lean: put elan's bin on PATH, so a fresh checkout finds lake - #38
Merged
Conversation
`make test` in the lean port failed with "lake: not found" on a machine whose Lean came from elan. elan installs to ~/.elan/bin and a stock install does not put that on a non-login shell's PATH, so make's own shell could not see it. CI never hit this: it unpacks a pinned lean release to /opt/lean/bin and puts that on GITHUB_PATH itself, so only a local checkout was affected. Prepending a directory that does not exist is a no-op, which is why CI stays as it was. This is the line sekreto's lean port already carries, for the same reason. Verified by running `make test` with a PATH that has no elan on it: 18 modules compile, 572 corpus entries pass. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SD5pZ6WjP8nMiUcAkRZjNN
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
rjrodger
pushed a commit
to voxgig/sekreto
that referenced
this pull request
Sep 5, 2026
main gained a house style guide and a two-part prose gate (vale + tools/check_prose.py) while this branch was converting ports. Merging it is also what unblocks CI: GitHub cannot build a merge ref for a conflicted PR, so no pull_request workflow had run on this branch all day, while voxgig/plugin#38 - opened from the same session minutes earlier and mergeable - got all 28 checks at once. Four conflicts, all README.md, all the same shape: main reworded the PRE-adoption text for the gate, this branch rewrote the same paragraphs to describe the adopted architecture. Resolved to this branch, because those ports have adopted and main's rows say "switch - pending" for sixteen that have moved. Style was then settled by running the gate rather than by guessing at main's intent. That turned up twelve findings, none of them from the merge itself: the adoption agents wrote these READMEs before the gate existed on main, so none of them could have known the rules. - six ports cited docs/design/plugin-providers.md, and one cited AGENTS.md. Documentation does not cite a working document; the citations are dropped, and each sentence still stands on its own. - c used "ours" and "our flags" for the port's own flags. - banned phrases: "sit with" (cpp), "load-bearing" and "quietly" (swift). One of my own fixes was wrong and is corrected here rather than shipped: rewriting c/README.md's "under our flags would weaken" dropped the verb and left "under this port's flags would / the gate to suit the dependency". It now reads "under these flags would weaken", which also stops the sentence saying "this port" twice. Verified after the merge: prose gate ok over 25 pages; omni register clean at 457 files; omni_isolation_selftest 85/85; make test-python green, since main moved `test:` to depend on scan-prose and that path needed exercising. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SD5pZ6WjP8nMiUcAkRZjNN
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
2 participants
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.
make testin the lean port fails withlake: not foundon a machinewhose Lean came from elan. elan installs to
~/.elan/bin, and a stockinstall does not put that on a non-login shell's
PATH— so the shellmakespawns for each recipe cannot see it.CI never hit this. It does not go through elan at all: it unpacks a
pinned release to
/opt/lean/binand puts that onGITHUB_PATHitself(
.github/workflows/ci.yml:235-242), so only a local checkout wasaffected. Prepending a directory that does not exist is a no-op, which
is why CI is unchanged by this.
The change
Two lines in
lean/Makefile, and the comment explaining why they arethere. They are the same two lines sekreto's lean port already carries,
for the same reason —
makeruns each recipe in its own shell, soexporting
PATHfrom the makefile is the only place all of them see it.Verification
Run with a
PATHthat has no elan on it — the condition that producedthe failure:
All 572 corpus entries pass. No other port needed this — the other
twenty-two build under the session's default
PATHas they did before.🤖 Generated with Claude Code
https://claude.ai/code/session_01SD5pZ6WjP8nMiUcAkRZjNN
Generated by Claude Code