Conversation
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds the
floorsjob the org standard requires (standard.md §7), following modern-di-fastapi#56. This repo surfaced four separate problems, three of them in the shipped extras.1. The pydantic extra could not install on 3.12 or 3.13
pydantic>=2.0resolves to pydantic 2.0, whose pydantic-core has nocp312wheel:The existing
>=2.12bound for 3.14 had the right idea; the gap below it was unguarded. Bisected on each interpreter — 2.0 fails and 2.1 passes on 3.12; 2.7 and 2.8 fail and 2.8.1 passes on 3.13:2. The msgspec extra could not install on 3.13
Same shape.
msgspec>=0.18.5is correct for thecan_decodereason the comment gives, but 0.19.0 is the first release with acp313wheel, and the< '3.14'band swallowed 3.13:3. hypothesis could not import on 3.14
hypothesiswas unbounded in[dependency-groups], so the floors leg resolved 6.0.0, which callsimportlib_metadata.entry_points().get(...)— removed fromimportlib.metadatalong ago, fatal on 3.14:Floored at
>=6.20, verified: 6.5.0 and 6.10.0 fail, 6.20.0 passes.4.
--no-install-projectbreaks the isolation tests hereWorth flagging beyond this repo.
httpwareis the only src-layout package in the org, and[tool.pytest.ini_options] pythonpath = ["src"]is what makesimport httpwarework without an installed copy. That setting applies to the pytest process only.tests/test_optional_extras_isolation.pyandtests/test_optional_extras_pydantic_missing.pydeliberately spawn a fresh interpreter, which inherits none of it, and--no-install-projectleaves nothing installed to fall back on:Four tests failed for a reason that has nothing to do with floors. Setting
PYTHONPATH: srcon the step gives the child the same path pytest gives the parent.--no-install-projectcannot be dropped — it is what makes--no-buildusable at all, since--no-buildotherwise refuses to build this project too.Why the job exists
pytestresolves every dependency at its newest, so the bottom of each declared range ships unexercised. This job resolves direct dependencies at their floors, wheel-only, on every entry of the same matrixpytestuses, and runs the suite against them. The matrix here folds in3.14t, whichpytestcovers through the separatepytest-freethreadedjob — §7 asks for every matrix entry, and free-threaded wheel coverage differs from the GIL build's.Verification
818 passed at the floors on 3.11, 3.12, 3.13, 3.14 and 3.14t. The job installs
httpx2==2.0.0,pydantic==2.8.1,msgspec==0.19.0andopentelemetry-api==1.20.0on 3.13 — each exactly at its declared bound, confirmed by reading back the install list rather than assuming. Resolved-newest path re-checked withjust install && just lint-ci && just test-ci. Local runs are macOS/arm64; CI is the verdict on manylinux wheel coverage.