ci(mergify): require the Python 3.13 and 3.14 checks - #311
Merged
mergify[bot] merged 2 commits intoSep 3, 2026
Merged
Conversation
The classifiers stopped at 3.12 and so did the matrix, while both internal consumers are already past it: the monorepo `engine` and `shadow-office` each declare `requires-python = "~=3.14.0"` and pin `sql-compare==0.2.0`. The library is consumed on 3.14 and tested on neither 3.13 nor 3.14. `requires-python = ">=3.10"` is an installability floor, not a support claim, and it stays unbounded on purpose: a ceiling there breaks `pip install` on every Python released after it. The classifiers are the support claim, so they are what this corrects. They only reach users on the next release; PyPI's current 0.2.0 still advertises `>=3.9` and stops at 3.12. Nothing else has to move. `python = "^3.10"` already spans both, so `poetry.lock`'s `python-versions` and `content-hash` are unchanged (`poetry check --lock` exits 0; classifiers are not hashed), and every locked dependency ships cp313 and cp314 wheels, so no leg builds from source. Free-threaded 3.14t is left out because nothing here exercises it, not because it would need different artifacts: `sql_compare` is one pure-Python module, so `poetry build` emits a single `py3-none-any` wheel that 3.14t installs unchanged. Adding that leg is its own decision. Verified on 3.13.10 and 3.14.6 in clean environments built from `poetry.lock`: ruff, ruff format, deptry and mypy --strict clean, 64 tests passing on both. The 3.14 leg prints two `SyntaxWarning: 'return' in a 'finally' block` from pluggy 1.5.0 whenever the bytecode cache is cold, which on CI is always. They are emitted while pluggy itself is compiled, before pytest installs `filterwarnings = ["error"]`, so they are noise on an exit-0 run. Bumping pluggy to 1.6.0 would silence them and belongs in its own change against the lock. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MKmEnjRuAM4NcTYxRXG3BA Change-Id: I08ce7a6db8d33895e7135f648c2afa9723ed0852
The two matrix legs added in the previous commit are advisory on their own. `.mergify.yml`'s `&CheckRuns` anchor is this repository's only merge gate: `branches/main/protection` returns no `required_status_checks`, and none of the eleven active rulesets carries one. So a red 3.13 or 3.14 leg would land silently. The ordering mirrors #299 -> #300 but not the reasoning, which is worth stating rather than assuming symmetric. Removing a check needed its own pull request first because Mergify reads its configuration from the default branch: a pull request deleting the job would still have been gated on a check that no longer reported, and could never merge. Adding one has no such deadlock. Both pull requests here are evaluated against main's configuration, which does not name the new checks, so a single atomic commit would have merged just as well. What the split actually buys is revertability. If a new leg turns out to be red on main, reverting this commit alone unblocks the repository while keeping the coverage the previous commit added; an atomic change would have to give up both. Two consequences worth knowing, neither of which ordering can avoid. Because main's configuration does not require the new checks until this lands, nothing forces either pull request of this stack to be green on 3.13 or 3.14. Both run the full five-leg matrix on their own head, so the legs are observable, but they have to be looked at before this one is merged rather than assumed. Every pull request already open when this lands keeps the check-run set from its last CI run, and no `pull_request` event fires when the base moves. `check-success=Test with Python 3.13` can therefore never go true on it, and `auto_merge_conditions: true` folds the success conditions into the queue conditions, so it cannot enter the queue to be rebased out either. Each one needs a push, a rebase or a close/reopen to pick up the new legs. Longer term, this list is a hand-maintained mirror of the job names and will want the same ceremony at 3.15. The monorepo and mergify-cli both replaced it with one aggregate job (`all-greens` / `ci-gate`) behind a single `check-success`, which is worth adopting here on its own. Note that `check-success~=^Test with Python ` is not a shortcut for it: check-success is a list attribute and `~=` matches any element, so one green leg would satisfy it while another fails. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MKmEnjRuAM4NcTYxRXG3BA Change-Id: I7054c8de8899a66bb549fbd9b29452d406e052ee
Member
Author
|
This pull request is part of a Mergify stack:
|
Contributor
Merge Protections🟢 All 7 merge protections satisfied — ready to merge. Show 7 satisfied protections🟢 ⛓️ Depends-On RequirementsRequirement based on the presence of
🟢 🤖 Continuous Integration
🟢 👀 Review Requirements
🟢 Enforce conventional commitMake sure that we follow https://www.conventionalcommits.org/en/v1.0.0/
🟢 🔎 Reviews
🟢 📕 PR description
🟢 🚦 Auto-queueWhen all merge protections are satisfied, this pull request will be queued automatically. |
sileht
marked this pull request as ready for review
September 2, 2026 07:56
Base automatically changed from
devs/sileht/py313-py314-matrix/test-against-python-3-13-3-14--08ce7a6d
to
main
September 3, 2026 07:56
remyduthu
approved these changes
Sep 3, 2026
Contributor
Merge Queue Status
This pull request spent 2 minutes 59 seconds in the queue, including 1 minute 46 seconds running CI. Required conditions to merge
|
44 tasks
mergify
Bot
deleted the
devs/sileht/py313-py314-matrix/require-python-3-13-3-14-checks--7054c8de
branch
September 3, 2026 08:01
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
2 participants
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.
The two matrix legs added in the previous commit are advisory on their
own.
.mergify.yml's&CheckRunsanchor is this repository's only mergegate:
branches/main/protectionreturns norequired_status_checks, andnone of the eleven active rulesets carries one. So a red 3.13 or 3.14 leg
would land silently.
The ordering mirrors #299 -> #300 but not the reasoning, which is worth
stating rather than assuming symmetric. Removing a check needed its own
pull request first because Mergify reads its configuration from the
default branch: a pull request deleting the job would still have been
gated on a check that no longer reported, and could never merge. Adding
one has no such deadlock. Both pull requests here are evaluated against
main's configuration, which does not name the new checks, so a single
atomic commit would have merged just as well.
What the split actually buys is revertability. If a new leg turns out to
be red on main, reverting this commit alone unblocks the repository while
keeping the coverage the previous commit added; an atomic change would
have to give up both.
Two consequences worth knowing, neither of which ordering can avoid.
Because main's configuration does not require the new checks until this
lands, nothing forces either pull request of this stack to be green on
3.13 or 3.14. Both run the full five-leg matrix on their own head, so the
legs are observable, but they have to be looked at before this one is
merged rather than assumed.
Every pull request already open when this lands keeps the check-run set
from its last CI run, and no
pull_requestevent fires when the basemoves.
check-success=Test with Python 3.13can therefore never go trueon it, and
auto_merge_conditions: truefolds the success conditionsinto the queue conditions, so it cannot enter the queue to be rebased out
either. Each one needs a push, a rebase or a close/reopen to pick up the
new legs.
Longer term, this list is a hand-maintained mirror of the job names and
will want the same ceremony at 3.15. The monorepo and mergify-cli both
replaced it with one aggregate job (
all-greens/ci-gate) behind asingle
check-success, which is worth adopting here on its own. Notethat
check-success~=^Test with Pythonis not a shortcut for it:check-success is a list attribute and
~=matches any element, so onegreen leg would satisfy it while another fails.
Co-Authored-By: Claude Opus 5 (1M context) noreply@anthropic.com
Claude-Session: https://claude.ai/code/session_01MKmEnjRuAM4NcTYxRXG3BA
Depends-On: #310