* Add support for WebHDFS
* Style and linting fixes
* Implement abstract methods and disable no-member pylint errors
* Add init unit test
* Yield client and fix redefinition of hdfs
* Get dynamic hdfs Docker port
* Add support for open in WebHDFSTree
* Remove hdfscli_config fallbacks in WebHDFSTree
* Use port 50070 for webhdfs
* WebHDFSTree supports http URL and proper walking
* Fix failing unit test after refactoring
* Add overwrite argument to upload
* Add remaining func tests for webhdfs
* Remove print statements
* Allow dynamic attributes in pylint for hdfs library
* Comment error on instantiating HDFS client from config
* Remove depth argument from call to walk
* Update dvc/tree/webhdfs.py
* Remove unnecessary deque
* Check exception messages hdfs client failure from config
* Return client directly
Co-authored-by: Ruslan Kuprieiev <[email protected]>
* Split long line
* Add size to HashInfo
* Implement move via hdfs.rename
* Add missing line
* Make parent directories before moving file in WebHDFS
* Add total to progress bar
Co-authored-by: Ruslan Kuprieiev <[email protected]>
* tree: use file_mode in upload() for trees that support modes
* experiments: preserve file_mode for files in the workspace
* skip permissions tests on windows
* LocalTree: separate chmod from protect
* fix rebase issue
* experiments: Allow branching from checkpoints with modified hyper params
* experiments: initial checkpoints tree functionality for `exp show`
* go back to flat/sortable view
* add test for resume w/modified params
* tests: move experiments fixtures into conftest
* add show checkpoints test
* test checkpoint table rendering
* remove dead code
* test show for checkpoint branches
* api: add support for simple wildcards
Related to #4816.
Signed-off-by: Ioana Grigoropol <[email protected]>
* api: make wildcard interpretation optional
Adds a new argument for the add command `glob` that is disabled by default and when enabled it passes
the input targets through glob filtering.
Related: #4816
Signed-off-by: Ioana Grigoropol <[email protected]>
* Update dvc/repo/add.py
* Update dvc/repo/add.py
* Update dvc/repo/add.py
* Update dvc/repo/add.py
* Update dvc/repo/add.py
Co-authored-by: Ruslan Kuprieiev <[email protected]>
* experiments: refactor regular/checkpoint experiment separation
* fix scm_checkout pull bug
* experiments: return stages from executor.reproduce
* since executor has to move to threadpool execution anyways, we no
longer have to work around the pickle repro result limitation
* refactor new/resume behavior
* refactor checkpoints tests
* use -f to force checkpoints reset/restart
* checkpoints: use 'exp run' and 'exp res[ume]' commands
* stage: all outs can be missing for checkpoint stages
* regular outputs may not be written at the time a checkpoint is
generated, but subsequent checkouts of that stage should not fail due
to the missing outs
* commit stage on checkpoint generation otherwise non-persist cached outs
are lost
* checkpoints: wait for stage proc to end after killing via terminate()
* stage: fix issues with killing checkpoint stage
* refactor signaling (use threading.Event instead of Condition)
* fix windows bug (kill all child processes via taskkill)
* experiments: refactor executor collection
* tests: move checkpoint tests into separate module
* refactor experiment tests to use stage fixture
We wrap every data to one of the `Node` object. But, since we were
passing `self.data` instead of `self`, it was skipping the conversion
and instead was just trying to set data as-is, which would have make
our latter assumptions incorrect, as other codebases assume it to be
a `Node`.
I have added asserts in a few places. Also, I noticed that, on a few
test assumptions, when `resolving`, we were not converting values
back to it's original form (`CtxDict` -> `dict`, `CtxList` -> `list`
and `Value` -> _). So, I have added `node.value` property, as it is
a complement of the above fix.
This caused a ripple effect for the changes in the API as `foreach`
now resolved and changed the data to the original form,
which prevents us from tracking the data later in the "field".
So, `resolve`/`select` got a real `unwrap` option that should
reliably work for all kinds of `Node` objects.
Currently we are converting dir_info to/from lists all the time.
The reason is that dir_info is stored as list of dicts in *.dir files,
but that makes it hard to work with. In addition to that, we will likely
be changing .dir file format in the near future #829, so we need to
abstract away dir_info into something that we won't care how it will be
stored on disk.
Related #3256
Related #4847
* add: warn on cache link failure instead of error
* add test case
* fix linted string
Co-authored-by: Paweł Redzyński <[email protected]>
Co-authored-by: Paweł Redzyński <[email protected]>
* metrics: include revision information when collecting metrics files
* plots: test warning message on no plot on revision
* plots/metrics: move to one package, introduce common tests
* tests: fix utils placement
* Allow setting of list parameteres
Check if dict is Mapping or list before using .get()
* Added test for dvc exp run with list paramater
* Add more tests for string parameter modification
* Add test, and fix resolving of paths for params
Improve merging of vars and imported params, improve tracking
Resolve wdir correctly
* Fix windows relpath
* isort imports
* command: print the current cache directory when no cmd flags are provided
tests: corrected the test for cache dir cmomand
comamnd 'dvc cache dir' now prints the current cache dir
when no cmd flags are provided and modified existing test for
the same.
Fixes#1998
Signed-off-by: Girish Joshi <[email protected]>
* command: updated validation for args.value and args.unset flag
according to current convention in dvc/command/cache.py
Signed-off-by: Girish Joshi <[email protected]>
* Update dvc/command/cache.py
* dont forget logger
Co-authored-by: Ruslan Kuprieiev <[email protected]>
Co-authored-by: Ruslan Kuprieiev <[email protected]>
* state: add simple reentrance
* state: pass Repo instead of LocalCache
* state: use granularly
Users could still load state preemptively if they want to bulk up
the operations, but other than that tree/cache will just load it
when they need it.
* stage: make runs checkpointable
* api: add make_checkpoint API call
* tests: add test for checkpointed stage
* experiments: implement checkpoints callback chain
* initial working state
* use .dvc/tmp for checkpoint signal file location
* force repro for checkpoint experiments
* checkout: add flag to allow missing persistent outputs
(i.e. not yet created)
* use allow_persist_missing on checkpoint experiment repro
* executor: force checkout before checkpoints experiment repro
* move checkpoint commands to `dvc exp run`
* support resuming checkpoint runs with --continue
* fix git bugs
* experiments: include checkpoint commits in `exp show`
* use 1sec sleep timers
* fix apply_workspace test conflict
* tests: add test for `dvc exp run`
* cleanup signal file behavior
* fix bool conversion
* tests: add tests for checkpoint and checkpoint_continue
* fix styling
* fix checkpoint stage test
* fix checkpoint monitor thread condition/notification
* repro: support --pull
This option allows run-cache to pull the cache for the outputs
when restoring a stage instead of pulling everything at once during
`dvc pull --run-cache`.
Fixes#4223
* repro: adjust --pull help
Tomás Farías SantanaandRuslan Kuprieiev
Peter Rowlands (변기호)
Saugat Pachhai
Sami Jawhar
Ioana GrigoropolandRuslan Kuprieiev
Michal NowikowskiandRuslan Kuprieiev
Paweł Redzyński
Kevin Lloyd Bernal
Juliano Efson Sales
Tomás Farías Santana
Harsh Patel
