Skip to content

Check out the full history in CI so the recorded checksum check runs - #113

Merged
scottjones merged 1 commit into
omacom:mainfrom
only21mil:ci-full-history-cryptsetup
Nov 9, 2026
Merged

scottjones merged 1 commit into
omacom:mainfrom
only21mil:ci-full-history-cryptsetup

Conversation

@only21mil

Copy link
Copy Markdown
Contributor

Problem

tools/release/test/omacom-lane checks that four real omarchy-mac-pkgs commits hash to the checksums makepkg recorded for them in omacom/omarchy-pkgs. It needs those commits in the checkout, and skips without them. The suites job checks out omarchy-mac-pkgs at actions/checkout's default depth of 1, so CI always skips it and prints ok. Main's latest run (42426384788, at 24430b4) shows:

ok - # skip: this checkout lacks the omarchy-mac-pkgs history the recorded checksums name

A change that broke how releases compute a git source's checksum would merge with green CI. AGENTS.md: "a skipped test, or a job that failed before reaching the tests, is not a pass."

Change

  • .github/workflows/tests.yml: fetch-depth: 0 on the suites job's omarchy-mac-pkgs checkout, so the four commits are present and the check runs.
  • tools/release/test/omacom-lane: the comment no longer says a shallow CI checkout lacks the history.

The repository is about 2 MiB packed, so the full checkout costs little. A fixed depth would break again as main moves on: the oldest of the four commits is already about 200 commits back.

Scope and approval

CI only. No package behaviour changes, so omarchy-mac/version stays as it is, and nothing here needs cold-boot evidence or a manual change.

The audit that found this also suggested installing cryptsetup so test/integration/factory-reset-luks-test.sh runs its real LUKS2 volume case. It is not needed: the Arch image already has cryptsetup, and main's runs already print ok - on a real LUKS2 volume with a token, ....

Verification

The two runs below are the real test in a depth-1 clone of 24430b4 (what CI checks out today) and in a full clone of this branch.

Before: depth-1 clone of main, the checksum check skips and the suite exits 0.

omacom-lane in a depth-1 clone prints the skip line

After: full clone of this branch, the checksum check runs and passes.

omacom-lane in a full clone prints the checksum line

  • actionlint .github/workflows/tests.yml: exit 0 on main and on this branch.
  • Depth-1 clone of 24430b4: bash tools/release/test/omacom-lane exits 0, with the skip line.
  • Full clone of this branch: the same command exits 0 and prints ok - omarchy-mac-pkgs commits hash to the checksums makepkg recorded in omarchy-pkgs.
  • With one recorded checksum changed: the full clone exits 1 with not ok - the checksum of omarchy-mac-pkgs 5b2c815f007d is not the 00000000e396 makepkg recorded, while the depth-1 clone still exits 0 on the skip.
  • No other test reads the repository's own history.

Not checked

  • The CI run on this pull request. It should print ok - omarchy-mac-pkgs commits hash to the checksums makepkg recorded in omarchy-pkgs under "Tooling suites", where main prints the skip.
  • The skip still prints ok, so a shallow checkout would hide the check again. Failing instead of skipping under CI would be a separate change.

Prior PRs

#27 added this check and its skip for shallow checkouts. #13, #30 and #76 also edit tests.yml; #30 adds fetch-depth: 2 to the separate scope job, so the two do not conflict.

Generated with Claude Opus 5.5 and GPT-6.1 Sol in T3 Code Nightly

The suites job checked out omarchy-mac-pkgs at depth 1, so
tools/release/test/omacom-lane skipped its check that real
omarchy-mac-pkgs commits hash to the checksums makepkg recorded in
omarchy-pkgs, and printed ok for it. Fetch the full history so the
check runs.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

@scottjones scottjones left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks, small and exactly right. It's a one-line change with no new permissions, secrets or actions. With full history the checksum check actually runs and passes, where before it skipped with "ok - # skip". The package resolution failure here comes from the live repo databases, not from this PR.

@only21mil

Copy link
Copy Markdown
Contributor Author

Note

🤖 Opus 5.5 replying on behalf of Victor.

@scottjones, thank you for the review. Checking that it adds no permissions, secrets or actions is the right bar for a workflow change, and thanks for confirming in this PR's own run that the checksum check now runs and passes, and that the package resolution failure comes from the live databases.

@scottjones
scottjones merged commit bee4c4e into omacom:main Nov 9, 2026
3 of 4 checks passed
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.

2 participants