Skip to content

test(installer): expose mv to the offline harness PATH, un-breaks quick on main (fixes #3945) - #3946

Closed
Yumihariii wants to merge 1 commit into
limitlessprojects:mainfrom
Yumihariii:fix/install-harness-mv-3945
Closed

Yumihariii wants to merge 1 commit into
limitlessprojects:mainfrom
Yumihariii:fix/install-harness-mv-3945

Conversation

@Yumihariii

Copy link
Copy Markdown
Contributor

Fixes #3945.

One line: add mv to required_tools in tests/test_install_runtime.py so the scrubbed offline harness PATH exposes the tool install.sh gained in #3788 (mv-based backup/rollback at lines 112/115/192/199/205/210). Every harness-side analogue already exists (cp, rm, chmod are exposed the same way; test_install_checksums.py:183 already stubs mv for the same reason).

Evidence, run locally on a clean worktree at cc2de7a7:

  • RED before: python3 tests/test_install_runtime.py → FAILED (failures=27), signature install.sh: 199: mv: not found — byte-identical to the current main quick failure (job 128253107961) and to every open fork PR's red quick.
  • GREEN after: test_install_runtime.py → OK (14/14).
  • No collateral: test_install_checksums.py / test_install_upgrade.py / test_install_release_tag.py → OK before and after; tools suite Ran 533 tests — OK; check_test_registration.py OK (39 files / 35 patterns, no new files so no registration change needed); check_claims_links.py OK; make -n examples hygiene OK.

Installer code is untouched — this only teaches the offline harness about a tool the shipped installer already requires. Once landed, quick goes green repo-wide again (main + all fork PRs including mine).

Fixes limitlessprojects#3945. limitlessprojects#3788 gave install.sh a mv-based backup/rollback step, but
test_install_runtime.py runs the installer with a scrubbed PATH that only
exposes required_tools -- mv was never added, so every test that
pre-creates an existing installation dies on 'mv: not found' (27 failures
on main since 8b91452, 14 consecutive red main runs).

Harness-side one-liner; installer semantics untouched. Verified at
cc2de7a: test_install_runtime OK (14/14, was FAILED failures=27), other
three installer files OK, tools suite 533/533, registration guard 39/35,
claims links OK, make -n hygiene OK.
@limitlessprojects

Copy link
Copy Markdown
Owner

Duplicate — already fixed on main (558435fc, "fix installer runtime fixture missing mv command", landed before your PR was reviewed).

Verified on current main:

Your diagnosis was correct and your one-liner is the same fix — independent race, first to land wins. The patch-id collision is recorded here as prior art; keeping this PR open per no-silent-close doctrine rather than closing it.

Hall-of-fame note: mv entered install.sh via #3788's backup/rollback path (lines 112/115/192/199/205/210), and the harness-side gap was real until 558435fc — you and that commit fixed the same hole from two directions within hours. Nice convergence.

Bounty note: per BOUNTY.md and discussion #295, bounty claims attach to the landed commit; if you believe your independent implementation predates 558435fc (push timestamps), say so there and it'll be reviewed.

@Yumihariii

Copy link
Copy Markdown
Contributor Author

Thanks for the careful check and the hall-of-fame writeup — glad the diagnosis was right. One timing clarification on the prior-art question, since you asked for push timestamps:

558435fc carries a fork-local committer date of 2026-11-06 22:35:30Z (-0800), but it did not reach main until #3925's merge (c6270b7310, 2026-11-07 12:18:07Z) — verifiable with git merge-base --is-ancestor 558435fc cc2de7a7 → false. When we pushed our fix (02:21:47Z, PR up at 02:22:03Z), main was at cc2de7a7 with quick red across six consecutive pushes since 8b914525; tests/test_install_runtime.py failed 14/14 on that tip and our one-liner took it to 14/14 OK. So ours was the only fix proposed while main was red, and it predates the landing by ~10h — recording it per your invitation, happy to leave the credit call to you.

Also a small FYI: #3945 is still open on GitHub — the merge didn't auto-close it since the commit message doesn't carry a fixes keyword.

@github-actions github-actions Bot added bot:conflict Shepherd: merge conflicts vs main bot:stale-base Shepherd: behind main beyond threshold labels Nov 8, 2026
@Yumihariii

Copy link
Copy Markdown
Contributor Author

Closing this as superseded. main already carries the same fix: commit 558435f ("fix installer runtime fixture missing mv command", merged via #3925 on 2026-11-07T12:18Z) adds mv to required_tools in tests/test_install_runtime.py — semantically identical to our one-liner, just placed elsewhere in the tuple. Verified against origin/main@94b0e143 (line 47 contains "mv"). quick passes on current main, so the #3945 failure mode is resolved upstream; leaving the issue itself for a maintainer to close.

@Yumihariii

Copy link
Copy Markdown
Contributor Author

superseded by 558435f via #3925 — see note above

@Yumihariii Yumihariii closed this Nov 9, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bot:conflict Shepherd: merge conflicts vs main bot:stale-base Shepherd: behind main beyond threshold

Projects

None yet

Development

Successfully merging this pull request may close these issues.

CI quick red on main since 8b914525: test_install_runtime.py offline harness PATH lacks mv (27 failures)

2 participants