8000 Fix benchmark publication audit gaps by rafatahmed · Pull Request #50 · rafatahmed/Flahax · GitHub
[go: up one dir, main page]

Skip to content
< 8000 a hidden="" id="code-view-repo-link" href="/rafatahmed/Flahax" data-discover="true">

Fix benchmark publication audit gaps - #50

Merged
rafatahmed merged 3 commits into
mainfrom
docs/benchmark-audit-fixes
Oct 20, 2026
Merged

rafatahmed merged 3 commits into
mainfrom
docs/benchmark-audit-fixes

Conversation

@rafatahmed
Copy link
Copy Markdown
Owner

Fixes the publication-quality gaps from the 2026-10-20 audit of #34, #35 and #43 (tracker #42).

Changes

The renderer (tools/publish_benchmark_report.py) now produces this text for future snapshots (new --capture-kind and --related-label options). The two published pages were updated with the renderer's own helpers. No evidence file, checksum or measured number changed.

Tests

  • New renderer tests: count wording, case labels, miss summary.
  • Published-page test: each page's miss summary matches its own snapshot CSV; no raw-ID rows; no "1 timing workloads".
  • Docs test: no P/G codes on public pages other than release evidence.
  • sphinx -W --keep-going builds cleanly with docs/site/requirements.txt.

#48 (stable docs version) stays open: it resolves when the next release tag includes the benchmark pages. No FlahaX link uses /en/stable/.

Closes #44, closes #45, closes #46, closes #47, closes #49.

Mark the Windows capture as a local supplement and state that passing
PR checks do not mean the scorecard passed (#44). Name the largest
target miss and list failing targets by nutrient (#45). Fix count
wording, show formula names beside catalogue IDs and explain the
memory-benchmark coverage (#46). The renderer produces the same text
for future snapshots; published evidence files are unchanged.
Repository-relative paths break on the PyPI project page, which renders
the README as the long description (#47).
API, model and assessment pages now use plain names instead of P/G
phase and gate codes. Release evidence keeps them because its records
cite them, and explains them once with a link to the roadmap. A test
keeps the codes off other public pages (#49).
@rafatahmed
rafatahmed merged commit a628be7 into main Oct 20, 2026
6 checks passed
@rafatahmed
rafatahmed deleted the docs/benchmark-audit-fixes branch October 20, 2026 12:52
@rafatahmed rafatahmed self-assigned this Oct 20, 2026
rafatahmed added a commit that referenced this pull request Nov 7, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
0