MLFLOW_EXPERIMENT_NAME env variable and --experiment-name arguments for CLI command #889
Conversation
… for CLI command - get_experiment_by_name() for SQL store - tests for env variables (MLFLOW_EXPERIMENT_NAME and MLFLOW_EXPERIMENT_ID)
…ridden 'get_experiment_by_name' method )
| @@ -50,6 +50,16 @@ def get_experiment(self, experiment_id): | |||
| """ | |||
| pass | |||
|
|
|||
| @abstractmethod | |||
| def get_experiment_by_name(self, experiment_name): | |||
There was a problem hiding this comment.
It looks like all three implementations of this method invoke list_experiments() and match against the provided experiment name. That seems to suggest that we can provide an implementation of the method in the base class. If this implementation is not performant enough for a given use case, the implementor can always override the method.
I noticed that the file_store implementation makes an additional call to _check_root_dir prior to calling list_experiments. However, list_experiments immediately calls _check_root_dir, so this call seems to be superfluous: all three implementations seem to share the same structure.
There was a problem hiding this comment.
Like your idea of base implementation or adding specific implementations for each.
Will take care of this in a follow on change since this is a PR for --experiment-name change (which needed this get_experiment_by_name method for completeness)
| Throws an exception if experiment is not found or permanently deleted. | ||
|
|
||
| :param experiment_name: Name of experiment | ||
| :return: A single Experiment object if it exists, otherwise raises an Exception. |
There was a problem hiding this comment.
Extra credit: is it possible to put some sphinx markup here so it links to the docs for the Experiment object?
There was a problem hiding this comment.
Not sure. Lemme explore that asynchronously...
There was a problem hiding this comment.
Yes! :ref:Experiment <fully_qualified_module_name_here>.
… has a specialized implementation. - Added sphinx markup for AbstractStore for pydocs ref: addressing comments in PR mlflow#889
* Basic implementation of "get_experiment_by_name" to base class, SQL has a specialized implementation. * Added sphinx markup for AbstractStore for pydocs ref: addressing comments in PR #889
MLFLOW_EXPERIMENT_NAMEenv variable and--experiment-namearguments for CLI commandget_experiment_by_namemethod for SQL storeMLFLOW_EXPERIMENT_NAMEandMLFLOW_EXPERIMENT_ID)mlflow runCLI command (tested only new options)