Make all `OpenML` classes to inherit `ReprMixin` by JATAYU000 · Pull Request #1567 · openml/openml-python

JATAYU000 · 2025-12-29T05:43:09Z

Metadata

Reference Issue: Fixes: Add __repr__ Methods to OpenMLSplit #1563
New Tests Added: No
Documentation Updated: No
Change Log Entry: Inherit ReprMixin for common __repr__ formatting across all OpenML classes
Dependent on [ENH] Added ReprMixin to share __repr__ formatting #1595

codecov-commenter · 2025-12-31T14:15:03Z

Codecov Report

❌ Patch coverage is 78.26087% with 10 lines in your changes missing coverage. Please review.
✅ Project coverage is 52.98%. Comparing base (d421b9e) to head (c76f884).

Files with missing lines	Patch %	Lines
openml/setups/setup.py	60.00%	4 Missing ⚠️
openml/datasets/data_feature.py	57.14%	3 Missing ⚠️
openml/tasks/split.py	62.50%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1567      +/-   ##
==========================================
+ Coverage   52.04%   52.98%   +0.94%     
==========================================
  Files          36       36              
  Lines        4333     4339       +6     
==========================================
+ Hits         2255     2299      +44     
+ Misses       2078     2040      -38

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

fkiraly

Question: have you considered inheriting from OpenMLBase, or at least move the __repr__ related logic to a common place, instead of writing a new one?

Not saying that this is how it has to be done, I would like to hear your rationale.

JATAYU000 · 2026-01-01T13:55:10Z

Question: have you considered inheriting from OpenMLBase, or at least move the __repr__ related logic to a common place, instead of writing a new one?

OpenMLSplit is independent, so making it inherit OpenMLBase just to reuse two methods did not feel right. I hadn’t considered moving __repr__ to a common place before, now that I think about it creating a small Mixin class would be better? ex ReprMixin which defines __repr__ and _apply_repr_template which expects the class inheriting from it implement _get_repr_body_fields
What is your opinion on that?

fkiraly · 2026-01-02T07:49:37Z

What is your opinion on that?

I think it might be a good idea, simply because of the DRY principle.
Although I am not 100% sure how it works out.

geetu040 · 2026-01-02T10:37:54Z

can we put this in utility functions? I don't think that's going to be clean, then ReprMixin class sounds good.

SimonBlanke · 2026-01-02T10:43:11Z

Question: have you considered inheriting from OpenMLBase, or at least move the __repr__ related logic to a common place, instead of writing a new one?

Not saying that this is how it has to be done, I would like to hear your rationale.

OpenMLSplit should not inherit from OpenMLBase, because it is not a server-entity. Moving the __repr__ logic can make sense. But I would not do it in this PR. I would rather do quick PRs and followup issues/PRs for refactoring.

…as non-strict expected fail. (openml#1587) #### Metadata * Reference Issue: Temporarily fix issue openml#1586 #### Details - Running the pytest locally, I found only one failed test which is: `tests/test_runs/test_run_functions.py::test__run_task_get_arffcontent_2` - However, when trying to go through the failed tests in the recent runed jobs in different recent PRs, I found many other failed tests, I picked some of them and tried to make a kind of analysis, and here are my findings: ##### Primary Failure Patterns 1. OpenML Test Server Issues (Most Common) The majority of failures are caused by: - `OpenMLServerError: Unexpected server error when calling https://test.openml.org/... with Status code: 500` - Database connection errors: `Database connection error. Usually due to high server load. Please wait N seconds and try again.` - Timeout errors: `TIMEOUT: Failed to fetch uploaded dataset` 2. Cache/Filesystem Issues - `ValueError: Cannot remove faulty tasks cache directory ... Please do this manually!` - `FileNotFoundError: No such file or directory` 3. Data Format Issues - `KeyError: ['type'] not found in axis` - `KeyError: ['class'] not found in axis` - `KeyError: ['Class'] not found in axis`

…openml#1556) #### Metadata * Reference Issue: Fixes openml#1542 #### Details Fixed sklearn models detection by safely importing openml-sklearn at `openml/runs/__init__.py`

…#1559) I have Refactored the `OpenMLEvaluation` class from a traditional Python class to use the `@dataclass` decorator to reduce boilerplate code and improve code maintainability. #### Metadata * Reference Issue: openml#1540 * New Tests Added: No * Documentation Updated: No * Change Log Entry: Refactored the `OpenMLEvaluation` class to use the `@dataclass` #### Details Edited the `OpenMLEvaluation` class in `openml\evaluations\evaluation.py` to use `@dataclass` decorator. This significantly reduces the boilerplate code in the following places: - Instance Variable Definitions **Before:** ```python def __init__( self, run_id: int, task_id: int, setup_id: int, flow_id: int, flow_name: str, data_id: int, data_name: str, function: str, upload_time: str, uploader: int, uploader_name: str, value: float | None, values: list[float] | None, array_data: str | None = None, ): self.run_id = run_id self.task_id = task_id self.setup_id = setup_id self.flow_id = flow_id self.flow_name = flow_name self.data_id = data_id self.data_name = data_name self.function = function self.upload_time = upload_time self.uploader = uploader self.uploader_name = uploader_name self.value = value self.values = values self.array_data = array_data ``` **After:** ```python run_id: int task_id: int setup_id: int flow_id: int flow_name: str data_id: int data_name: str function: str upload_time: str uploader: int uploader_name: str value: float | None values: list[float] | None array_data: str | None = None ``` - _to_dict Method Simplification **Before:** ```python def _to_dict(self) -> dict: return { "run_id": self.run_id, "task_id": self.task_id, "setup_id": self.setup_id, "flow_id": self.flow_id, "flow_name": self.flow_name, "data_id": self.data_id, "data_name": self.data_name, "function": self.function, "upload_time": self.upload_time, "uploader": self.uploader, "uploader_name": self.uploader_name, "value": self.value, "values": self.values, "array_data": self.array_data, } ``` **After:** ```python def _to_dict(self) -> dict: return asdict(self) ``` All tests are passing with accordnce to the changes: ```bash PS C:\Users\ASUS\Documents\work\opensource\openml-python> pytest tests/test_evaluations/ ======================================= test session starts ======================================= platform win32 -- Python 3.14.0, pytest-9.0.2, pluggy-1.6.0 rootdir: C:\Users\ASUS\Documents\work\opensource\openml-python configfile: pyproject.toml plugins: anyio-4.12.0, flaky-3.8.1, asyncio-1.3.0, cov-7.0.0, mock-3.15.1, rerunfailures-16.1, timeout-2.4.0, xdist-3.8.0, requests-mock-1.12.1 asyncio: mode=Mode.STRICT, debug=False, asyncio_default_fixture_loop_scope=None, asyncio_default_test_loop_scope=function collected 13 items tests\test_evaluations\test_evaluation_functions.py ............ [ 92%] tests\test_evaluations\test_evaluations_example.py . [100%] ================================= 13 passed in 274.80s (0:04:34) ================================== ```

…enml#1566) #### Metadata * Reference Issue: Fixes openml#1531 * New Tests Added: No * Documentation Updated: Yes * Change Log Entry: Update supported Python version range to 3.10–3.14 and extend CI testing to Python 3.14 #### Details This pull request updates the officially supported Python version range for openml-python from 3.8–3.13 to 3.10–3.14, in line with currently supported Python releases. The following changes were made: Updated pyproject.toml to reflect the new supported Python range (3.10–3.14). Extended GitHub Actions CI workflows (test.yml, dist.yaml, docs.yaml) to include Python 3.14. Updated documentation (README.md) wherever Python version support is mentioned. No new functionality or tests were introduced; this is a maintenance update to keep Python version support and CI configuration up to date. This change ensures that users and contributors can use and test openml-python on the latest supported Python versions.

Fixes openml#1598 This PR adds the `@pytest.mark.uses_test_server()` marker to tests that depend on the OpenML test server. Changes * added `uses_test_server` on the relevant test sets. * replaced all the `server` markers with `uses_test_server` marker * removed all the `@pytest.mark.xfail(reason="failures_issue_1544", strict=False)` where the failure was due to race-conditions or server connectivity

fkiraly

Not ready to merge, requires conflicts to be resolved

for more information, see https://pre-commit.ci

JATAYU000 · 2026-01-22T09:55:49Z

I have resolved the conflicts, but there seems to be unrelated test failures
For reviewers note, ruff hook in pre-commit raised error for OpenMLSplit and OpenMLDataFeature for not implementing __hash__ whilst having __eq__ since its __hash__ method automatically set to None, So have added it please review it as well.

geetu040 · 2026-02-06T07:36:23Z

openml/datasets/data_feature.py

    def __eq__(self, other: Any) -> bool:
        return isinstance(other, OpenMLDataFeature) and self.__dict__ == other.__dict__

+    def __hash__(self) -> int:
+        return hash(
+            (
+                self.index,
+                self.name,
+                self.data_type,
+                tuple(self.nominal_values) if self.nominal_values is not None else None,
+                self.number_missing_values,
+                tuple(self.ontologies) if self.ontologies is not None else None,
+            )
+        )
+


I am not so sure about the custom implementation of __hash__, I know it's a requirement from pre-commit but we need to make sure we don't just write a bad implementation to satisfy the pre-commit checks

I think if it can be set to None, and that shuts the pre-commit and is right choice in code and no sdk code currently depends on hashing then do it like that:

If we want to implement __hash__, given the implementation of __eq__, doesn't it make more sense to create hash by creating a tuple of tuples by looping over all (key, value) pairs of self.__dict__

pairs of self.dict

self.__dict__ would return unhashable items which would raise errors, Thats Why I picked immutable/hashable fields

I think if it can be set to None, and that shuts the pre-commit and is right choice in code and no sdk code currently depends on hashing then do it like that:I think if it can be set to None, and that shuts the pre-commit and is right choice in code and no sdk code currently depends on hashing then do it like that:

I have set it None and it does shut the pre-commit failure

@fkiraly please have a look at this thread.

Is it fine to have __hash__ = None for a class?

JATAYU000 and others added 2 commits December 29, 2025 11:07

Added __repr__

e89ed3f

Merge branch 'main' into repr_openmlsplit

e7006d8

fkiraly requested changes Jan 1, 2026

View reviewed changes

JATAYU000 mentioned this pull request Jan 2, 2026

[ENH] Add ReprMixin to share __repr__ formatting across OpenML classes #1591

Open

JATAYU000 changed the title ~~Added __repr__ method from OpenMLSplit~~ Make all OpenML classes to inherit ReprMixin Jan 7, 2026

EmanAbdelhaleem and others added 8 commits January 8, 2026 10:51

[BUG] Fix Sklearn Models detection by safely importing openml-sklearn (…

4a3aae6

…openml#1556) #### Metadata * Reference Issue: Fixes openml#1542 #### Details Fixed sklearn models detection by safely importing openml-sklearn at `openml/runs/__init__.py`

Added RepeMixin in utils

5dfa47d

Updated OpenML classes which require repr

d755d4c

Merge branch 'main' into repr_openmlsplit

8450715

geetu040 assigned JATAYU000 Jan 19, 2026

SimonBlanke approved these changes Jan 20, 2026

View reviewed changes

fkiraly requested changes Jan 22, 2026

View reviewed changes

JATAYU000 and others added 3 commits January 22, 2026 14:48

Merge branch 'main' into repr_openmlsplit

47bfec3

[pre-commit.ci] auto fixes from pre-commit.com hooks

d7c6406

for more information, see https://pre-commit.ci

Class requires __hash__

b295d38

Merge branch 'main' into repr_openmlsplit

ff648f8

geetu040 suggested changes Feb 6, 2026

View reviewed changes

set __hash__ None

c76f884

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Make all `OpenML` classes to inherit `ReprMixin`#1567

Make all `OpenML` classes to inherit `ReprMixin`#1567
JATAYU000 wants to merge 15 commits intoopenml:mainfrom
JATAYU000:repr_openmlsplit

JATAYU000 commented Dec 29, 2025 •

edited

Loading

Uh oh!

codecov-commenter commented Dec 31, 2025 •

edited

Loading

Uh oh!

fkiraly left a comment

Uh oh!

JATAYU000 commented Jan 1, 2026

Uh oh!

fkiraly commented Jan 2, 2026

Uh oh!

geetu040 commented Jan 2, 2026

Uh oh!

SimonBlanke commented Jan 2, 2026

Uh oh!

fkiraly left a comment

Uh oh!

JATAYU000 commented Jan 22, 2026

Uh oh!

geetu040 Feb 6, 2026

Uh oh!

JATAYU000 Feb 6, 2026 •

edited

Loading

Uh oh!

geetu040 Feb 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

Uh oh!

Conversation

JATAYU000 commented Dec 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Metadata

Uh oh!

codecov-commenter commented Dec 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

fkiraly left a comment

Choose a reason for hiding this comment

Uh oh!

JATAYU000 commented Jan 1, 2026

Uh oh!

fkiraly commented Jan 2, 2026

Uh oh!

geetu040 commented Jan 2, 2026

Uh oh!

SimonBlanke commented Jan 2, 2026

Uh oh!

fkiraly left a comment

Choose a reason for hiding this comment

Uh oh!

JATAYU000 commented Jan 22, 2026

Uh oh!

geetu040 Feb 6, 2026

Choose a reason for hiding this comment

Uh oh!

JATAYU000 Feb 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

geetu040 Feb 6, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

JATAYU000 commented Dec 29, 2025 •

edited

Loading

codecov-commenter commented Dec 31, 2025 •

edited

Loading

JATAYU000 Feb 6, 2026 •

edited

Loading