Skip to content

chore: cut v4.8.0rc1 - #347

Merged
danielplohmann merged 1 commit into
masterfrom
chore/prerelease-v4.8.0rc1
Sep 14, 2026
Merged

danielplohmann merged 1 commit into
masterfrom
chore/prerelease-v4.8.0rc1

Conversation

@danielplohmann

Copy link
Copy Markdown
Owner

The release candidate that rehearses #343's workflow. No code changes: __version__ moves to 4.8.0rc1 and ## [Unreleased] becomes ## [v4.8.0rc1] - 2026-09-14.

This exists to run the release path once before a real version number is at stake. Two defects have already come out of preparing it — #345 and #346 — and neither was reachable without trying to cut a candidate.

What it carries

Five entries, each written by its author at merge time:

Added — the release-on-a-tag workflow (#343).
Changed — SmdaConfig.VERSION read from smda.__version__, and the PEP 639 licence expression with the project URLs (#343).
Deprecated — the 3.11 support policy: the floor follows MCRIT's and moves in the same wave (#343).
Fixed — a pre-release version is now ordered on report import rather than treated as no version at all (#345).

Removed, Security and Compatibility were empty and are dropped. A fresh ## [Unreleased] sits above with the full scaffold. None of the three compatibility constants moved, so no Compatibility section is owed.

Why a candidate rather than going straight to v4.8.0

Pushing any v* tag runs the real path to PyPI, and a manual TestPyPI run needs the tag to already exist on the remote — so pushing is both what makes a rehearsal possible and what would publish it. The required reviewer on the pypi environment is what separates the two: the push run stops at the publish job for approval, that gets rejected, and the same tag is then dispatched manually to TestPyPI.

A version number on PyPI is permanent even when yanked. Spending 4.8.0rc1 to find out whether the workflow holds is the cheap way to learn it.

Verification

On this branch, saturn:

  • python -m pytest tests/ → 2,172 passed, 1 skipped, 2,623 subtests
  • ruff check . and ruff format --check . → clean
  • release_guard.py --tag v4.8.0rc1 → passes, version=4.8.0rc1 prerelease=true, so the GitHub release for it would be marked a pre-release and never become latest
  • compare links: [Unreleased] spans v4.8.0rc1...HEAD, [v4.8.0rc1] spans v4.7.0...v4.8.0rc1

After the rehearsal

The final cut renames ## [v4.8.0rc1] to ## [v4.8.0] - <date> rather than adding a second section repeating it — the candidate's own release notes were generated from the tree at its tag, so nothing is lost by the heading moving on. RELEASING.md does not currently say which of those two it wants; worth settling there once the rehearsal has shown what the notes actually look like.

🤖 Generated with Claude Code

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown

📊 SMDA Performance Evaluation Benchmark Results

Generated on: 2026-09-14 13:53:29

Summary

Metric Result
Correctness ✅ PASS — 0 / 155 common file(s) differ
Determinism ✅ PASS — base 3 run(s), PR 3 run(s)
Verdict inconclusive — within run-to-run noise (±8.1%)
Median paired speedup -0.06% (95% CI [-0.76%, +0.60%])
Timing noise band ±8.1%

Per-side Timing Context (best of 3/3 runs per file)

Side Files Functions Median best time/file (s) Sum of best times (s) Throughput estimate (func/s)
base 155 173496 0.6573 254.75 ~681
pr 155 173496 0.7049 254.16 ~683

These context rows sum each file's best observed time across repeated runs; they are normalized comparison estimates, not single-run CI wall-clock times.

Paired per-file timing (positive speedup = PR faster):

Statistic Value
Files compared 155
Median paired speedup -0.06% (95% CI [-0.76%, +0.60%])
Mean speedup -2.96% (95% CI [-6.50%, +0.46%])
Std dev / IQR 22.16% / 8.50%
Wilcoxon signed-rank p 0.9822 (n=155)

ℹ️ base and PR are timed in the same job on the same runner, with their passes interleaved base/PR/PR/base/base/PR so that drift over the job cancels rather than landing on whichever side ran last. The noise band above is therefore run-to-run variation on one machine, which is what it is able to measure; a difference inside it is reported as inconclusive. Correctness and determinism are not timing-based and are unaffected either way.

Determinism (self-check across repeated runs)

Side Runs Files Deterministic Median timing CV
base 3 155 ✅ 8.1%
pr 3 155 ✅ 6.8%
Pairwise run matrix (individual run medians, diagnostic)

Diagnostic only: each row compares one raw PR run against one raw base run. The headline verdict above uses paired per-file best-of-runs timings.

Comparison Run PR Files Base Files Common Function Set Matches Med PR (s) Med Base (s) Med Diff (s) Speedup %
pr_0 vs base_0 155 155 155 155/155 0.8762s 0.9720s -0.0958s +9.86%
pr_0 vs base_1 155 155 155 155/155 0.8762s 0.9293s -0.0531s +5.71%
pr_0 vs base_2 155 155 155 155/155 0.8762s 0.8509s +0.0253s -2.97%
pr_1 vs base_0 155 155 155 155/155 0.9121s 0.9720s -0.0599s +6.16%
pr_1 vs base_1 155 155 155 155/155 0.9121s 0.9293s -0.0172s +1.85%
pr_1 vs base_2 155 155 155 155/155 0.9121s 0.8509s +0.0612s -7.19%
pr_2 vs base_0 155 155 155 155/155 0.9128s 0.9720s -0.0592s +6.09%
pr_2 vs base_1 155 155 155 155/155 0.9128s 0.9293s -0.0165s +1.78%
pr_2 vs base_2 155 155 155 155/155 0.9128s 0.8509s +0.0619s -7.27%

@danielplohmann
danielplohmann merged commit 5479aae into master Sep 14, 2026
47 checks passed
@danielplohmann
danielplohmann deleted the chore/prerelease-v4.8.0rc1 branch September 14, 2026 13:58
@danielplohmann danielplohmann added this to the v4.8.0 milestone Sep 14, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant