Bump propcache from 0.4.1 to 0.5.2

[dependabot]

Bumps propcache from 0.4.1 to 0.5.2.

Release notes

Sourced from propcache's releases.

0.5.2

0.5.0 and 0.5.1 were tagged earlier today but never reached PyPI: 0.5.0's deploy failed at cibuildwheel's post-build pytest on free-threaded armv7l musllinux (SIGBUS under QEMU emulation while importing the C extension), and 0.5.1's deploy hit a transient sigstore Rekor 502 during the attestation step. 0.5.2 is the first of the three to actually publish.

Features

• Added support for newer type hints and remove Optional and Union from all annotations -- by :user:Vizonex

  Related issues and pull requests on GitHub: #193.

Removals and backward incompatible breaking changes

• Dropped support for Python 3.9 as it has reached end of life.

  Related issues and pull requests on GitHub: #216.

Packaging updates and notes for downstreams

• Changed the Cython build dependency from ~= 3.1.0 to >= 3.2.0, removing the upper version bound to avoid conflicts for downstream packagers -- by :user:jameshilliard and :user:gundalow.

  The upstream Cython version is pinned to 3.2.4 in the CI/CD environment.

  Related issues and pull requests on GitHub: #184, #188, #214.

• Start building and shipping riscv64 wheels -- by :user:justeph.

  Related issues and pull requests on GitHub: #194.

• The :pep:517 build backend now supports a new build-inplace config setting (and PROPCACHE_BUILD_INPLACE environment variable) for controlling whether to build the project in-tree or in a temporary directory. It only affects wheels and is set up to build in a temporary directory by default. It does not affect editable wheel builds; they will keep being built in-tree regardless.

  Here's an example of using this setting:

  .. code-block:: console

... (truncated)

Changelog

Sourced from propcache's changelog.

0.5.2

(2026-05-08)

No significant changes.

---

0.5.1

(2026-05-08)

No significant changes.

---

0.5.0

(2026-05-08)

Features

• Added support for newer type hints and remove Optional and Union from all annotations -- by :user:Vizonex

  Related issues and pull requests on GitHub: :issue:193.

Removals and backward incompatible breaking changes

• Dropped support for Python 3.9 as it has reached end of life.

  Related issues and pull requests on GitHub: :issue:216.

Packaging updates and notes for downstreams

... (truncated)

Commits

• 041f7af Release 0.5.2 (#222)
• 9c374aa Release 0.5.1 (#221)
• c4e2b78 Skip cibuildwheel test on free threaded armv7l musllinux (#220)
• e60bded Release 0.5.0 (#219)
• b88f846 Make wheel builds reproducible (#218)
• fdacd6c Update Typehint annotations (#193)
• 483e9d3 [pre-commit.ci] pre-commit autoupdate (#204)
• d0d45e1 Drop Python 3.9 support (#216)
• c0bacf5 Update pytest-cov requirement from >=2.3.1 to >=7.1.0 (#210)
• 9bfb3e4 Bump pytest-codspeed from 4.3.0 to 4.5.0 (#211)
• Additional commits viewable in compare view

Note
Automatic rebases have been disabled on this pull request as it has been open for over 30 days.

[codspeed-hq]

Merging this PR will degrade performance by 12.66%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 21 regressed benchmarks
✅ 76 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_empty_query	44.3 µs	53.9 µs	-17.8%
❌	test_query	44.4 µs	53.8 µs	-17.52%
❌	test_query_string	45.1 µs	54.7 µs	-17.51%
❌	test_empty_query_string	44.9 µs	54.1 µs	-17.05%
❌	test_empty_path_safe	49.2 µs	58.6 µs	-15.95%
❌	test_empty_path	49.2 µs	58.5 µs	-15.92%
❌	test_raw_path_qs_with_query	50.7 µs	60.1 µs	-15.59%
❌	test_empty_raw_path_qs	50.5 µs	59.8 µs	-15.44%
❌	test_raw_path_qs	50.7 µs	59.6 µs	-14.98%
❌	test_path_safe	51.5 µs	58.8 µs	-12.51%
❌	test_url_host_port_subcomponent	451.6 µs	514.1 µs	-12.16%
❌	test_url_origin_with_user_pass	76.7 µs	86.2 µs	-11.04%
❌	test_url_with_path_origin	77.3 µs	86.8 µs	-10.93%
❌	test_url_origin	77.2 µs	86.6 µs	-10.82%
❌	test_is_default_port	165.2 µs	183.1 µs	-9.79%
❌	test_empty_query_string_uncached	72.3 µs	79.5 µs	-8.98%
❌	test_empty_path_safe_uncached	80.9 µs	88.3 µs	-8.45%
❌	test_url_with_password	300.5 µs	328.1 µs	-8.41%
❌	test_url_with_port	307.4 µs	334.3 µs	-8.04%
❌	test_url_with_user	299.3 µs	324.8 µs	-7.86%
...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing dependabot/pip/propcache-0.5.2 (cdf896e) with master (ccd2648)¹}

Footnotes

1. No successful run was found on master (dccaecb) during the generation of this report, so ccd2648 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[bdraco]

might be a real regression .. need to check

[bdraco]

@dependabot recreate

[bdraco]

regression is real but I think its Cython as nothing of substance has changed in propcache

[Vizonex]

regression is real but I think its Cython as nothing of substance has changed in propcache

@bdraco Incase I did propose one time to try migrating the project to C at some point, but a second option I have is that we could bring this up with the Cython developers incase there were any hiccups in the newer code generation.

[webknjaz]

@bdraco I'm wondering if the Cython version is still being pinned after the uv switch — with the new frontend, I'm not sure if it picks up the PIP_CONSTRAINT set in the cibuildwheel config and apply it to the ephemeral build env the same way pip does. Have you checked? Also, I remember that sometimes uv would pick up the cache when we'd rather not have it.

Oh, and if we get to return to pip, it's probably a good idea to look into also setting the PIP_BUILD_CONSTRAINT env var that the recent pip versions now support.

[bdraco]

I'm wondering if the Cython version is still being pinned after the uv switch — with the new frontend, I'm not sure if it picks up the PIP_CONSTRAINT set in the cibuildwheel config and apply it to the ephemeral build env the same way pip does. Have you checked?

Yes, its a problem. All sorted now

Also, I remember that sometimes uv would pick up the cache when we'd rather not have it.

There is specific logic to disable the cache on pre-release builds added

[coveralls]

Coverage Report for CI Build 27616098591

Coverage remained the same at 98.74%

Details

• Coverage remained the same as the base build.
• Patch coverage: No coverable lines changed in this PR.
• No coverage regressions found.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

No coverage regressions found.

---

Coverage Stats

Relevant Lines:	6210
Covered Lines:	6131
Line Coverage:	98.73%
Relevant Branches:	61
Covered Branches:	61
Branch Coverage:	100.0%
Branches in Coverage %:	Yes
Coverage Strength:	0.99 hits per line

---

💛 - Coveralls

[Vizonex]

Regression is still visible I would recommend aio-libs/propcache#245 Which seeks to defend against these regressions using a small amount of C code.