chore(skore): Switch from mypy to ty

[Sharkyii]

Since ty surfaced a large number of pre-existing diagnostics, I added scoped ignore rules in each project’s pyproject.toml to establish a stable baseline:

• skore/: 12 rules ignored
• skore-hub-project/: 5 rules ignored
• skore-local-project/: 5 rules ignored

Fixes: #2376, replaces #2455.

[Sharkyii]

@thomass-dev When you have some time, I would appreciate another review. I’ve updated the PR to address the earlier concerns and verified the CI behavior.

[thomasloux]

@Sharkyii you probably tagged the wrong person :)

[github-actions]

Documentation preview @ 9b733ce

[github-actions]

Coverage Report for skore-hub-project/

File	Stmts	Miss	Cover	Missing
skore-hub-project/src/skore_hub_project
__init__.py	12	1	91%	34
exception.py	2	0	100%
json.py	10	1	90%	16
protocol.py	47	0	100%
skore-hub-project/src/skore_hub_project/artifact
__init__.py	0	0	100%
artifact.py	25	0	100%
serializer.py	28	0	100%
upload.py	36	4	88%	175, 177–178, 180
skore-hub-project/src/skore_hub_project/artifact/media
__init__.py	5	0	100%
data.py	20	0	100%
inspection.py	46	1	97%	33
media.py	10	0	100%
model.py	10	0	100%
performance.py	45	0	100%
skore-hub-project/src/skore_hub_project/artifact/pickle
__init__.py	2	0	100%
pickle.py	24	0	100%
skore-hub-project/src/skore_hub_project/authentication
__init__.py	0	0	100%
apikey.py	7	0	100%
login.py	24	3	87%	32–33, 42
token.py	80	0	100%
uri.py	5	0	100%
skore-hub-project/src/skore_hub_project/client
__init__.py	0	0	100%
client.py	63	0	100%
skore-hub-project/src/skore_hub_project/metric
__init__.py	10	0	100%
accuracy.py	35	0	100%
brier_score.py	35	0	100%
log_loss.py	35	0	100%
metric.py	55	4	92%	38, 77–78, 84
precision.py	53	0	100%
r2.py	35	0	100%
recall.py	55	0	100%
rmse.py	35	0	100%
roc_auc.py	35	0	100%
timing.py	76	4	94%	45–46, 104–105
skore-hub-project/src/skore_hub_project/project
__init__.py	0	0	100%
project.py	144	6	95%	86, 111, 126, 314, 394, 424
skore-hub-project/src/skore_hub_project/report
__init__.py	3	0	100%
cross_validation_report.py	73	0	100%
estimator_report.py	10	0	100%
report.py	59	0	100%
TOTAL	1249	24	98%

Tests	Skipped	Failures	Errors	Time
249	0 💤	0 ❌	0 🔥	2m 1s ⏱️

[github-actions]

Coverage Report for skore-local-project/

File	Stmts	Miss	Cover	Missing
skore-local-project/src/skore_local_project
__init__.py	5	0	100%
metadata.py	94	5	94%	28, 36, 49, 165–166
project.py	102	1	99%	259
storage.py	40	2	95%	45, 187
TOTAL	241	8	96%

Tests	Skipped	Failures	Errors	Time
23	0 💤	0 ❌	0 🔥	7.869s ⏱️

[github-actions]

Coverage Report for skore/

File	Stmts	Miss	Cover	Missing
skore/src/skore
__init__.py	28	0	100%
_config.py	44	1	97%	57
exceptions.py	4	4	0%	4, 15, 19, 23
skore/src/skore/_project
__init__.py	0	0	100%
_summary.py	81	1	98%	123
_widget.py	195	0	100%
login.py	10	0	100%
plugin.py	12	0	100%
project.py	54	2	96%	148, 157
types.py	3	0	100%
skore/src/skore/_sklearn
__init__.py	6	0	100%
_base.py	70	8	88%	268–275
feature_names.py	27	0	100%
find_ml_task.py	61	0	100%
types.py	28	1	96%	30
skore/src/skore/_sklearn/_comparison
__init__.py	7	0	100%
inspection_accessor.py	40	1	97%	195
metrics_accessor.py	155	2	98%	179, 1124
report.py	111	3	97%	514, 517, 523
utils.py	58	0	100%
skore/src/skore/_sklearn/_cross_validation
__init__.py	9	0	100%
data_accessor.py	31	1	96%	48
inspection_accessor.py	19	1	94%	121
metrics_accessor.py	158	4	97%	1046, 1107, 1140–1141
report.py	109	4	96%	403, 445, 448, 454
skore/src/skore/_sklearn/_estimator
__init__.py	9	0	100%
data_accessor.py	45	1	97%	167
inspection_accessor.py	58	1	98%	321
metrics_accessor.py	352	5	98%	427, 431, 446, 476, 1647
report.py	166	3	98%	273, 477, 480
skore/src/skore/_sklearn/_plot
__init__.py	3	0	100%
base.py	59	2	96%	60–61
utils.py	121	2	98%	273–274
skore/src/skore/_sklearn/_plot/data
__init__.py	2	0	100%
table_report.py	178	1	99%	661
skore/src/skore/_sklearn/_plot/inspection
__init__.py	0	0	100%
coefficients.py	204	0	100%
impurity_decrease.py	84	2	97%	365, 409
permutation_importance.py	149	0	100%
utils.py	9	0	100%
skore/src/skore/_sklearn/_plot/metrics
__init__.py	6	0	100%
confusion_matrix.py	163	0	100%
metrics_summary_display.py	8	0	100%
precision_recall_curve.py	108	0	100%
prediction_error.py	167	0	100%
roc_curve.py	112	0	100%
skore/src/skore/_sklearn/train_test_split
__init__.py	0	0	100%
train_test_split.py	59	0	100%
skore/src/skore/_sklearn/train_test_split/warning
__init__.py	8	0	100%
high_class_imbalance_too_few_examples_warning.py	21	1	95%	86
high_class_imbalance_warning.py	22	0	100%
random_state_unset_warning.py	10	0	100%
shuffle_true_warning.py	9	0	100%
stratify_is_set_warning.py	10	0	100%
time_based_column_warning.py	21	0	100%
train_test_split_warning.py	3	0	100%
skore/src/skore/_utils
__init__.py	6	2	66%	8, 13
_accessor.py	118	4	96%	38, 214, 268, 288
_cache.py	37	0	100%
_cache_key.py	29	3	89%	35, 43, 52
_dataframe.py	34	1	97%	43
_environment.py	32	2	93%	41, 44
_fixes.py	8	0	100%
_index.py	5	0	100%
_jupyter.py	8	2	75%	13–14
_logger.py	22	4	81%	15–17, 19
_measure_time.py	10	0	100%
_parallel.py	17	0	100%
_patch.py	22	12	45%	31, 36–40, 43–44, 47–48, 59, 61
_progress_bar.py	41	4	90%	55–56, 66–67
_show_versions.py	38	0	100%
_testing.py	110	11	90%	21, 30, 158, 167, 178–183, 185
skore/src/skore/_utils/repr
__init__.py	2	0	100%
base.py	54	0	100%
data.py	164	0	100%
html_repr.py	38	0	100%
rich_repr.py	81	0	100%
TOTAL	4362	96	97%

Tests	Skipped	Failures	Errors	Time
1765	5 💤	0 ❌	0 🔥	7m 27s ⏱️

[Sharkyii]

Why ty flags errors that mypy didn’t

Import & attribute resolution

Errors: unresolved-import, unresolved-attribute, possibly-missing-attribute
mypy often supresses missing or dynamic attributes that are used in the repo as Any.
Docs:
mypy: here

Function call & signature checking

Errors: unknown-argument, invalid-argument-type, missing-argument, no-matching-overload, call-non-callable
mypy skips checking many untyped function bodies and treats unknown types as Any, so invalid calls often pass silently. ty performs stricter call-site checking even when types are incomplete.

Docs:
mypy untyped defs not checked → here
mypy Any behavior → here
ty diagnostics philosophy → here

Assignment & return typing

Errors: invalid-assignment, invalid-return-type, invalid-type-form
mypy widens many inferred types to avoid false positives and accepts more loose assignments. ty keeps stricter inference and flags mismatches instead of widening automatically.

Docs:
mypy assignment rules → here

Hygiene / redundancy checks

Errors: redundant-cast, unused-type-ignore-comment
mypy only reports these if explicitly enabled. ty reports them by default to keep annotations clean.

Docs:
mypy unused ignore behavior → here
ty ignore handling → here

[thomass-dev]

Just to ensure i can review.

[auguste-probabl]

Suggested change

	warning = cast(Any, warning_class).check(**kwargs)
	warning = cast(TrainTestSplitWarning, warning_class).check(**kwargs)

[auguste-probabl]

Working on it, hope that's okay ^^'

[Sharkyii]

Yupp, works for me.

[thomass-dev]

This is redundant with the import in TYPE_CHECKING, isn't?

[auguste-probabl]

No, I got test failures when this wasn't imported out of TYPE_CHECKING. So it looks like cast statements are not simply removed at runtime...

[auguste-probabl]

I'm having a lot of trouble with this because I never get the same check results between CI and my machine. I'm going to let this sit for a while and hope that ty matures more and provides a more robust solution.

[thomass-dev]

This change has to be discussed, as ty doesn't have a strict mode for now and we would to be strict as much as possible in skore-hub-project and skore-local-project.

https://docs.astral.sh/ty/reference/typing-faq/#does-ty-have-a-strict-mode

Could you please list me the rules that will not be applied with this change?

[thomass-dev]

I know it violates the Liskov substitution principle, but your change is not true. Please revert.

[thomass-dev]

We can fix the violation in a dedicated issue as we have done in https://github.com/probabl-ai/skore/blob/main/skore-hub-project/src/skore_hub_project/report/estimator_report.py, but it is not important here.

[thomass-dev]

Please revert.