* Remove all of the experiments in the workspace
fix#5676
1. remove all the experiments in workspace
2. add tests for it
* Only remove baseline ones
* Add a new arguments --all
* Update dvc/repo/experiments/remove.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* Rename and shortcut for these commands
* Solve the issue from removing a changing list.
* Update dvc/repo/experiments/remove.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* Update dvc/repo/experiments/remove.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* delete workspace option
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
Replacing DVC_ANALYTICS. This is following our other env conventions and
will not have effect if set to false/no/etc, and will only work if set
to true/yes/y/etc. Will disable analytics even if it is explicitly
enabled in our configs.
Kudos @skshetry
* add support for Python 3.10
pyarrow is not available, so we are skipping tests for it.
And, restricting `hdfs` extras to be not installed in Python3.10.
Pathlib removed `_init` method and `init` parameter, so just working
around that. It's not a big change.
* skip windows on 3.10; update pip/wheel/setuptools during install
pywin32 is not yet available for 3.10 Python, so skipping those for now
* install libgit2 with Python3.10 as pygit2 wheel is not available
* install libgit2
* only run on macOS, libgit2-1.1 not available on ubuntu
* Remove a special queued experiments
fix#6157
1. dvc exp remove not accept queued experiments name
2. add some tests for this feature
* Extract tests and add revision support
1. Extract remove experiments to a new file.
2. revision can be used to remove special queued experiment
* Accept shortened revisions
* Split removing committed and queued exp functions
* Api name change
* Return to the old API name
* shorten some functions
* Error message change
* Better test cases, more corner case
* Still raise exception in a mixed case
* Push running checkpoint to remote
Fix#6182
* Update dvc/repo/experiments/executor/base.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* Get the branch name
* Finish this PR
1. move env name to dvc/env.py.
2. add some tests for it.
* Update after finally commit
* Some problems found in the review
1. change the behavior of self remote
2. do not use string DVC_EXP_AUTO_PUSH
3. downgrade the logger level in auto push
4. use full branch ref
* Update dvc/repo/experiments/executor/base.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* add a remote check before auto push
* Split current one env into two
* Rename DVC_EXP_CHECKPOINT_PUSH to DVC_EXP_AUTO_PUSH
* value error
* Remove getenv_bool switch to env2bool
* Better on handling Dulwich exceptions
* Name changing removing prints
* Update dvc/env.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* Some changes in review
* Add validation to repo URL
* Add new type of SCMError
* Update dvc/repo/experiments/executor/base.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* Use spy tests call counts and args
* Update dvc/scm/git/backend/base.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* Update dvc/scm/git/backend/dulwich.py
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Remove validate_git_repo to validate_git_remote
Co-authored-by: Peter Rowlands (변기호) <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* demonstrate issue while sending analytics report
When sending analytics, we are not flushing the data to the temp file
which causes failure at times, especially on Windows
* close file before calling daemonized analytics send command
This way, the contents bufferred will get a chance to be flushed.
* objects.status: use object IDs instead of naive objects
* objects.transfer: use object IDs instead of naive objects
* objects: add ReferenceHashFile
* odb: add ReferenceObjectDB
* repo: collect used object IDs instead of naive objs
* support basic/naive fs serialization
* objects.stage: stage file objects in addition to trees
* objects: remove save(), update transfer()
* output/dep: update for osave removal
* path_info: support both url and path attributes
* status/gc: use object IDs instead of naive objs
* diff: update staging usage
* commit: update staging usage
* tests: clean staging when needed
* tests: update for save/transfer/stage changes
* dvcfs: don't allow config serialization (repofs should be used instead)
* utils: return mtime as str, size as int
* tests: update unit tests for save/stage/transfer changes
* objects: use Repo.open when needed in ReferenceHashFile
* fix ls test staging
* objects.stage: don't skip hash check when loading from state
- individual hashes will still be loaded from state if they are still
valid
* ReferenceHashFile: serialize repo fs paths relative to repo root
* preserve state when staging
* objects.tree: skip hash computation in digest() when we have already
loaded hash from state
* always check file existence in referencehashfile
* update repofs test since stage no longer allows nonexistent files
* odb: only protect on check() if hash was verified
* staging: use unique memfs namespace for staging per dest ODB
* tests: clean full staging namespace
* refodb: preserve repofs cache dir
* repodependency: stage imports to local ODB and strict check circular
imports
* fetch: fetch staging last
* staging: make dest odb mandatory and preserve state
* refodb: cache nonlocal fs's
* fix rebase errors
* repofs: support full instantiation from fs config
* repofs: preserve subrepo factory for erepos
* tests: clarify repofs dirty file behavior
* Fix merging of params when empty
Recent release of python-benedict==0.24.1 started ignoring empty dictionaries
during merge. The way we expected to work was questionable and also how
benedict itself tries to handle the references to empty dictionary is
also questionable.
I consider this to be a proper way to handle this even if this gets fixed
on the benedict side.
* Revert "setup: limit python-benedict version (#6377)"
This reverts commit 15de826266.
* add comment for why passing through benedict
* objects: migrate remote status to objects.status
* data_cloud: use objects.status
* objects: migrate remote._process to objects.transfer
* data_cloud: use objects.transfer
* objects: handle memfs staging objects in transfer
* fetch/import: use unified transfer()
* update get_remote and transfer exception usage
* update tests
* objects.transfer: use src ODB verify rules after xfer
* odb: move index from remote into odb
* objects.transfer: skip src status check when possible
* update tests
* remove dvc.remote
* odb.index: migrate to diskcache
* objects: fix status/transfer optimization
* update index usage
* state: remove dead sqlite related code
* update cloud tests
* drop unnecessary remove in tests
* use abstract base class in odb index
* status: load trees from local cache instead of remote odb when possible
* handle status when dir cache is explicitly removed
* params/metrics/plots/exp show: introduce onerror
* fixup
* plots/metrics/params: removing errors
* removing more errors
* removing errors
* some more removing
* new plots ux
* new ux for metrics
* new ux for params
* fixing tests
* fixups
* minor test fixes
* minor test fixes
* commands: plots/params: fix ui imports
* working on error handlng decorator, stopped at show checkpoints
* error_handler, further dvelopment
* plots just before moving loading to collect
* appending errors to results: initial passing
* fixup
* onerror: remove onerror fixture
* fixups
* plots/metrics/params/experiments: temporarily remove error warnings
* plots: extract parsing out of class method
* metrics/experiments: provide exception serializer for show-json
* params/plots/metrics: signal errors
* rebase fixups
* fix TODOs
* fixup
* skshetry's review refactor
* Migrate to ossfs
fix 6177
* Add verification to object stored on oss
* Change the oss remote's dependency.
pass all of the tests
* Performance optimaztion and ossfs version update
* Remove the argument to set the resumable get/put
1. Experiments shows resumable get/put has little influnence on performance.
2. Just use resumable ones as default
* Use a new api for performance optimization
* Without filesize progress bar version
* With progress bar version
* Bump and add comments
1. Bump ossfs from 2021.6.2 to 2021.7.0
2. add some comments to the verification trigger
* Change the name of kwrags following parent class.
* Bump to new ossfs and rewrite upload_fobj
Because some problem with oss-emulater
1. bump to ossfs 2021.7.1
2. rewrite upload_fobj
* Optimization on rm_files
* Some optimization follow the old version.
* Update dvc/fs/oss.py
Co-authored-by: Ruslan Kuprieiev <[email protected]>
* Update dvc/objects/db/oss.py
Co-authored-by: Ruslan Kuprieiev <[email protected]>
* bump into ossfs 2021.7.2
* Bump into ossfs 2021.7.3OC
* Switch to real_oss_server
* Switch to a real oss server
* Use the oss bucket located in US.
* Remove some emulator related code
* Add envs to ci
* remove some oss pytest XFAIL
Because we didn't use emulator, we can remove some pytest XFAILs
Co-authored-by: Ruslan Kuprieiev <[email protected]>
* exp: allow fetching from arbitrary executor by git URL
* exp: write info for running executors to .dvc/tmp/exp/run
currently only includes PID and git fetch URL
* exp: refactor stash/queue cleanup
* exp: add ability to collect dict of currently running experiments
* exp show: display run/queue state in the table
* exp run: unlock main repo while running tmpdir executors
* add tests for new exp show behavior
* fix workspace checkpoint run handling
* fix issue where running stash revs were shown as queued
* use "running"/"queued"
* add executor (location) column
* compare: add TabularData.is_empty
* use td.is_empty to drop state/executor
* use "workspace" and "temp" as executor names
* fs: use fsspecwrapper in memfs
* odb: add memfs staging ODB in ODBManager
* odb: add read_only attribute for odb instances
* objects: handle memfs paths when staging trees
* odb: remove git ODB
* output: use staging ODB
* RepoDependency: use staging odb for imports
* dvcfs: handle case where dir cache is unavailable
* fix circular import check
* odb: use temp local ODB for staging
* objects: use staging ODB in stage() and tree load
* objects: add typing for add/check/load/stage
* output/dep: use new staging
* add unit tests for local odb staging
* localfs: account for string paths in move()
* include local in odb.by_scheme
* tests: count dir cache like any other object in test_pull_git_imports
* odb: clear staging on gc()
* objects: add tree.get/filter/insert
* output: use obj.filter/get for granular dir commit
* squash filter logic output
* dependency: don't filter imports at the dep level
* don't init staging ODB for external cache
* use staging ODB when computing tree hash for external outs
* remove unneeded force in test granular commit
* use unique-per-manager memfs URLs for staging
* odb: make staging per-ODBManager
* objects: move state dependent staging into ODBManager
* objects: support checking multiple odbs
- staging dependent checks can be done via ODBManager.check
* checkout: do state lookup before explicit stage() in _changed()
* output: use odbmanager based staging
* RepoDependency: use odbmanager based staging
* update obj staging unit tests
* checkout: remove check(), handle individual file exceptions on link()
* diff: use odbmanager based staging
* objects: add ObjectPermissionError for read/write perm errors
* odb: move staging out of odbmanager
* objects: handle staging ODB in objects.stage()
* objects.tree: don't stage anything in tree methods
* odb: clear staged objects on gc()
* remote: don't filter objs by scheme, objs are already grouped by ODB
* diff: update objects.stage usage
* output/dep: update stage() usage
* update ODB tests
* update func tests
* remove memfs from FS_MAP
* revert unneeded objects.check() changes
* remove unused tree.insert()
* revert unused odbmanager changes
* add separate odb error classes
* stage external outs directly in ODB
* check for obj existence in main ODB before staging
* use .dvc/tmp/staging instead of per-odb staging
* gc: only gc loal staging once
* Fixed diff summary to only show states with values.
* Removed redundant comma by using join.
* Swithced for loop to generator for better readability.
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* add: do not delete stage files before add
Because of the way we collect stages and cache them, we were not able to
collect them for the `add` without removing them from the workspace.
As doing so, we'd have two same/similar stages - one collected from the
workspace and the other just created from the `dvc add` in-memory.
This would raise errors during graph checks, so we started to delete them
and reset them (which is very recently, see #2886 and #3349).
By deleting the file before we even do any checks, we are making DVC
fragile, and results in data loss for the users with even simple mistakes.
This should make it more reliable and robust.
And, recently, we have started to keep state of a lot of things, that by
resetting them on each stage, we waste a lot of performance, especially
on gitignores. We cache the dulwich's IgnoreManager, which when resetted
too many times, will waste a lot of our time just collecting them again
next time (see #6227).
It's hard to say how much this improves, as this very much depends on
no. of gitignores in the repo (which can be assumed to be quite in number
for a dvc repo) and the amount of files that we are adding
(eg: `-R` adding a large directory).
On a directory with 10,000 files (in a datadet-registry repo),
creating stages on `dvc add -R` went from 64 files/sec to 1.1k files/sec.
* add tests
* make the test more specific
Paweł Redzyński
GaoandPeter Rowlands
Saugat Pachhai
Ruslan Kuprieiev
Batuhan Taskaya
Nikita KodenkoandPeter Rowlands
dependabot[bot]andPeter Rowlands
Batuhan Taskaya
Christian Clauss
Hadas Arikandpre-commit-ci[bot]