docs: restore the gate facts dropped from CLAUDE.md - #145
Open
prorochestvo wants to merge 2 commits into
Open
Conversation
Four references survived only on docs/slim-claude-md, whose pull request (#140) was closed rather than merged. PR #142 salvaged that branch's canon into the project skills but did not carry these, and the branch has now been deleted, so this brings them back into the file. The load-bearing one is that both make lint and make lint-new run two steps, golangci-lint run and scripts/lint-checks.sh. A green golangci-lint is therefore not a green gate, and nothing outside the Makefile said so. make lint-new is also the mergeable gate, linting only what changed since LINT_BASE, while make lint scans the whole tree as a worklist; CLAUDE.md named only the latter, so the documented command was the one that reports standing findings on untouched code. Also recorded: make test runs the WASM suite, and skips it with a warning and a zero exit when node is absent, which takes the gate green with that coverage silently missing. On the 8 GB Pi the modernc.org/sqlite rebuild is OOM-killed under -race, where go test -race -p 1 is the way through. Both were verified against the Makefile; the Pi constraint is corroborated by deploy/README.md and the completed long-range-outlook plan. The remaining two are pointers to the canon this file is a delta of: standards-layout R21 for the measure-and-account procedure, and pipeline:working-agreement for the pipeline the working agreement below restates. CLAUDE.md grows from 17234 to 17890 bytes, inside its own 20k budget. Note for the reviewer: make lint-new cannot run on the dev Mac right now. Go 1.27.0 is installed while golangci-lint 2.12.2 is built against go1.26.2, so gocritic's ruleguard panics decoding export data version 4. The failure reproduces on a clean tree with this change stashed and is unrelated to it. scripts/lint-checks.sh, the half that inspects text, was run and is clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Nz77JABRj78e92q1sRPuyF
Claude Code matches file-tool permission rules on Edit(path) only, so every Write(path) entry is dead config that emits a startup warning on each session. Every one already has an Edit() twin, so the .env and .idea guards are unchanged; only the noise goes.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Restores four references into
CLAUDE.mdthat survived only ondocs/slim-claude-md, the branch behind PR #140. That pull request was closed rather than merged; PR #142 salvaged its canon into the project skills but did not carry these. The branch has now been deleted as part of the repository cleanup, so this is the last chance to keep them.CLAUDE.md: 17234 → 17890 bytes, inside its own 20k budget.Why it matters
The load-bearing item is a silent failure, which is exactly what
CLAUDE.mdis chartered to hold:make lintandmake lint-newrun two steps —golangci-lint runandscripts/lint-checks.sh. A greengolangci-lintis not a green gate, and nothing outside the Makefile said so.make lint-newis the mergeable gate, linting only what changed sinceLINT_BASE(origin/alpha);make lintscans the whole tree as a worklist.CLAUDE.mdnamed only the latter, so the one documented command is the one that reports standing findings on untouched code.make testruns the WASM suite, and skips it with a warning and a zero exit whennodeis absent — the gate goes green with that coverage quietly missing.modernc.org/sqliterebuild under-race;go test -race -p 1is the way through, or ans_*tag hands the gate to CI.The remaining two are pointers to the canon this file is a delta of:
standards-layoutR21 for the measure-and-account procedure, andpipeline:working-agreementfor the pipeline the working agreement restates.Verification
Every claim was checked against the source, not copied on trust:
lint-newis the mergeable gateMakefile:246target commentMakefile:238-253LINT_BASE ?= origin/alphaMakefile:40make testruns the WASM suite, exits 0 withoutnodetest:targetdeploy/README.md:11,plans/completed/260823.0001.long-range-weather-outlook.md:302make lint(pre-existing claim).golangci.yml:107-114,depguardscripts/lint-checks.sh origin/alpha— clean, exit 0.Known issue, unrelated to this change
make lint-newcannot complete on the dev Mac right now. Go 1.27.0 is installed whilegolangci-lint2.12.2 is built against go1.26.2, sogocritic's bundled ruleguard panics decodingnet/http:The failure reproduces on a clean tree with this change stashed, so it predates it. The fix is a
GOLANGCI_VERSIONbump to a build made against Go 1.27 — worth its own issue rather than riding along in a docs change.🤖 Generated with Claude Code
https://claude.ai/code/session_01Nz77JABRj78e92q1sRPuyF