Repository navigation
docs(agents): document running targeted tests with BLS cgo flags - #1622
Conversation
A bare `go test` fails with `dashbls/bls.hpp: No such file or directory` because only Makefile targets pass the BLS cgo flags. Document the flags, worktree/macOS/CI hints, and correct the make target descriptions. CLAUDE.md now imports AGENTS.md with `@AGENTS.md`; a plain-text reference does not load the file into Claude Code's context. <sub>🤖 Co-authored by [Claudius the Magnificent](https://github.com/lklimek/claudius) AI Agent</sub> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Important Review skippedAuto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the
|
|
🕓 Review not started yet because the new head is waiting for the 30-minute push debounce.
Commit 3bdab33. Normal review starts when eligible; priority review starts as soon as a slot is available. |
|
Waiting for bot review — thepastaclaw not yet. Wait for the missing reviews, or a writer can post |
TL;DR: Contributor and AI-agent instructions now explain how to run a single test package without a native-library compile error, and Claude Code actually loads those instructions.
User story
As a contributor or AI coding agent, I want to run just the tests I touched, to get fast feedback without running the whole suite through
make.Scenario
Base flow
A contributor changes one package and runs its tests directly with
go test.Actual behavior
The build fails with
blschia.cpp:17:10: fatal error: dashbls/bls.hpp: No such file or directory, and the docs do not say why. Claude Code also does not loadAGENTS.md:CLAUDE.mdonly mentions it in prose.Expected behavior
The docs give the exact environment needed to run
go teston one package, and Claude Code sessions start withAGENTS.mdin context.Detailed discussion
What was done
AGENTS.md:CGO_ENABLED/CGO_CXXFLAGS/CGO_LDFLAGSexports, copied from the Makefile.build/exists only wheremake build-blsran (fresh clones and git worktrees need it, orBLS_DIRpointing at a checkout that built it); macOS gmp flags; how CI runs tests;libpcap-devforcmd/abcidump;-count=1.make build-blswith its prerequisites, documentsLINT_BASEfor stacked PRs, and corrects the target descriptions.make testuses-tags deadlockwithout-race,make test_raceuses-racewithout the tag, and neither is how CI runs tests.CLAUDE.md: replaces the prose reference with an@AGENTS.mdimport. Per the Claude Code memory docs, aCLAUDE.mdthat only tells Claude to readAGENTS.mddoes not load it, and aCLAUDE.mdin the repo stops Claude Code from readingAGENTS.mdon its own.Testing
go test -count=1 ./crypto/bls12381/: passes. The same flags withBLS_DIRpointing at another checkout letgo test -race ./internal/p2p/pass in a git worktree that had no BLS build.markdownlintwith.github/linters/markdownlint.yml: no new findings. The 4 MD007 findings on the existing Protobufs list predate this PR.Breaking changes
None.
🤖 Co-authored by Claudius the Magnificent AI Agent
🤖 Generated with Claude Code
PR Hygiene ·
3bdab33/skip-botsproceeds without the ones not yet reportedAGENTS.md,CLAUDE.md) — lklimek or shumkovWhen every merge requirement is met, the
PR Hygienecheck passes. Reviewer limits do not block merging; other required GitHub checks and protections still apply.