote = out_fields
+        old_out = old_outs.get(out.def_path, None)
+        if old_out is not None:
+            out.restore_fields(old_out)
 
 
 class Stage(params.StageParams):
@@ -280,22 +281,7 @@ class Stage(params.StageParams):
 
     def _read_env(self, out, checkpoint_func=None) -> Env:
         env: Env = {}
-        if out.live:
-            from dvc.env import DVCLIVE_HTML, DVCLIVE_PATH, DVCLIVE_SUMMARY
-            from dvc.output import Output
-            from dvc.schema import LIVE_PROPS
-
-            env[DVCLIVE_PATH] = relpath(out.fs_path, self.wdir)
-            if isinstance(out.live, dict):
-                config = project(out.live, LIVE_PROPS)
-
-                env[DVCLIVE_SUMMARY] = str(
-                    int(config.get(Output.PARAM_LIVE_SUMMARY, True))
-                )
-                env[DVCLIVE_HTML] = str(
-                    int(config.get(Output.PARAM_LIVE_HTML, True))
-                )
-        elif out.checkpoint and checkpoint_func:
+        if out.checkpoint and checkpoint_func:
             from dvc.env import DVC_CHECKPOINT
 
             env.update({DVC_CHECKPOINT: "1"})
@@ -379,7 +365,7 @@ class Stage(params.StageParams):
     def remove_outs(self, ignore_remove=False, force=False):
         """Used mainly for `dvc remove --outs` and :func:`Stage.reproduce`."""
         for out in self.outs:
-            if (out.persist or out.checkpoint or out.live) and not force:
+            if (out.persist or out.checkpoint) and not force:
                 out.unprotect()
                 continue
 
@@ -477,13 +463,21 @@ class Stage(params.StageParams):
         logger.debug("Computed %s md5: '%s'", self, m)
         return m
 
-    def save(self, allow_missing=False):
+    def save(
+        self,
+        allow_missing: bool = False,
+        merge_versioned: bool = False,
+        run_cache: bool = True,
+    ):
         self.save_deps(allow_missing=allow_missing)
 
-        self.save_outs(allow_missing=allow_missing)
+        self.save_outs(
+            allow_missing=allow_missing, merge_versioned=merge_versioned
+        )
         self.md5 = self.compute_md5()
 
-        self.repo.stage_cache.save(self)
+        if run_cache:
+            self.repo.stage_cache.save(self)
 
     def save_deps(self, allow_missing=False):
         from dvc.dependency.base import DependencyDoesNotExistError
@@ -495,15 +489,40 @@ class Stage(params.StageParams):
                 if not allow_missing:
                     raise
 
-    def save_outs(self, allow_missing=False):
+    def save_outs(
+        self, allow_missing: bool = False, merge_versioned: bool = False
+    ):
         from dvc.output import OutputDoesNotExistError
 
+        from .exceptions import StageFileDoesNotExistError, StageNotFound
+
+        if merge_versioned:
+            try:
+                old = self.reload()
+                old_outs = {out.def_path: out for out in old.outs}
+                merge_versioned = any(
+                    (
+                        out.files is not None
+                        or (
+                            out.meta is not None
+                            and out.meta.version_id is not None
+                        )
+                    )
+                    for out in old_outs.values()
+                )
+            except (StageFileDoesNotExistError, StageNotFound):
+                merge_versioned = False
+
         for out in self.outs:
             try:
                 out.save()
             except OutputDoesNotExistError:
                 if not (allow_missing or out.checkpoint):
                     raise
+            if merge_versioned:
+                old_out = old_outs.get(out.def_path)
+                if old_out is not None:
+                    out.merge_version_meta(old_out)
 
     def ignore_outs(self):
         for out in self.outs:
@@ -571,7 +590,9 @@ class Stage(params.StageParams):
         if not dry:
             if kwargs.get("checkpoint_func", None) or no_download:
                 allow_missing = True
-            self.save(allow_missing=allow_missing)
+
+            self.save(allow_missing=allow_missing, run_cache=not no_commit)
+
             if no_download:
                 self.ignore_outs()
             if not no_commit:
@@ -762,6 +783,7 @@ class PipelineStage(Stage):
         return f"{super().addressing}:{self.name}"
 
     def reload(self):
+        self.dvcfile._reset()  # pylint: disable=protected-access
         return self.dvcfile.stages[self.name]
 
     def _status_stage(self, ret):

dvc/stage/cache.py

@@ -180,7 +180,7 @@ class StageCache:
         dump_yaml(tmp, cache)
         self.repo.odb.local.move(tmp, path)
 
-    def restore(self, stage, run_cache=True, pull=False):
+    def restore(self, stage, run_cache=True, pull=False, dry=False):
         from .serialize import to_single_stage_lockfile
 
         if not _can_hash(stage):
@@ -195,14 +195,15 @@ class StageCache:
         else:
             if not run_cache:  # backward compatibility
                 raise RunCacheNotFoundError(stage)
-            stage.save_deps()
+            if not dry:
+                stage.save_deps()
             cache = self._load(stage)
             if not cache:
                 raise RunCacheNotFoundError(stage)
 
         cached_stage = self._create_stage(cache, wdir=stage.wdir)
 
-        if pull:
+        if pull and not dry:
             for objs in cached_stage.get_used_objs().values():
                 self.repo.cloud.pull(objs)
 
@@ -213,7 +214,8 @@ class StageCache:
             "Stage '%s' is cached - skipping run, checking out outputs",
             stage.addressing,
         )
-        cached_stage.checkout()
+        if not dry:
+            cached_stage.checkout()
 
     def transfer(self, from_odb, to_odb):
         from dvc.fs import HTTPFileSystem, LocalFileSystem
@@ -265,11 +267,15 @@ class StageCache:
         return ret
 
     def push(self, remote: Optional[str], odb: Optional["ObjectDB"] = None):
-        dest_odb = odb or self.repo.cloud.get_remote_odb(remote)
+        dest_odb = odb or self.repo.cloud.get_remote_odb(
+            remote, "push --run-cache"
+        )
         return self.transfer(self.repo.odb.local, dest_odb)
 
     def pull(self, remote: Optional[str], odb: Optional["ObjectDB"] = None):
-        odb = odb or self.repo.cloud.get_remote_odb(remote)
+        odb = odb or self.repo.cloud.get_remote_odb(
+            remote, "fetch --run-cache"
+        )
         return self.transfer(odb, self.repo.odb.local)
 
     def get_used_objs(self, used_run_cache, *args, **kwargs):

dvc/stage/exceptions.py

@@ -9,6 +9,14 @@ class StageCmdFailedError(DvcException):
         super().__init__(msg)
 
 
+class CheckpointKilledError(DvcException):
+    def __init__(self, cmd, status=None):
+        msg = f"failed to finish: {cmd}"
+        if status is not None:
+            msg += f", exited with {status}"
+        super().__init__(msg)
+
+
 class StageFileDoesNotExistError(DvcException):
     DVC_IGNORED = "is dvc-ignored"
     DOES_NOT_EXIST = "does not exist"

dvc/stage/imports.py

@@ -21,22 +21,20 @@ def update_import(
     stage, rev=None, to_remote=False, remote=None, no_download=None, jobs=None
 ):
     stage.deps[0].update(rev=rev)
-    outs = stage.outs
-    deps = stage.deps
 
     frozen = stage.frozen
     stage.frozen = False
+    changed = stage.changed()
 
-    if stage.outs:
-        stage.outs[0].clear()
     try:
         if to_remote:
             _update_import_on_remote(stage, remote, jobs)
         else:
             stage.reproduce(no_download=no_download, jobs=jobs)
     finally:
-        if deps == stage.deps:
-            stage.outs = outs
+        if no_download and changed:
+            # Avoid retaining stale information
+            stage.outs[0].clear()
         stage.frozen = frozen