Skip to content

Arch/Omarchy package, and both packaging files on v0.2.0 - #3

Merged
tlehman merged 3 commits into
mainfrom
arch-packaging
Nov 11, 2026
Merged

tlehman merged 3 commits into
mainfrom
arch-packaging

Conversation

@tlehman

@tlehman tlehman commented Sep 2, 2026

Copy link
Copy Markdown
Owner

Adds packaging/arch/PKGBUILD next to the Homebrew formula, so dotbar installs on Arch — and therefore on Omarchy, which is Arch with the same pacman, so there is no second artifact to keep in sync.

Shape follows the omarchy-pkgs house style (ttfx): cargo fetch --locked in prepare() so build() and check() run --frozen, options=('!debug'), depends=('gcc-libs' 'glibc'). check() runs the suite against the release binary, since tests/cli.rs is process-level and has to exercise the binary package() installs. Both LICENSE texts ship: MIT OR Apache-2.0 is a choice, not a conjunction.

.SRCINFO is not in the tree on purpose — it is generated, only the AUR repo needs it, and a copy here would go stale on the first bump. The PKGBUILD header carries the publish and release-bump procedure.

The second commit moves the Homebrew formula to v0.2.0, which is now tagged, so both packaging files sit on the same tag.

Verified on an Omarchy 4.0.2 box: makepkg builds from the real v0.2.0 tarball, all 26 tests pass in check(), and the package contains usr/bin/dotbar plus both licenses and the README. The installed binary renders (NO_COLOR=1 dotbar --dense 76 → ⣿⡿⣀ 76%), exits 0 with no output on {}, and carries install-claude.

Omarchy is Arch (ID_LIKE=arch, same pacman), so one package covers both
and there is no separate Omarchy artifact to keep in sync. Shaped after
the omarchy-pkgs house style: cargo fetch --locked in prepare() so
build() and check() can run --frozen, !debug because nobody attaches a
debugger to a statusline segment, and both LICENSE texts because
MIT OR Apache-2.0 is the consumer's choice, not a conjunction.

check() runs the suite against the release binary — tests/cli.rs is
process-level, so it has to be the same binary package() installs.

.SRCINFO is deliberately not kept here: it is generated, only the AUR
repo needs it, and a copy in this tree would go stale on the first bump.
v0.2.0 is tagged now, so the formula and the PKGBUILD both point at it.
Same tarball, same sha256.
The README calls this file the source of truth for the tap, but the tap
had drifted ahead: 0533a08 there tightened the test block to assert the
exact 13-cell and dense renderings, the statusline stdin path, and the
unknown-subcommand exit, and that never came back here. Copying this
file over the tap — which is what the README tells you to do on a
release — would have quietly reverted it.

Below the header comment the two files are now byte-identical.
@tlehman
tlehman merged commit b61e0b0 into main Nov 11, 2026
2 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.

1 participant