--- title: Version 1.9 --- ### ClearML 1.9.3 **Bug Fixes** * Fix broken `Task._get_status()`, which was breaking `clearml-session` in the latest version * Fix path substitution, making it possible to store unsubstituted URL for models ([ClearML GitHub PR #935](https://github.com/allegroai/clearml/pull/935)) ### ClearML 1.9.2 **New Features and Improvements** * Support parsing queue name when providing execution queue in pipelines code ([ClearML GitHub PR #857](https://github.com/allegroai/clearml/pull/857)) * Ignore `None` values for keys in the `click` argument parser ([ClearML GitHub issue #902](https://github.com/allegroai/clearml/issues/902)) * Improve docstrings for `Task.mark_completed()` and `Task.close()` (ClearML GitHub PRs [#920](https://github.com/allegroai/clearml/pull/920) and [#921](https://github.com/allegroai/clearml/pull/921)) * Add pre/post execution callbacks to pipeline steps through `@PipelineDecorator.component` * Add status-change callback to pipeline steps through `PipelineController.add_step()`, `PipelineController.add_function_step()`, and `@PipelineDecorator.component` **Bug Fixes** * Fix missing debug samples when reporting using TensorBoard ([ClearML GitHub issue #923](https://github.com/allegroai/clearml/issues/923)) * Fix wrong Jupyter API token during repository detection ([ClearML GitHub PR #904](https://github.com/allegroai/clearml/pull/904)) * Fix typo in the warning for very large git diffs ([ClearML GitHub PR #932](https://github.com/allegroai/clearml/pull/932)) * Fix pipelines from tasks don't propagate `parameter_override` values in `PipelineController.add_step()` * Fix folders and files uploaded to S3 and Azure with `StorageManager.upload_file()` receive wrong MIME types * Fix the CSV file preview in Datasets * Fix `Task.connect_configuration()` doesn't work with non-string dictionary keys * Fix `lightgbm_example` deprecation warning * Fix potential race condition in `get_or_create_project()` ### ClearML 1.9.1 **New Features and Improvements** * Add signature version to `boto3` configuration ([ClearML GitHub issue #883](https://github.com/allegroai/clearml/issues/883)) * Allow requesting custom token expiration using the `api.auth.req_token_expiration_sec` configuration setting * Add Python 3.11 support **Bug Fixes** * Fix `UniformParameterRange.to_list` throws error when step size is not defined ([ClearML GitHub issue #858](https://github.com/allegroai/clearml/issues/858)) * Fix `StorageManager.list()` does not return size metadata ([ClearML GitHub issue #865](https://github.com/allegroai/clearml/issues/865)) * Fix storage with path substitutions ([ClearML GitHub issue #825](https://github.com/allegroai/clearml/issues/825)) * Fix extras in ClearML installation prevents clearml from being included in requirements ([ClearML GitHub issue #867](https://github.com/allegroai/clearml/issues/867)) * Fix metadata set on an uploaded model object is not accessible ([ClearML GitHub issue #890](https://github.com/allegroai/clearml/issues/890)) * Fix Azure storage upload not working ([ClearML GitHub issue #868](https://github.com/allegroai/clearml/issues/868)) * Fix `task.connect` list of dicts parsed incorrectly in remote * Fix casting `None` to `int` fails uploads and permission checks * Fix numpy 1.24 support * Fix `clearml-data` previews are saved on file server even when `output_uri` is specified * Fix connecting a dictionary to task sometimes raises an exception * Fix authentication headers are not set on substituted fileserver URLs * Fix `Task.get_project_id()` cannot find hidden projects * Fix TriggerScheduler docstrings ([ClearML GitHub issue #881](https://github.com/allegroai/clearml/issues/881)) ### ClearML 1.9.0 **New Features and Improvements** * Add `r` prefix to `re.match()` strings ([ClearML GitHub issue #834](https://github.com/allegroai/clearml/issues/834)) * Add `path_substitution` to `clearml.conf` example file ([ClearML GitHub PR #842](https://github.com/allegroai/clearml/pull/842)) * Clarify `deferred_init` usage in `Task.init()` ([ClearML GitHub issue #855](https://github.com/allegroai/clearml/issues/855)) * Add pipeline decorator argument to control docker image ([ClearML GitHub issue #856](https://github.com/allegroai/clearml/issues/856)) * Add `StorageManager.set_report_upload_chunk_size()` and `StorageManager.set_report_download_chunk_size()` to set chunk size for upload and download * Add `allow_archived` argument in `Task.get_tasks()` * Support querying model metadata in `Model.query_models()` * Add `Dataset.set_metadata()` and `Dataset.get_metadata()` * Add `delete_from_storage` (default `True`) to `Task.delete_artifacts()` **Bug Fixes** * Fix jsonargparse and pytorch lightning integration broken for remote execution ([ClearML GitHub issue #403](https://github.com/allegroai/clearml/issues/403)) * Fix error when using `TaskScheduler` with `limit_execution_time` ([ClearML GitHub issue #648](https://github.com/allegroai/clearml/issues/648)) * Fix dataset not synced if the changes are only modified files ([ClearML GitHub issue #822](https://github.com/allegroai/clearml/issues/822)) * Fix `StorageHelper.delete()` does not respect path substitutions ([ClearML GitHub issue #825](https://github.com/allegroai/clearml/issues/825)) * Fix can't write more than 2 GB to a file * Fix `StorageManager.get_file_size_bytes()` returns `ClientError` instead of `None` for invalid S3 links * Fix Dataset lineage view is broken with multiple dataset dependencies * Fix `tensorflow_macos` support * Fix crash when calling `task.flush(wait_for_uploads=True)` while executing remotely * Fix `None` values get casted to empty strings when connecting a dictionary