Skip to content

cut: v0.42.0 - #1052

Merged
InauguralPhysicist merged 1 commit into
mainfrom
cut-v0.42.0
Aug 27, 2026
Merged

cut: v0.42.0#1052
InauguralPhysicist merged 1 commit into
mainfrom
cut-v0.42.0

Conversation

@InauguralPhysicist

Copy link
Copy Markdown
Collaborator

Cuts v0.42.0. Updates the four sites the v0.41.0 cut touched: VERSION, CLAUDE.md's "Latest release" line, the CHANGELOG.md section, and docs/llms.txt's version stamp.

Contents since v0.41.0

Why now

The tag is the version pin for consumers. phugoid's devcontainer clones EigenScript at a released tag; lib/complex.eigs does not exist in v0.41.0, so phugoid's modes.eigs stdlib swap — 45 lines of hand-rolled complex arithmetic and root-finding deleted, chain eigenvalues verified bit-identical — is blocked until this tag exists.

On the apparent doc-drift contradiction

CLAUDE.md now names a tag that does not exist yet. That is the documented ordering, not an oversight: doc-drift rule 2 compares that line to the latest tag, and the release build runs after the tag is created, so a stale line fails the release's own suite (it bit the v0.27.0 cut).

It does not fail this PR's CI because only the first ci.yml job checks out with fetch-depth: 0; the suite job's default checkout fetches no tags, so the rule reports "no tags; skipping" there. Verified against the v0.41.0 cut (#1037), which passed 18/18 in exactly this state.

Verification

Rebuilt and re-ran tests/test_cli.sh, because --version is baked in at compile time from VERSION — a bump without a rebuild makes the binary disagree with the file. 15/15, and --version prints 0.42.0.

After merge: dispatch Actions → Release → Run workflow (this environment's git proxy cannot push tags, and a GITHUB_TOKEN-pushed tag would not retrigger the workflow).

🤖 Generated with Claude Code

https://claude.ai/code/session_01Aj9b82JBb8WS3b8ExoV5Yt

VERSION, CLAUDE.md's "Latest release" line, the CHANGELOG section and
docs/llms.txt's stamp — the four sites the v0.41.0 cut touched.

Contents since v0.41.0: the stdlib complex/eigenvalues work (#1051,
closing #1042 and #1043) — lib/complex.eigs, linalg.charpoly and
linalg.eigenvalues, engineering.phase_spectrum with magnitude/power_spectrum
delegating to complex.mag/mag2 — plus the github-actions bump (#1050).

Cutting now because the tag IS the version pin for consumers: phugoid's
devcontainer clones EigenScript at a released tag, lib/complex.eigs does
not exist in v0.41.0, and its modes.eigs stdlib swap (45 lines of
hand-rolled complex arithmetic and root-finding deleted, chain eigenvalues
verified bit-identical) is blocked until this tag exists.

CLAUDE.md names a tag that does not exist yet, which is the documented
ordering: doc-drift rule 2 compares that line to the latest tag, and the
RELEASE build runs after the tag is created, so a stale line fails the
release's own suite (it bit the v0.27.0 cut). It does not fail this PR's
CI because only the first ci.yml job checks out with fetch-depth: 0 — the
suite job's default checkout fetches no tags, so the rule reports "no tags;
skipping" there. Verified against the v0.41.0 cut, which passed 18/18 in
the same state.

Rebuilt and re-ran tests/test_cli.sh because --version is baked in at
compile time from VERSION: 15/15, --version prints 0.42.0.
@InauguralPhysicist
InauguralPhysicist merged commit feb99d6 into main Aug 27, 2026
18 checks passed
@InauguralPhysicist
InauguralPhysicist deleted the cut-v0.42.0 branch August 27, 2026 22:31
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