stillwater-sc / universal / 36208611327
87%
master: 84%

LAST BUILD BRANCH: main
DEFAULT BRANCH: master
Ran
Jobs 1
Files 676
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 85.975% (-0.02%) from 85.992%
36208611327

push

github

web-flow 
chore: prepare the v5.1.0 release (#1611)

* chore: prepare the v5.1.0 release

Fallback version 5.0.0 -> 5.1.0. It only applies when the version cannot be read
from a git tag, which is the tarball and git-less build case, so a source archive
of the v5.1.0 tree would otherwise report 5.0.0.

A minor bump rather than a patch, deliberately. There are no feat commits, so
conventional-commit tooling would compute 5.0.1, but a patch release says "take
this without reading" and these nine commits change values a caller reads:
single's digits10, duble's min_exponent10, bfloat16's digits, and xtndd's entire
dynamic range. Four BEHAVIOUR-CHANGE trailers carry the detail into the notes.

CHANGELOG gets the [5.1.0] section, covering the seven numeric_limits issues, the
CodeQL removal, and the documentation and CHANGELOG-attribution work that
preceded them.

Session log for 2026-09-24 to 2026-09-25. Its subject is mostly what went wrong:
every defect in the session's own work came from verifying the code path that
runs rather than the path that exists, and the test suites were green through all
of them. Also records the three measurement artifacts reported as results, the
same guard mistake made three times, and the two wrong hypotheses that preceded
finding the CI slowdown was a spend budget rather than the eviction limit.

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

* docs: describe the session's work rather than asserting a release state

Review caught that the index entry said v5.1.0 was released in a commit that
predates the v5.1.0 tag. It reads "cut v5.0.0 and v5.1.0" now, which describes
what the session did and is true whenever it is read.

Small, but it is the same failure as claiming a fix before pushing it, which
happened twice this session. Worth not shipping a third instance in the document
that records the other two.

Relates to #1611

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

---------

Co-authored-... (continued)

45505 of 52928 relevant lines covered (85.98%)

9401920.95 hits per line

Coverage Regressions

Lines Coverage ∆ File
9
95.35
-0.58% include/sw/universal/number/cfloat/cfloat_impl.hpp
Jobs
ID Job ID Ran Files Coverage
1 36208611327.1 676
85.98
GitHub Action Run
Source Files on build 36208611327