fix(deps): update all non-major dependencies #51
No reviewers
Labels
No labels
ansible
bug
ci
dependencies
digest
docker
documentation
duplicate
enhancement
fix
go
good first issue
help wanted
invalid
javascript
major
minor
patch
php
python
question
released
security
wontfix
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
dawid/langchain-memvid!51
Loading…
Reference in a new issue
No description provided.
Delete branch "renovate/all-minor-patch"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
This PR contains the following updates:
==1.15.1→==1.16.0==9.1.1→==9.2.0==1.4.1.post1→==1.5.0<3.13.11→<3.17.6Release Notes
facebookresearch/faiss (faiss-cpu)
v1.16.0Compare Source
Added
fc1095ae-means: add benchmarks reproducing the paper (#5845)2079677e-means: add max_points_per_centroid (#5855)8e5843cAdd regression test for flat search in a nested OpenMP region (#5754)b80093cAdd DINO resident and SSD streaming e-means demos (#5815)97fdfb0Add optional IVF assignment for centroid-sharded e-means (#5817)f6a82dcAdd centroid-sharded e-means training (#5823)0136502Add native INT8 CuTe assignment to e-means (#5816)15b14b2Add FP8 CuTe assignment to e-means (#5820)13a33bfAdd CuTe assignment backend to e-means (#5818)7a038e7Add INT8 Triton assignment to e-means (#5821)87b5339Add FP8 Triton assignment to e-means (#5819)c8ab4f5Add Triton assignment backend to e-means (#5822)f61f80aAdd reduced-precision assignment to e-means (#5814)8ac2fdcAdd distributed E-means training (#5813)3795ee6Add reference e-means implementations (#5812)951067dPrint native crash diagnostics in Linux CMake CI (#5828)7648f5fInvertedListScanner::distances_to_codes, with a block kernel for 8-bit PQ (#5807)a6505c5Expose caller-owned Dataset APIs for CAGRA (#5739)4caf504Add MetalIndexBinaryFlat with GPU Hamming search and exact top-k selection (#5759)71c7f48Add remove_ids to IndexShards and IndexReplicas, and add_with_ids to IndexReplicas (#5773)6964ecdMake the multi-GPU Python helpers work on the Metal backend (#5747)b156ccefeat(rvv): add bit-exact RVV decode_vector for raw-codec quantizers (#5689)8ff93a8Serialization, clone and index_factory support for IndexPiPNN (#5758)840aa1fAdd HNSW level-0 links from a sample to remove search sinks (#5705)89e5eb5Add IndexPiPNN, a PiPNN graph index with beam search (#5734)3effdb2PiPNN build assembly (#5729)07979b1Randomized Ball Carving partition (#5714)7069478Numeric kernels7a05795HashPrune reservoir and sketch kernel (#5708)b0074a3Add fp16 support toSuperKMeans(#5692)e7c44ebAdd AVX-512 fp16 conversion kernels (#5694)ea29bacRaBitQ: add batch-four estimates and 9-bit SIMD scoring (#5647)76c0e01Add GPU-accelerated IVFPQ search to the Metal backend (#5449)0db2f30rabitq neon kernels (#5532)430a41eNative fp16 k-means training for Faiss IVF (#5671)6c8378eAdd RVV implementation for decode_bf16_simd (#5641)e24a371feat(rvv): add vector arithmetic kernels (#5655)94a4331Add reranked types to IndexSQFastScan and fix flat RaBitQ filtering (#5661)dcf8ebbSupport IDSelector in flat fast-scan search (#5660)1768a5aAdd RVV SIMD specialization for product quantizer distance-table computation (#5543)f323c18Add RVV encoding for QT_8bit_direct (#5640)5e74b1bfeat: implement RVV optimized scalar quantizer codecs and distance computation (#5535)b0d9e80Add fast scan variants of SQ indexes - IndexSQFastScan and IndexIVFSQFastScan (#5341)4864f4dSVS: Add ParameterSpace.set_index_parameter support for Vamana (#5644)0f23e6aAdd RVV SIMD specialization for scalar quantizer distance compute (#5539)Changed
f34c0b8Move the macOS arm64 CI jobs from macos-14 to macos-15 (#5875)741138aE-means: quality batch at d <= 16 and no revival at d <= 2 by default (#5854)731286fe-means: faiss-style verbose output that reports epochs and asks for a citation (#5841)277b3c7e-means: rename niter to nepochs (#5848)a15797aCompute the coarse distances in search_preassigned when they are not given (#5748)b955264Replace the e-means README with a user guide (#5842)e4500c2Use BF16 defaults and share quantized rows across E-means sharding (#5844)0a6f936Emeans ssd speedups (#5839)9246aa7Rename e-means operating_point to batch_law, allow batch_size without alpha0, and default niter to 5 (#5833)22a0cb7Separate Emeans implementations (#5840)1d40eebSpeed up the x86 8-bit scalar quantizer inner-product distance (#5829)efd79c9Upgrade CPU CMake CI to PyTorch 2.15 (#5827)0b0250dSQ fp16 and bf16: four accumulators per code at d >= 384, any d at AVX-512 (#5806)a3217cbGive each SIMD level its own copy of the 65 shared functions (#5810)2a6e785Define the AVX-512 SQ specializations once for AVX512 and AVX512_SPR (#5808)c2d27c6Search only the sample vectors at sinks after each round of HNSW level-0 links (#5804)9d5b442Codify the Faiss SIMD guard rules and check them with a test (#5793)0d3c380Move SIMD code to dedicated translation units and align guards (#5792)03aa1bdGive fvec_L2sqr_ny_y_transposed_D internal linkage in the AVX2 and AVX512 files (#5788)167f443Prefetch graph rows and vectors in PiPNN search (#5805)ea99cbcSVS 0.5.0 (#5785)0c2cad8Run the Windows conda CI steps in bash (#5795)35a4ec2Use 3 sample vectors per node by default for HNSW level-0 links (#5789)1ec0011Hamming: vector popcount for the default computer at AVX512 and AVX512_VPOPCNT (#5787)fc65960Use a SIMD masked sum for the fp32 RaBitQ 1-bit estimate (#5782)d37bedeTrain ProductQuantizer subquantizers in parallel (#5783)386c96bDefault alpha to 1.2 for all metrics (#5786)0a32a42Scale the HNSW and PiPNN search check period by the thread count (#5784)c919a60Dispatch RaBitQ scans once per inverted list (#5597)6907ed1Prune inner-product graphs on cosines when norms vary (#5774)e353ed78-bit PQ code distance: 4-sum scalar kernel instead of gathers on AMD Zen 3/4/5 (#5751)9e16309Build PiPNN graphs over distinct codes (#5768)7834abaSize IVF fast-scan slice LUTs on the calling thread (#5721)2732125Dispatch more dimensions to the SIMD fvec_L2sqr_ny_transposed kernels (#5757)ee6927divflib::search_with_distance_to_code: search through InvertedListScanner::distance_to_code (#5755)56c96d5Move Faiss C++ tests to Python (#5743)a55e0b2Upgrade cuVS to 26.10 (#5633)494330fSpeed up the NEON 8-bit scalar quantizer L2 distance (#5733)1a9ddc6Use monotonic dynamic scheduling so libomp 12 does not take its static-steal path (#5723)3fc6bcdOn ROCm, stage pageable host copies in pinned memory (#5712)cec5702DD: give fvec_L2sqr_ny_nearest_y_transposed_D internal linkage (#5720)db6aa06Hoist per-list allocations out of the IVF fast-scan search loops (#5717)3907966Select OpenBLAS OMP build in conda for AArch64 + Linux (#5669)44afa77Stop allocating per (query, probe) on OpenMP workers in IVF RaBitQ fast-scan (#5716)6692d7aStop copying QueryFactorsData per block in RaBitQHeapHandler3a98062faiss/conda: extend LSQ NEON-flaky test skip to all arm64 (#5713)2839c39Make the GPU IVF search return the same neighbours on every run (#5682)a2f7943Optimize 4-bit PQ fastscan accumulation on Arm CPUs (#5709)83ae8b0Move the RVV distance kernels into free-standing headers (#5697)1dab0c1Quarantine unstable cuVS Python GPU tests (#5699)ae4466dAvoid widening fp16 data for k-means++ initialization (#5693)ea9151dRaBitQ: broadcast one sign word per 32 dims in the AVX2 RBQ9 kernel (#5696)4102312Defer IVF list scan setup until needed (#5595)fdb9535Make search stats collection configurable (#5670)7f6bf10Halve IndexSQFastScan rerank memory with a split-nibble layout (#5662)b1c1496migrate from bit-manipulation builtins in fbcode/faiss (#5667)2496f63Avoid search stats contention during concurrent search (#5663)3f4838eBatch the IVF ScalarQuantizer scan four codes at a time (#5456)1dbb3d2Update backward compatibility test to v1.15.1 (#5646)Fixed
29782a0Fix verbose e-means tests after thenepochsrename (#5869)8babd49Fix assignment after progressive-dimension Kmeans training (#5861)b7c44bcInitialize SVS Vamana graph parameters in the default constructor. (#5834)9acf8b0Guard fp16 threshold in mock builds [unused-const-variable] (#5835)0c269acValidate codebook size in read_ResidualQuantizer_old and bounds-check IndexBinaryIVF::reconstruct_from_offset (#5764)7ac4a6cFix database-parallel kNN when the parallel region gets fewer threads (#5824)98a23e4Fix OMP exception safety in ProductQuantizer (#5771)d87b58bfbcode/faiss/tests/test_fast_scan_distance_to_code.cpp (#5772)48f2129Drop constants [unused-const-variable] (#5794)2f890e8Fix train_NonUniform skipping training row 0 and racing on a shared buffer (#5791)d4a6548Faiss OSS nightly autofix (#5790)5712d82Fix cuVS IVF-Flat workspace sizing after CPU-to-GPU cloning (#5763)bce5e24Re-enable the cuVS CAGRA Python tests that hung on cuVS 26.06 (#5770)450dd79Fix AArch64's conda multi output recipe (#5727)55afdffcmake: install utils/popcount.h (#5744)d4a4e3eFix build for CPUTYPE=znver4 (#5752)56797f9Match scalar tie-break in AVX2/AVX512 Lloyd-Max encode (#5715)5b4aa46fbcode/faiss/gpu/utils (#5737)a643100Faiss OSS nightly autofix (#5731)4f63b42Keep AVX512-TU copies of the AVX2 QBS loop out of the AVX2 path (#5724)c966dabFaiss OSS nightly autofix (#5728)5d635f3DD: give PQCodeDistanceScalar a SIMD level so the AVX2 path cannot run AVX512 code (#5719)f46050eDeflake test_io TestIVFPQRead.test_reader (#5718)8b884ecFix IndexFastScan::reset() leaving stale ntotal2 (crash on search) (#5614)d380b7dFix IndexFlatCodes::reset on mmap-loaded codes (#5707)6c51305Avoid double-advancing HNSW visited tables (#5596)76978eeFix selector handling for IVF iterator scans (#5592)38460eeFix Faiss fp16 parity tests across architectures (#5695)88a28bcFaiss OSS nightly autofix (#5686)ec94295Bound the fast scan LUT scale by the accumulated sum (#5658)3f08095fix: release SuperKMeans sample buffer after rotation (#5648)02cacc5Fix TSAN races in sidehsnw_statsandrabitq_stats(#5657)bf8be4afix(ci): use GCC 14 riscv64 cross-compiler for RVV tuple/zvfhmin intrinsics (#5653)98af2cfFix osx-arm64 conda build: drop redundant openblas pin, allow single-platform release dispatch (#5652)3778f58Fix IDSelector handling in unbounded HNSW search (#5508)Deprecated
8f176c3Remove the replicas parameter (#5809)de495c9Remove redundant contrib, torch, Manifold and telemetry tests (#5801)ce25a10Remove redundant Panorama, general index, factory and SWIG tests (#5802)ca60fb5Remove redundant clustering, binary index and SIMD distance tests (#5800)4915580Remove redundant serialization, mmap and deserialization tests (#5798)2f8c1d2Remove redundant SVS, HNSW, NSG and PiPNN tests (#5799)f2130c2Remove redundant IVF and search parameter tests (#5803)1c84f08Remove redundant PQ, AQ and SQ fast-scan tests (#5797)2010413Remove redundant PQ, AQ, SQ and RaBitQ tests (#5796)42aac2fRemove the cuVS 26.02 compatibility paths from the cuVS GPU code (#5750)pytest-dev/pytest (pytest)
v9.2.0Compare Source
pytest 9.2.0 (2026-10-27)
Removals and backward incompatible breaking changes
#14716:
-cnow raises a usage error for a path that does not exist, is a directory, or is in a format pytest has no loader for. Such paths were previously either ignored or reported as an unhandledFileNotFoundError.#8593: The
indirect,idsandscopearguments ofMetafunc.parametrize() <pytest.Metafunc.parametrize>-- and therefore of@pytest.mark.parametrize <pytest.mark.parametrize ref>-- are now keyword-only.Passing them positionally was a common source of confusing errors, e.g.
parametrize("arg1", "arg2", [(1, 2)])made the argvalues land inindirectand failed withindirect fixture '(1, 2)' doesn't exist; such mistakes now fail with an understandableTypeError. The type annotations have always declared these arguments as keyword-only.#993: Plugin import failures at startup are now reported and classified instead of escaping as a raw traceback with the accidental exit code
1:-p,pytest_pluginsorPYTEST_PLUGINS-- exits withpytest.ExitCodeUSAGE_ERROR(4), like aconftest.pythat fails to import;pytest11entry point or a missing plugin dependency -- exits withpytest.ExitCodeINTERNAL_ERROR(3), with the traceback preserved.pytest.mainnow returns these codes instead of raisingImportError. Also fixed anIndexErrorfrom pytest's own internals when the plugin raised an exception with no arguments.Deprecations (removal in next major release)
#14742: The private
_pytest.python.CallSpec2name has been renamed toCallSpec.The old
CallSpec2alias remains available for now and will be removed in pytest 10 or 11.#14755: Schedule yield_fixture decorator for removal in pytest 10 or 11.
#14774: Similarly to the deprecation of class-scoped fixtures defined as instance methods in pytest 9.1,
all higher-scoped fixtures (
module,package, andsession) definedas instance methods are now deprecated.
They suffer from the same issue the original deprecation only fixed for
class-scoped fixtures, see
10819and14011.#14808: Passing a non-string value to a
string-typed ini option now emits apytest.PytestRemovedIn10Warning. This affects[tool.pytest.ini_options]inpyproject.toml. In pytest 10 this will raise aTypeError, matching the behavior of the corresponding TOML config path.#7323: Passing
bytestoTerminalReporter.write_line()is now deprecated, pass astrinstead.New features
strict_markersis enabled, marker names used in-mexpressions are now validated against registered markers.parametrize_long_str_id_strategyconfiguration option to control how longstr/bytesparameter values are handled in auto-generated test IDs. Explicit IDs set viaids=[...]orpytest.param(..., id=...)are not affected. Available strategies:short(default, values over 100 chars fall back toargname+index),sha256(SHA-256 hex digest),legacy(keep full value), anddisallow(error requesting explicit IDs).Improvements in existing functionality
#10745: A
RecursionErrorwhose recursion origin cannot be located now shows only thefirst and last 10 stack frames, matching what pytest already did when locating
the origin failed outright. Previously the full traceback -- often around a
thousand near-identical frames -- was rendered, which was both slow and
unreadable. Pass
--full-traceto see every frame.#10745: The line numbers of a module's statements are now computed once per traceback
instead of once per traceback entry, which noticeably speeds up rendering
failures raised from large test modules.
#11418: The cached rewritten modules in
__pycache__are now invalidated by a hash ofthe source instead of its modification time, using the checked-hash pyc format
from
552.Timestamps make the cache useless wherever source files get fresh modification
times, most notably in CI: a checkout gives every file a new mtime, so a
restored cache is discarded in full and every test module is rewritten again.
Hashing the source costs roughly 15 microseconds per file and makes the cache
survive that.
Existing timestamp based caches are ignored and rewritten once.
#12624: The
pytest_pluginsvariable and thePYTEST_PLUGINSenvironment variable now also accept entry point names of installed plugins (e.g.xdist), in addition to importable module names, matching the behavior of the-pcommand-line option.As part of this change, plugins specified by entry point name through these mechanisms are now also listed in the
plugins:line of the terminal header (12615).Based on the original implementation by Marcel Telka in
12616.#12697: Warnings emitted while importing plugins through
-p,PYTEST_PLUGINS, or entry point autoloading are now captured, filtered, and reported like other warnings.#12860: Assertion diffs for
dict.items()anddict.keys()comparisons (>=,<=,>,<,==) now show which items are missing, the same way set comparisons do.#13354:
Node.get_closest_marker <_pytest.nodes.Node.get_closest_marker>now also accepts a~pytest.MarkDecoratoras itsdefault, for exampleitem.get_closest_marker("foo", pytest.mark.foo(1)), and returns the wrapped~pytest.Mark.Previously the only way to build a
defaultwas to instantiate~pytest.Markdirectly, which is private and warns.#13369: Improved the error from
pytest.RunResult.assert_outcomeswhen a plugin changes or removes pytest's standard terminal summary, and documented how to isolate output-changing plugins.#13431: Exposed
pytest.ParameterSetas the type of the return value ofpytest.param().This can be used in type annotations and
isinstancechecks.#13564: Issue a warning when fixtures are wrapped with a decorator that prevents
correct discovery, including
@classmethodabove@pytest.fixtureand@staticmethodabove@pytest.fixture. Put@pytest.fixtureabovethose decorators on test classes.
#14164: Exposed
pytest.Approxas the type of the return value ofpytest.approx().This can be used in type annotations and
isinstancechecks.#14463: Assertion diffs for mappings now match up keys the way the compared mappings match them themselves.
Key comparison goes through the mappings' key views instead of plain sets of their keys, so a mapping
with its own notion of key equivalence, such as case-insensitive header mappings, no longer
reports its common keys as missing on either side.
#14523: Large assertion comparison diffs are now built lazily and capped to the
truncation budget, so a huge diff is no longer formatted in full just to
be truncated. As a consequence, the truncation footer no longer
reports the exact number of hidden lines.
#14687: The default fd-level output capturing (
--capture=fd) now skips reading the capture buffer for tests which produce no output, reducing per-test overhead in large test suites.#14724:
--no-summarynow only applies directly to pytest's own summary.The flag no longer skips the
pytest_terminal_summaryhook, so third-partyplugins (for example coverage) can still write their terminal summaries.
#14751:
parser.addini <pytest.Parser.addini>now also accepts type expressions for itstypeargument, in addition to the existing string tags: the plain Python typesstr,bool,int,float, aLiteralof strings, and unions of these (for exampleint | strorint | Literal["auto"]).#14754:
pytest.approx()now rejects infinite relative tolerances fordatetime.timedeltavalues with a clearValueErrorinstead of raisingOverflowError.#14791: The
junit_family,junit_logging,junit_duration_report,console_output_style,empty_parameter_set_mark,assertion_text_diff_styleandtmp_path_retention_policyconfiguration options are now registered with the set of values they accept: an invalid value now fails with a clean usage error instead of being silently ignored or crashing later, andpytest --helplists the accepted values.config.getini <pytest.Config.getini>now raisespytest.UsageErrorinstead ofTypeErrororValueErrorwhen the value read from the configuration file does not match the registered type, so a bad value is always reported as a short usage error message instead of an internal error traceback. Asking for an option name that was never registered still raisesValueError.#15006:
pytest.approxnow emitspytest.PytestApproxDecimalToleranceWarningwhen arelorabstolerance is given as afloatthat cannot be represented exactly for a~decimal.Decimalcomparison, since such a tolerance is widened to the float's exact binary value.#14675, #14953, #15153, #15132: The
truncation_limit_lines,truncation_limit_chars,max_warnings,verbosity_assertions,verbosity_test_cases,verbosity_subtests,tmp_path_retention_count,log_level,log_cli_level,log_file_level,log_auto_indentandfaulthandler_timeoutconfiguration options now accept native integer (and, forlog_auto_indent, boolean) values in TOML configuration files, while still accepting strings for backward compatibility.log_file_modenow reports invalid values as a usage error.#15254:
tmp_pathandTempPathFactory.mktemp <pytest.TempPathFactory.mktemp>no longer list the whole base temporary directory on every call, which made sessions with many temporary directories quadratically slow. Numbers are now handed out from a per-prefix counter, skipping names that already exist.#4569:
Node.add_marker <_pytest.nodes.Node.add_marker>now stores a~pytest.Markinnode.keywords, matching what marks applied during collection store. Previously it stored a~pytest.MarkDecorator.#9436: Avoided formatting tracebacks for tests marked with
xfail(run=False).Bug fixes
#10128: Configured warning filters now apply to warnings emitted by
pytest_configurehooks.#10210:
pytest.approxnow raises aTypeErrorfor any nested container, notonly for one of the same type as its parent. Previously
approx([{"k": 1}])and
approx({"k": [1]})were accepted and then compared unequal to everyvalue, silently ignoring the tolerance, while
approx([[1]])correctlyreported that nesting is unsupported.
#10644:
monkeypatch.setattr()no longer leaves a new entry in the instance__dict__when it patches an attribute that the instance inherits from its class.Previously
undo()assigned the inherited value onto the instance, which shadowed the class attribute -- permanently freezing the result for descriptors that resolve dynamically.#1139: Fixed tracebacks showing
???instead of source lines for code compiled with a filename relative to a directory other than the current working directory -- e.g. compiled Cython modules, when running pytest from a subdirectory of the project root.The source lookup now falls back to searching
sys.pathfor such filenames, as the standardtracebackmodule does.#11502: The
rootdiris no longer derived from a configuration file that is not a regular file, such as--config-file=/dev/null.#11783: Plugins installed in editable mode (
660) are marked for assertion rewriting again.Such installs do not record the package's Python files in the distribution metadata, so pytest found nothing to rewrite and assertions inside the plugin lost their introspection. The plugin's entry point is now used to locate the package in that case.
#11933:
pytest.warnsno longer re-emits unmatched warnings withmodule="warnings", which causedmodule-based warning filters to be applied to the wrong warnings. Sincewarnings.WarningMessagedoes not record the module a warning was emitted with, re-emitted warnings now use the standardwarnings.warn_explicitdefault of deriving the module from the warning's filename.#12175:
ExceptionInfo.exconly(tryshort=True) <pytest.ExceptionInfo.exconly>now strips theAssertionError:prefix of failedassertstatements for~pytest.ExceptionInfoobjects created bypytest.raises, and failures insideunittest.TestCase.subTestreport the same short crash line as regular tests.#12223: The
messageattribute of<failure>and<error>elements in--junitxmlreports is now shortened to whole lines within 500 characters. When something is dropped, it ends with a note pointing to the element text, which still holds the full report. Previously, very verbose assertion output (e.g. with-vv) produced attributes tens of thousands of characters long, which some XML parsers fail on.#12307: Fixed truncation of long collections (lists, dicts, sets, etc.) in the "message" part of failed assertions (as in
assert condition, msg) when using-vv: the message now uses the same unlimited repr as the rest of the assertion output, instead of eliding collection elements with....#12824: Custom command-line options now fail with a clear error when their implicit
destination conflicts with an existing option, avoiding silent overrides of
built-in options such as
-k.#13292: Fixed a stale rewritten module being used when a test file is modified twice
within the same second without changing its size -- the pyc header could only
record whole-second timestamps. The cache is now keyed on a hash of the source.
#14036:
__tracebackhide__is now respected inExceptionGrouptracebacks, both for the group itself and for its sub-exceptions.#14189: Nested usage of
caplog.filtering <pytest.LogCaptureFixture.filtering>no longer removes filters early if they were already present.#14412: Fixed incorrect timing information for subtests and parent tests when
console_output_style = "times".#14445: Assertion rewriting no longer evaluates walrus operator (
:=) expressions more than once, which reported results the test never computed when the expression had side effects.An operand evaluated before a later walrus is now reported with the value it actually saw, so
assert value != identity(value := value.lower())keeps Python's left-to-right order.#14552: Fixed stale
co_filenamevalues after a test module or directory is moved. Assertion-rewrite caches are reused when the source hash still matches, and in-memory code objects are pointed at the current path.#14586: Negative values passed to
log_auto_indentnow correctly fall back to no indentation, matching the documented behavior.#14619: Fixed a crash in
pytest.mark.parametrizewhen a non-sequence value is passed for a parameter set. It now provides a clear error message during collection.#14635: Fixed fixture resolution failures when specifying multiple test paths on the command line that share common parent directories. Parametrized fixtures defined in a
conftest.pycould be silently dropped depending on the order of the specified paths.#14637: Pytest now correctly skips identical trailing characters in assertion text diffs even when the compared strings differ at the beginning.
#14638: Top-level options in
pytest.tomland.pytest.tomlfiles now raise an error instead of being silently ignored when no[pytest]table is present.#14683: Fixed a regression in pytest 9.1 where a conftest.py located in the rootdir was no longer visible to tests and doctests collected from outside the rootdir (for example when passing a parent directory of the rootdir as a collection argument, or setting
--rootdirto a subdirectory).As a result, fixtures defined in such a conftest -- including
doctest_namespaceinjections -- were not available, causing errors such asNameErrorin doctests.#14700: Fixed an internal error on Python 3.12 when a doctest on a fixture-decorated function was skipped.
#14705: TOML configuration files passed via
-cnow also read their configuration from the[pytest]table, the one pytest's own configuration files use. The[tool.pytest]and[tool.pytest.ini_options]tables such files were previously restricted to keep working; using both styles in one file is an error.#14828: Exception groups raised from another exception now print their cause chain only once in the traceback output.
#14864: Fixed collection on Windows collecting the whole suite instead of the given path, on file systems which do not support file IDs (
st_inois0for every file, as seen for example onsshfs-win/WinFsp mounts).#14884:
Cache.get(),Cache.set()andCache.mkdir()now reject keys that resolve outside the cache directory, and--cache-showignores glob matches outside it.#14909: Fixed
MonkeyPatch.delattr <pytest.MonkeyPatch.delattr>,MonkeyPatch.setitem <pytest.MonkeyPatch.setitem>andMonkeyPatch.delitem <pytest.MonkeyPatch.delitem>recording a stale undo entry when the underlying mutation failed, which could cause spurious errors during teardown. The undo entry is now recorded only after the mutation succeeds, matching the behavior ofMonkeyPatch.setattr <pytest.MonkeyPatch.setattr>.#14912:
~pytest.PytestWarninginstances defined in_pytest.deprecatedare no longer reused acrosswarnings.warncalls, so warnings raised as errors (e.g. with-W error) no longer accumulate stale traceback frames from previous raises.#14942: Fixed a regression in 9.1.0 that made collection very slow (quadratic) when many fixtures are defined with the same name.
This can happen particularly when many tests are defined in separate classes which inherit from a base class which defines fixtures (the fixtures are repeated per class).
#15005:
pytest.approxno longer treats finite~decimal.Decimalvalues outside thefloatrange (such asDecimal("1e400")) as infinite, which made in-tolerance comparisons fail.#15006: Restored the tolerance shown by
pytest.approxfor~decimal.Decimalvalues, which since 9.0 displayed???for the default tolerance and reported the rawrelratio instead of the absolute tolerance actually compared against.#15006: The failure message for
pytest.approxcomparisons of~decimal.Decimalcollections no longer reportsMax absolute difference: -inf, or crashes for mappings, when thedecimal.FloatOperationtrap is set.#15006:
pytest.approxno longer raisesAttributeErrorwhen comparingnumpy.ndarrayinstances withobjectdtype, such as arrays of~decimal.Decimal.#15032: Help texts for command-line options and configuration options now keep their indentation
and blank lines. Previously each line was stripped before wrapping, so a wrapped
list item was indistinguishable from a new one and blank lines were dropped; configuration
option help lost its line breaks entirely.
#15043: Cache writes now emit
pytest.PytestCacheWarningwhen writing or closing the cache file fails, consistent with errors opening the file.#15052: Fixed possible off-by-one line numbers in
--fixturesand--fixtures-per-testoutput.#15145: With
console_output_styleset totimes, durations just below a unit boundary are now shown in the next unit (e.g.1.000msinstead of1000.0us).#15172: Assertion diffs now pretty-print the built-in
frozendict(Python 3.15+) across multiple lines, similarly to howdictis displayed.#4569:
-kno longer matches attributes that end up in a test function's__dict__without being meant as keywords, such as pytest's ownpytestmarkattribute and the bookkeeping left behind byfunctools.wrapsandfunctools.lru_cache.#6750: Fix fixture argument discovery for custom collectors that pass bound methods to
pytest.Function.#6997: XFAIL reports from teardown phases now include the phase in the short test
summary.
#7940: Fixed cleanup of temporary directories failing when subdirectories have their execute permission removed.
#8321: Fixed exception chains being duplicated in the traceback output when an exception in the chain (
__cause__/__context__) has no traceback of its own, and that exception is itself linked to further exceptions.#8914: Tests parametrized with higher-than-function scope are now reordered by
parameter value rather than by parameter index, so that items using equal
parameter values are grouped together and share a single fixture setup, even
when they are parametrized by separate
@pytest.mark.parametrize <pytest.mark.parametrize ref>calls. Previously, an expensive higher-scoped fixture could be set up and torn
down repeatedly for the same parameter value.
Unhashable parameter values (e.g. dicts) keep the previous index-based
grouping behavior.
As a consequence, the order in which parametrized tests run may change
(fixture setup/teardown counts can only decrease, not increase).
#9924: Fixed doctest option flags leaking between docstrings after a failure, skip, or expected failure.
Improved documentation
setup.pyusage withpyproject.tomlinwriting-plugins.--confcutdiraffects initialconftest.pydiscovery.pytest-flakefightersplugin to the list of flaky test plugins inplugins.rst.Added flaky test papers published on the Test FLARE project to the list of flaky papers in
plugins.rst.installation instructions <installation>andinstall-package-with-pipnow show how to install pytest into a virtual environment with venv and pip, uv or virtualenv, and explain theexternally-managed-environmenterror.pytest --help, the-kreference entry andselect-testsno longer describe-kas a Python expression matched against test names only;keyword expressionsnow shows marker names being matched as well. The examples that looked up a marker useNode.get_closest_marker <_pytest.nodes.Node.get_closest_marker>rather thanitem.keywords.session.itemsandsession.testscollectedexclude deselected tests in thepytest_collectiondocumentation.Packaging updates and notes for downstreams
#11680: Source archives of arbitrary commits -- the tarballs GitHub serves under
/archive/, not just the ones for tags -- can now be built into an sdist or awheel with a correct version. A
.git_archival.txtfile, expanded bygit archiveviaexport-subst, carries the describe output thatsetuptools-scmwould otherwise have to askgitfor.The archival file deliberately omits
ref-names. Recording it would make thecontent of an archive -- and therefore its checksum -- change whenever a branch
that points at the archived commit is moved afterwards, which is how the
equivalent tarballs of tagged releases became unstable for other projects.
Describe output alone depends only on the commit and on the tags reachable from
it.
The archival file is not shipped in the sdist -- an sdist carries its version
in its metadata, and the template is only ever expanded by
git archive.Building pytest from a git checkout now requires
setuptools-scm >= 10.1.#9236:
pluggy>=1.6is now required.Older pluggy versions stopped running the remaining hook wrapper teardowns once an
old-style
hookwrapperraised in its own teardown. The logging plugin's wrapper wasleft suspended, and when it was finalised later it detached the handler that
caplogof a subsequent test was recording through, so that test saw no records.Contributor-facing changes
DistFacadecompatibility class.Miscellaneous internal changes
FixtureRequest.node <pytest.FixtureRequest.node>is now explicitly annotated as returningAny, so strict pyright no longer reports it asUnknown. The return type stays deliberately broad, see13888.::-separated strings identifying collected items, e.g.path/to/test_file.py::TestClass::test_method[param]) are now represented internally by a single structured~_pytest.nodeid.NodeIddataclass instead of being repeatedly re-parsed as plain strings. The publicnodeid: strattribute on nodes and reports is unchanged and remains fully backward compatible for plugins.pytest-dev/pytest-asyncio (pytest-asyncio)
v1.5.0: pytest-asyncio v1.5.0Compare Source
1.5.0 - 2026-11-12
Added
fixture-scopevalue forasyncio_default_fixture_loop_scope. With it, an asynchronous fixture withoutloop_scopeuses itsscopeas its loop scope. This is the behavior of an unset option in pytest-asyncio 1.x. (#1095)Fixed
asynciomarker, including an invalidloop_scope, are reported with a message that names the test instead of a traceback. (#1584)tox-dev/sphinx-autodoc-typehints (sphinx-autodoc-typehints)
v3.17.5Compare Source
🐍 PyPI release
📜 Full Changelog
v3.17.4Compare Source
Bug fixes
typealias an inventory lists as data, such asnumpy.typing.NDArray, resolves through intersphinx (#1105) by @gaborbernat.🐍 PyPI release
📜 Full Changelog
v3.17.3Compare Source
Packaging
🐍 PyPI release
📜 Full Changelog
v3.17.2Compare Source
Bug fixes
🐍 PyPI release
📜 Full Changelog
v3.17.1Compare Source
Packaging
🐍 PyPI release
📜 Full Changelog
v3.17.0Compare Source
Features
typehints_modernize_annotationsto render legacytypingaliases with current Python spelling. #1085 by @gaborbernat."optional"default style that marks parameters callers may omit without printing their default values. #1086 by @gaborbernat.Annotatedmetadata classes other thanDoc. #1088 by @gaborbernat.Documentation
autodoc_preserve_defaultsapplies to default values in parameter descriptions. #1087 by @gaborbernat.🐍 PyPI release
📜 Full Changelog
v3.16.0Compare Source
Features
Docdescriptions from annotated-doc as well as typing_extensions. #1083 by @gaborbernat.Bug fixes
html_static_path. #1079 by @gaborbernat.autodoc_use_type_comments = Falsewhen reading types from a function's# type:comment. #1080 by @gaborbernat.TYPE_CHECKINGguard's branch; skip guards whose other conditions cannot be determined without running code. #1081 by @gaborbernat.:Returns:fields, and recognize numpydoc return types containing spaces. #1082 by @gaborbernat.Documentation
NamedTuplereferences in both modes. #1084 by @gaborbernat.🐍 PyPI release
📜 Full Changelog
v3.15.0Compare Source
Features
Annotatedmetadata in rendered types withtypehints_annotated_metadata. #1071 by @gaborbernat.typehints_formatterreceive whether a type appears in a parameter, return, attribute or base whentypehints_formatter_pass_contextis enabled. #1073 by @gaborbernat.Bug fixes
typehints_formatterchanges in an incremental build, so their rendered types update withoutsphinx-build -E. #1069 by @gaborbernat.html_static_pathas changed after the extension adds its stylesheet. #1070 by @gaborbernat.Cannot subscript already-subscripted typing.Generic. #1074 #1075 by @gaborbernat.Annotatedmetadata raises during comparison with the getter type, instead of failing the build. #1076 by @gaborbernat.:Returns:field instead of interpreting it as a numpydoc return type. #1077 by @gaborbernat.if TYPE_CHECKING and Falseinstead of resolving types the branch never defines. #1078 by @gaborbernat.Packaging
🐍 PyPI release
📜 Full Changelog
v3.14.0Compare Source
Features
Generic[T]subclasses in their signature, such asbounded[T: int](value)andclass Stack[T], with bounds, constraints and defaults. #867 by @gaborbernat.**kwargs: Unpack[TypedDict]documented as keyword arguments. #869 by @gaborbernat.Overridesfield linking the base class member a@typing.overridemethod replaces. #871 by @gaborbernat.TypedDictkeys a caller may omit asNotRequired[...], and add anExtra itemsfield for PEP 728extra_itemsandclosedclasses. #873 by @gaborbernat.@warnings.deprecatedmessage as a deprecation notice below the summary. #874 by @gaborbernat.Tinstead ofTypeVar(T, bound=int), andP.argsandP.kwargsof a declared**Pas written. #888 by @gaborbernat.Yieldssection when its docstring has aReturnssection and the third argument of itsGeneratorannotation is notNone. #889 by @gaborbernat.typehints_use_signature, render the types kept in the signature the same way as the fields, so stubs, guarded imports, documented aliases,typehints_fixup_module_name,typehints_fully_qualifiedandtypehints_formatterapply to both. #895 by @gaborbernat.sentinelbuiltin andtyping_extensions.Sentinel, by their name as apy:datareference instead of linking a nonexistentsentinelclass. #912 by @gaborbernat.:ivar:,:cvar:and:var:fields from class-body and base class annotations, droppingClassVarandFinalqualifiers. #916 by @gaborbernat.Yieldsentry of a generator, keeping a type the docstring already states. #917 by @gaborbernat.napoleon_use_ivar = False, type.. attribute::entries fromself.x: Tannotations in__init__unlessnapoleon_attr_annotations = False; a:type:already in the docstring wins. #918 by @gaborbernat.Setter acceptsfield to a property whose setter annotation differs from the getter's return type. #919 by @gaborbernat.intforTunderclass IntStore(Store[int]), in what it inherits from a generic base, covering parameter and return fields, overloads, attribute and property headers,:vartype:fields andTypedDictkeys. #921, #926, #938 by @gaborbernat.list(), in the signature and fields instead of<factory>orNOTHING; lambdas, partials and nested functions keep<factory>. #922 by @gaborbernat.typehints_fully_qualified,typehints_formatterandtypehints_fixup_module_nameapply there and the header dropsAnnotatedmetadata. #923 by @gaborbernat.typehints_use_signature = True, show one typed signature line per@overloadinstead of the untyped implementation signature and anOverloadsfield. #927 by @gaborbernat.TypeVar,ParamSpecorTypeVarTuplein docstring fields by its name, as a PEP 695 type parameter renders, instead of as aTypeVar(B, bound=Sized)call, andP.argsas written instead ofParamSpecArgs(P); the name links to the variable's own entry, which shows its bound or constraints, when the docs document it. #1044 by @gaborbernat.'fill'inLiteral["fill", 1, None], now take the colors of the project's Pygments style in light and dark mode, the way a code block does. #1046 by @gaborbernat.TypeVarTupleor tuple with the PEP 646 star,*Tsand*tuple[int, ...], instead ofUnpack[Ts];Unpack[TypedDict]for**kwargskeeps its spelling. #1047 by @gaborbernat.typing.ContextManager[None]renders asContextManager[None]instead ofContextManager[None, bool | None]on Python 3.13 and later; a writtentyping.Generator[int, None, None]renders asGenerator[int]. #1054 by @gaborbernat.Bases:line of a class documented with:show-inheritance:as a docstring field renders a type, so a generic base such asMapping[str, int]followstypehints_fully_qualified,typehints_fixup_module_nameandtypehints_formatter, and a PEP 695 base such asStack[T]keeps its argument where Sphinx left it off. #1058 by @gaborbernat.Bug fixes
"a positive float", as written with aforward_referencewarning instead of aborting the build. #789 by @gaborbernat.sphinx.ext.autodocbefore connecting to its events, so listing this extension first or omitting autodoc no longer fails withUnknown event name. #790 by @gaborbernat.QuerySet[...]withoutdjango-stubs-ext, also for a function re-exported with a rewritten__module__, instead of showing them all as plain text. #791, #901 by @gaborbernat.__new__from__new__, asinspect.signaturedoes, and take a__new__-only class's parameter types from__new__rather than the class body. #792 by @gaborbernat.x, y : int, and singleYieldsentries. #793 by @gaborbernat.typehints_defaults = "braces-after"defaults on their own parameter when the docstring already has:type:lines or napoleon types. #794 by @gaborbernat.:returns:withtypehints_use_rtype = Falsewhen the docstring also has:rtype:or its:returns:text starts on the next line. #795 by @gaborbernat.Attributesentries from the class,__init__and module annotations, linked and rendered like other fields withtypehints_fully_qualified, type aliases andtypehints_formatterapplied, unlessnapoleon_attr_annotationsisFalse. #796, #966 by @gaborbernat.staticmethod(func)and of name-mangled private static methods. #797 by @gaborbernat.Overloadslist, leave outselfandcls, and show constructor overloads. #798 by @gaborbernat.always_document_param_typeschanges, so incremental builds pick up the new setting. #799 by @gaborbernat.:no-overloads:marker from the output whentypehints_document_overloadsisFalse. #800 by @gaborbernat.typehints_use_signature_return = True. #801 by @gaborbernat.strip_signature_backslashis on. #802 by @gaborbernat.sphinx.ext.doctestfailures on the line the example starts on instead of one line late. #804 by @gaborbernat.Callableparameterized by aParamSpecorConcatenateasCallable[P, int]instead ofCallable[[P], int]. #805 by @gaborbernat.P.argsandP.kwargsasParamSpecArgs(P)andParamSpecKwargs(P), naming theParamSpecthey belong to. #806 by @gaborbernat.TypeVarorParamSpecafter its bound, as inTypeVar(T, bound=int, default=bool). #807 by @gaborbernat.ParamSpecargument list of a generic as a list,Box[[int, str]], instead of a tuple. #808 by @gaborbernat.TypeVarTupleasTypeVarTuple(Ts)linking totyping.TypeVarTuple, instead of a broken reference that loses its name. #809 by @gaborbernat.Optional[Union[...]]withsimplify_optional_unions = Falseon Python before 3.14. #810 by @gaborbernat.typing_extensions.TypeAliasTypeliketypestatement aliases instead of as theTypeAliasTypeclass. #811 by @gaborbernat.TYPE_CHECKINGguards written through a module alias, on one line, with a comment at column zero, or at the end of the file. #813 by @gaborbernat.from __future__ import annotations. #814, #941 by @gaborbernat.TYPE_CHECKINGregardless of which class Sphinx documents first. #815 by @gaborbernat.Keyword Argssection, and stopalways_document_param_typeslisting them again under Parameters. #816 by @gaborbernat.autoclass_content = "both", list each constructor parameter once, including those added throughalways_document_param_typesorDoc(). #817, #925 by @gaborbernat.sys.version_infoandsys.platformblocks, merge all overload variants of a stub function, and check re-exports. #818 by @gaborbernat.typesclasses, so references resolve againstdocs.python.org/3when building on Python 3.12. #820 by @gaborbernat.TYPE_CHECKINGin attrs class fields, fixing aNameErrorcrash on Python 3.14 and a forward-reference warning on older versions. #822 by @gaborbernat.AttributeErrorbuilding man pages for projects that document a function taking atypealias. #823 by @gaborbernat.typehints_formatterwhen rendering signatures withtypehints_use_signatureortypehints_use_signature_return. #824 by @gaborbernat."* weird"annotation stays inline instead of producing a list or crashing. #825 by @gaborbernat.forward_referencewhen an annotation names a missing module attribute, such astyping.TypeIson Python 3.12. #826 by @gaborbernat..pyistub refers to throughimport pkg.suborfrom . import sub, so their types are no longer dropped. #827 by @gaborbernat.__module__find theirTYPE_CHECKINGimports. #828 by @gaborbernat... py:type::by name only where the annotation writes that name, whatever order Sphinx reads the entries in, sointparameters no longer render as an alias ofint. #829, #920 by @gaborbernat.typealias to its own entry, not to a different documented alias with the same name. #831 by @gaborbernat.TYPE_CHECKINGguards written asif not TYPE_CHECKING: ... else:or combined withand/or, so their imports resolve. #833 by @gaborbernat.self.x: Tannotations nested inif,try,with, loop andmatchblocks of__init__, and handle a positional-onlyself. #834 by @gaborbernat.functools.cacheandfunctools.lru_cachedecorated functions on Python 3.14. #835 by @gaborbernat.napoleon_use_param = Falseornapoleon_use_keyword = False, type the entries of napoleon's combined parameter field, add their defaults, and list the parameters the extension adds there instead of in a second list. #836, #886 by @gaborbernat.typehints_defaults, show the default of a parameter typed only in its docstring, whether through a:type:line, a:param str c:field or napoleon's combined parameter field. #837, #884 by @gaborbernat.functools.singledispatchimplementations, so each registered implementation shows the type it dispatches on. #838 by @gaborbernat.typehints_defaults, render defaults the way the signature does, without memory addresses, with enum members by name and with sets in a stable order. #839 by @gaborbernat.Literalvalues, soLiteral["\n"]renders as written instead ofLiteral['n']. #840 by @gaborbernat.tuple[()]andTuple[()]as the empty tuple instead of a baretuple. #841 by @gaborbernat.typealias passed to itself, such asPair[Pair[int]], instead of dropping the inner arguments. #842 by @gaborbernat.Literalmember, asLiteral[Color.RED, Light.RED]instead ofLiteral[RED, RED]. #843 by @gaborbernat.autodoc_mock_importsmodules, such asNDArray[float64], instead of a garbled name. #844 by @gaborbernat.Literalstring values that contain backticks without breaking the surrounding markup. #845 by @gaborbernat.|of a union around anautodoc_type_aliasesname on Python 3.12 and 3.13 instead of renderingOptionalorUnion. #846 by @gaborbernat.typehints_defaults = "braces-after", put the default at the end of the last plain paragraph of the description instead of inside a trailing list, directive or literal block. #847 by @gaborbernat...placeholder as no description, so entries without one render asx (int)instead ofx (int) -- ... #848 by @gaborbernat.Returnsdescription of a function without parameters instead of adding a second one from the return annotation'sDoc(). #849 by @gaborbernat.Returnssection that only describes the value. #850 by @gaborbernat.:rtype:as a whole instead of leaving its continuation lines behind. #851 by @gaborbernat.typehints_use_rtype = False, add the return type to a:returns:description that contains--. #852 by @gaborbernat.always_document_param_typesorDoc()in signature order among the documented ones instead of in a second list at the end of the docstring. #853 by @gaborbernat.versionadded,deprecatedandseealso, as before docutils directives likenote. #854 by @gaborbernat.autodoc_type_aliasestargets documented as module data or in a document read later. #855 by @gaborbernat.typealias even when the docs have no entry for its arguments, instead of expanding it. #856 by @gaborbernat.typing_extensionsprovides on the running Python, such asTypeIson 3.12 andDoc, totyping_extensionsinstead of a missingtypingtarget. #857 by @gaborbernat.@overloadsignatures from other Sphinx apps in the same process that do not load the extension. #858 by @gaborbernat... include::. #859 by @gaborbernat.typehints_defaultsortypehints_formattervalue as a configuration error instead of an extension crash with a traceback. #860 by @gaborbernat.Unparsable type hint commentfor a lambda assigned with a variable type comment. #861 by @gaborbernat.Returnsentry, renderingratio (float) -- The ratio.. #862 by @gaborbernat.undoc-members. #863 by @gaborbernat.Other Parameterssection under its own heading instead of merging it intoParameters. #864 by @gaborbernat.*tuple[...], rendering it asUnpack[tuple[...]]. #865 by @gaborbernat.TypeVar(T, bound=int)without a gap after the=. #866 by @gaborbernat.InitVar[...]instead of rendering a bareInitVar. #875 by @gaborbernat.:param:,:returns:and:yields:lines inside literal blocks and directives alone instead of treating them as the object's fields. #876 by @gaborbernat.functools.partialobjects,partialmethodand callable instances from the function they call, moving the annotations from the signature into the fields. #877 by @gaborbernat.typealias whose value raises, as written instead of crashing the build or dropping the object. #878, #897 by @gaborbernat.list assignment index out of range. #879 by @gaborbernat.clsfrom the signature of a@classmethodstacked on another decorator such asfunctools.cache. #880 by @gaborbernat.cannot cache unpickleable configuration valuewarning fortypehints_fixup_module_name, and re-read the documents only when that function's source changes instead of on every build. #881, #937 by @gaborbernat.__signature__, such as a pydantic model, by that signature with its fields typed. #883 by @gaborbernat.typehints_use_rtype = False, keep the inline return type when the docstring had an:rtype:field. #885 by @gaborbernat.__init__that dataclasses and attrs generate, instead of pointing at a nonexistent<string>.rst. #887 by @gaborbernat.autodoc_typehints_format = "fully-qualified":typehints_fully_qualifieddefaults to it, so fields, signatures and type parameters show full paths. An explicittypehints_fully_qualifiedstill wins. #891 by @gaborbernat.Gen[int] | Noneunderautodoc_mock_imports, which rendered as unlinked text on Python 3.14 or withfrom __future__ import annotations. #892 by @gaborbernat..pyistub imports with a plainimportbut that is not installed, so types likenpt.NDArrayrender as cross-references instead of raising aforward_referencewarning. #893 by @gaborbernat... autofunction:: mod.f(x, y=3), or read from the docstring throughautodoc_docstring_signature, when it names other parameters than the object takes, instead of replacing it with the object's own. One made of*argsand**kwargsalone still gets the object's parameters (#903). #896 by @gaborbernat.typealiases that parameter types use, fixing broken cross-references. #902 by @gaborbernat..pyistub when the stub changes, so an incremental build picks up the edit. #905 by @gaborbernat.Overloadslist after a literal block that the summary introduces with::, instead of inside it. #907 by @gaborbernat./and keyword-only*markers, defaults and the stars of*argsand**kwargsinOverloadsentries. #910 by @gaborbernat.bound="None"on an unboundParamSpecon Python 3.12 and early 3.13 releases. #911 by @gaborbernat.typehints_defaultsmode. #913 by @gaborbernat.Explicit markup ends without a blank linewarning when numpydoc processes a docstring without aParameterssection and the extension adds parameter types. #914 by @gaborbernat.Annotated[Iterator[int], {"unit": "m"}], when its docstring has aYieldssection. #928 by @gaborbernat.__signature__a decorator sets on its wrapper, so the signature and the parameter fields list the parameters callers pass instead of those of the wrapped function; the signature shows types only undertypehints_use_signature. #931, #946 by @gaborbernat.Annotatedreturn annotation before checking it, so anAnnotatedgenerator types itsYieldsentry instead of getting a return type, andtypehints_document_rtype_none = FalsehidesAnnotated[None, ...]. #939 by @gaborbernat.if TYPE_CHECKING:block without writing to the module, so the names it defines, such as@overloadstubs orProtocolclasses, no longer replace runtime bindings or show up as module members. #960, #984 by @gaborbernat.__class_getitem__returns the class itself, such as SQLAlchemy'sTypeEngine[Any], in headers, parameter and return fields, signatures, overloads and:ivar:types. #963 by @gaborbernat.if TYPE_CHECKING:blocks import from each other to their guarded types, whichever module Sphinx documents first. #969 by @gaborbernat.:Parameters:definition lists, alone instead of converting them as numpydoc output into empty:param :fields. #979 by @gaborbernat.TYPE_CHECKING, such as_t.UriTypein requests, instead of dropping the type. #985 by @gaborbernat..pyistub defines to the module the stub describes, such asext.Config, instead of rendering it as an unresolved bareConfigwhen the function is documented through a re-export. #1011 by @gaborbernat.typing.NewTypeclass ortyping.Annotatedform; the field names it as a reference, as the signature does. #1013 by @gaborbernat... autofunction::no longer listselfand-> None, and overloads of a function documented with.. autoclass::keep their first parameter and return types. #1014 by @gaborbernat.typeno longer links to an unrelated documented attribute such asEvent.type. #1017 by @gaborbernat.py:classand apy:dataname to the class entry on every build, instead of a name that changed withPYTHONHASHSEED. #1018 by @gaborbernat.NamedTuple's__new__on its own, withautomethodor:special-members: __new__, no longer warns that names its fields use, such asPathorint, are not defined. #1019 by @gaborbernat.__init__now shows the type a subclass binds, such asintforclass Child(Base[int]), in that subclass's:ivar:fields instead of the bare type parameter. #1020 by @gaborbernat.autodoc_mock_importswithout aborting the build withwrapper loop when unwrapping. #1022 by @gaborbernat.__annotations__, so the object reads the same after a build as before it. #1025 by @gaborbernat.typehints_formatterthat returns an empty string now hides the type instead of showing raw:sphinx_autodoc_typehints_type:text in docstring fields or an empty annotation such asattr: = 1in headers and signatures, and drops theReturn type,Extra itemsandSetter acceptsfields it empties. #1026 by @gaborbernat.typestatement or a module-levelCount = int, used by anautomodulemember when itspy:typecomes later on the page, instead of expanding it to its value. #1027 by @gaborbernat.TYPE_CHECKINGblock of each module again in every build, so a module edited and imported again between builds renders its new types. #1030 by @gaborbernat.Literalto its enum when the member has no entry of its own, such as a member without a docstring under:members:, instead of rendering it unlinked with anitpickywarning. #1035 by @gaborbernat.None,TrueandFalseinsideLiteral[...]to the Python docs, the same wayNonelinks outside it, instead of rendering them as plain text. #1038 by @gaborbernat.__module__isNonenow renders as its bare name instead of a broken cross-reference toNone.ClassName. #1039 by @gaborbernat.typehints_formatterwrites between bare backticks inside an expanded type alias now use the document'sdefault_role, as they do outside an alias, instead of rendering as italic text. #1040 by @gaborbernat.napoleon_attr_annotationsoff for napoleon in its own project only, so the first build after upgrading re-reads every document once. #1048 by @gaborbernat.SphinxTransform.appwhen choosing between an alias and its value, so builds that run Python with-W error::PendingDeprecationWarningno longer fail, and the extension keeps working on Sphinx 11. #1051 by @gaborbernat.sphinx_autodoc_typehintsinextensions; the order the two are listed in no longer matters. #1052 by @gaborbernat.:rtype:or:return:now goes where Sphinx puts the types it adds withautodoc_typehints = "description": after the last field of the first list with parameters, else of the last field list, else in a new list at the end of the docstring. It no longer sits before a definition list, a doctest block, a rubric or a section, nor below a table that follows the parameters, and a warning from a docstring reports its own line rather than one counting the lines added above it. #1060 by @gaborbernat.Callableparameter list. #1064 by @gaborbernat.napoleon_attr_annotationsat its configured boolean value for other Sphinx extensions. #1067 by @gaborbernat.Documentation
typingwith the extension, with controls in each example'sconf.pytab that switch the options it renders with. #1029 by @gaborbernat.Packaging
typing-extensions>=4.16on Python versions before 3.15. #986 by @gaborbernat.🐍 PyPI release
📜 Full Changelog
Configuration
📅 Schedule: (in timezone Europe/Berlin)
* 0-5 * * 1)🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
👻 Immortal: This PR will be recreated if closed unmerged. Get config help if that's undesired.
This PR has been generated by Mend Renovate CLI.
acb84724c1tod7471029f0fix(deps): update dependency sphinx-autodoc-typehints to <3.14.1to fix(deps): update dependency sphinx-autodoc-typehints to <3.15.1d7471029f0toe1ebc33b9ffix(deps): update dependency sphinx-autodoc-typehints to <3.15.1to fix(deps): update dependency sphinx-autodoc-typehints to <3.16.1e1ebc33b9fto037a550cddfix(deps): update dependency sphinx-autodoc-typehints to <3.16.1to fix(deps): update dependency sphinx-autodoc-typehints to <3.17.2037a550cddto9919627efbfix(deps): update dependency sphinx-autodoc-typehints to <3.17.2to fix(deps): update dependency sphinx-autodoc-typehints to <3.17.39919627efbto137738473efix(deps): update dependency sphinx-autodoc-typehints to <3.17.3to fix(deps): update all non-major dependencies137738473etoba4b7e1036Edited/Blocked Notification
Renovate will not automatically rebase this PR, because it does not recognize the last commit author and assumes somebody else may have edited the PR.
You can manually request rebase by checking the rebase/retry box above.
⚠️ Warning: custom changes will be lost.
ba4b7e1036toc081418ccbc081418ccbe1201cc426e1201cc426ab71c29a47View command line instructions
Checkout
From your project repository, check out a new branch and test the changes.Merge
Merge the changes and update on Forgejo.Warning: The "Autodetect manual merge" setting is not enabled for this repository, you will have to mark this pull request as manually merged afterwards.