Repository navigation
chore(release): 0.14.0 - #379
Conversation
Signed-off-by: client-software-ci <129794699+client-software-ci@users.noreply.github.com>
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
|
|
||
| ### Features | ||
| * Bump openjd-* Rust crates to the openjd-model 0.13.0 release (#378) ([`d9322c9`](https://github.com/OpenJobDescription/openjd-model-for-python/commit/d9322c91c90e40d27d0c3bbee0af531e83a67a7b)) | ||
| * Bump openjd-* Rust crates to the openjd-model 0.13.0 release ([`d9322c9`](https://github.com/OpenJobDescription/openjd-model-for-python/commit/d9322c91c90e40d27d0c3bbee0af531e83a67a7b)) |
There was a problem hiding this comment.
Nit: The 0.14.0 Features section lists the same change twice. Lines 5 and 6 both describe the Rust crate bump and link the same commit d9322c9; the only difference is the (#378) suffix. Most likely the generator picked up both the squash-merge title and the commit body. Delete line 6 and keep the entry that has the PR reference.
There was a problem hiding this comment.
Addressed: The duplicate Features entry was removed, and a single combined entry with the (#378) reference replaces both.
|
Claude review · advisory Reviewed Open: 0 blocking · 0 should-fix · 0 nit. ✅ Nothing blocking. Fix or reply to each thread; the next revision's review re-checks open threads and resolves those it agrees are handled. Resolving a thread yourself also closes it. Later revisions review only what changed. |
Signed-off-by: David Leong <116610336+leongdl@users.noreply.github.com>
|
The Windows test failures on this PR are a known flaky test, not a problem with the release. The fix is in #380, which I will merge after this release ships. What fails. The first tests in This run, same SHA
In attempt 1, windows 3.12 ran all 13 tests in the class and passed. When PowerShell starts in time, the output is correct. History. The same timeout caused failures in 8 of the last 60 Code Quality runs (2026-09-04 to 2026-10-15), on Python 3.10, 3.12, 3.13 and 3.14.
Run 35389550086 failed on attempt 1 and passed on attempt 2 at the same SHA, |
TestReprPwshWindowsValidation fails intermittently on windows-latest with subprocess.TimeoutExpired after 10 seconds. The first PowerShell launch on a fresh hosted runner pays a one-time cold start that can exceed the 10-second per-call timeout in _pwsh_eval, so whichever test launches PowerShell first is killed, and a killed launch can leave the next one cold too. Run 38812686088 (PR #379, a CHANGELOG-only change) lost the first four tests of the class this way, each at about 10.1 s, while the other nine passed. The same timeout appears in 8 of the last 60 Code Quality runs, on Python 3.10, 3.12, 3.13 and 3.14, and run 35389550086 passed on rerun at the same SHA. Add a class-scoped autouse fixture that pays the cold start once, with a 120-second deadline, through the same _pwsh_eval path the tests use: a .ps1 file run with -File that calls Write-Output. It asserts the output, so a broken PowerShell still fails the class instead of being skipped. The per-call timeout stays at 10 seconds and no test assertion changes. Mark the class timeout(180), because pytest-timeout counts class-scoped fixture setup against the first test and the project-wide --timeout=30 would otherwise cut a long warm-up short. Verified on macOS with PowerShell 7.6.7 behind a shim that simulates the cold start. The class is Windows-only, so this is a simulation, not a Windows reproduction. Without the fix, a 41 s cold start fails the same four tests as run 38812686088, at 10.01 s each, and a 12 s cold start fails test_pwsh_string_simple_roundtrip alone. With the fix both pass 13/13, and the cold start lands in the first test's setup. Each of these mutants fails the class under a 41 s cold start: no warm-up, a 10 s warm-up deadline, and no timeout marker. Only repeated green Code Quality runs on real Windows runners prove the fix. At roughly one failure in 50 Windows jobs, one green run is weak evidence. Signed-off-by: David Leong <116610336+leongdl@users.noreply.github.com>
3 participants
0.14.0 (2026-10-15)
BREAKING CHANGES
create_jobraisesModelValidationErrorinstead ofExpressionErrorfor a let-binding failure, some templates that decoded before are now rejected (and/orafter an unresolved operand, nested lists with mismatched element types, parameter-dependent invalid timeouts), and building from source needs Rust 1.97 (feat!: Bump openjd-* Rust crates to the openjd-model 0.13.0 release #378) (d9322c9)Features
d9322c9)Bug Fixes
d9322c9)