omacom/omarchy#13362 sets HOOKS per platform inside 00-omarchy-hooks.conf,
with no column-0 HOOKS= line, so the aarch64 build failed on it; settings-dev
tracks quattro automatically and would break as soon as it lands. Wrap only a
single-line HOOKS=(...), refuse any other column-0 HOOKS=, then source the
shipped files onto a Mac line and a stock line: the Mac line must come through
unchanged and the stock one must not. Back up 00-omarchy-hooks.conf when it
ships, and refuse it without the omarchy-hw-platform copy it asks.
The test now uses the real v4.0.4 hooks file, #13362's two files and
omarchy-mac-boot's 90-94 fragments, and covers upgrades over a hand-restored
file.
Apple Silicon Macs install the aarch64 package too, and an unconditional
HOOKS= line would replace their asahi hooks. Apply Omarchy's HOOKS line only
when the incoming hooks lack asahi, in whichever drop-in carries it, and
rebuild omarchy-settings (4.0.4-3) so existing installs get the fix.
CI builds a changed recipe and needs a new package; 4.0.4-1 is already
published. The pinned upstream source has no guard files, so 4.0.4-2 has the
same contents.
Moving omarchy-hw-platform from omarchy to omarchy-settings loses the file
when the dev pair installs in separate transactions, since omarchy-dev does
not pin omarchy-settings-dev's version. Settings now installs a copy beside
the guard script and omarchy keeps its detector, so no file changes owner.
The settings package's platform guard runs omarchy-hw-platform before the
omarchy runtime is installed. omarchy leaves the detector out of its bin/
loop and settings ships it, with the /usr/share/omarchy/bin link the Apple
predicate and the CLI router use, whenever the source has it.
The guard hook and the script it runs install together from omarchy-settings,
so they are resident before the runtime and hardware packages. Conditional, so
builds from sources without them are unchanged.
Exercise both settings recipes with synthetic runtime files. Check the exact Limine templates, boot drop-ins and backup metadata, installer-owned configuration, notifier copies and architecture-specific Thunderbolt handling. Run package regressions in the existing self-test job.
Preserve the v4.0.4 source pin and checksum while retaining the settings package revision bump. Keep Jim Martin’s boot-configuration fix and original commit intact.
basecamp/omarchy#11653 added etc/udev/rules.d/60-omarchy-io-scheduler.rules. package() already ships the whole etc/ tree, so the rule lands on new installs with the next build; list it in backup= so pacman keeps a user's local edit of the rule across upgrades, like the other package-owned /etc drop-ins.
Claude-Session: https://claude.ai/code/session_01EqYNBv5cERVom2zox5epyE
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
The aarch64 packages removed etc/mkinitcpio.conf.d and
etc/limine-entry-tool.d wholesale. On an encrypted aarch64 install that boots
with Limine (DGX Spark, Snapdragon X), installing the published package
therefore drops omarchy_hooks.conf, HOOKS falls back to /etc/mkinitcpio.conf,
and the next kernel update builds an initramfs with no encrypt hook: the
machine hangs waiting for /dev/mapper/root with no prompt.
Ship both drop-in directories on aarch64 and remove only
thunderbolt_module.conf, whose module ARM kernels do not have. Keep the
Limine template and the snapper notifier autostart for the same reason. The
memory-stack removals (zram, oomd, zswap, USB autosuspend) are unchanged.
The Omarchy payload is architecture-independent, but the package is not.
On x86_64 omarchy pulls the Limine + mkinitcpio hook + Snapper boot stack;
on Apple Silicon the system boots through m1n1 + GRUB from the Asahi
packages on Arch Linux ARM's kernel, so that stack does not apply, and
Wi-Fi on the Broadcom parts needs the iwd backend. The shipped /etc tree
differs too: mkinitcpio reads every file under /etc/mkinitcpio.conf.d/,
so shipping omarchy_hooks.conf on aarch64 injects the Limine hooks into
the Asahi kernel's initramfs, and the zram/zswap/oomd drop-ins and the
zram-tuned vm.* sysctls belong to the x86_64 memory stack.
makepkg only honours depends_<arch> and optdepends_<arch> on
arch-specific packages, so arch=('any') becomes ('x86_64' 'aarch64').
backup=() has no arch-suffixed form, so the x86_64-only entries are
appended under CARCH and each of those paths is removed from the aarch64
package in package(). The x86_64 package keeps exactly the contents it
had; only its filename suffix changes.
The install scriptlet applies the hardened cups-files.conf on every
platform, then on Apple Silicon keeps the Arch Linux ARM system identity
(/etc/os-release stays the distribution's symlink) instead of the
etc-overrides. The -dev pair carries the same change so the pairs stay in
lockstep.
Edge is the channel the 4.0.2 RC ISOs point pacman at until the rc-channel
migration lands, but edge only carried omarchy 4.0.1 — older than the
installed 4.0.2rc1, so those installs never saw an upgrade. Bumping master's
pin lets the normal edge pipeline build and publish 4.0.2rc2, whose migration
repoints [omarchy] at the rc channel.
Stable is unaffected: pinned packages never build for stable
(package_builds_for_mirror), it only receives them via the rc->stable advance
at ship time.
(cherry picked from rc-branch pin commit 11767d6)
The rc channel's Arch base can sit anywhere between stable's snapshot and
edge's, so a package built against stable's libraries is not necessarily
correct for rc. Copying stable's fast-ring artifacts into rc therefore shipped
possibly-mislinked packages to RC testers. Fast-ring packages now build
natively for all three channels, each in its own image against its own base
mirror, and the stable release's replication step is gone.
That required separating 'may be built here' from 'whose version wins'. The
release pair is now marked "pinned": its version is set per release on the rc
branch, so it builds for rc only from that branch's worktree
(OMARCHY_RC_PINS=1, set by omarchy-release rc) — master's shipped pins can
never overwrite an in-flight RC, even though check-versions now discovers rc
work like it does for edge and stable.
A package's .omarchy/package.json may now pin where it lives with
"channels": [...]. With the key present the package builds in each listed
channel except stable (stable is only fed by promotion); without it, today's
defaults hold (build for edge; fast-ring also builds stable directly).
package_moves_to_channel() is the advance/promote eligibility rule: a member
of the destination channel that is not built there natively.
The release pair (omarchy, omarchy-settings) is edge+rc+stable — edge stays
during the client-migration overlap window and drops later. The dev pair is
pinned to edge only.
#6746 added the unit, the binary, the enable-user-units.sh entry, and a
migration, but not the install line here -- so omarchy-settings shipped
omarchy-crash-watch.service only into the default/ template tree and never
into the search path systemd actually reads.
install/user/first-run/enable-user-units.sh enables its six units in a single
`systemctl --user enable --now` call, so the missing unit failed the whole
call. omarchy-provision-first-run only marks first-run-user when every step
succeeds, which meant first-run never completed and replayed on every login,
re-firing the "Learn Keybindings" and "Update System" notifications. Because
enable is atomic, it also left the other five units disabled -- no bluetooth
agent, sleep lock, monitor recovery, migrate notifier, or fcitx5.
Migration 1786539345 falls back to writing the wants symlink by hand when
there is no live user manager, pointing at the /usr/lib path this omission
left empty, so existing installs got a dangling symlink too.
No new migration is needed: affected installs retry first-run on the next
login and succeed, and the dangling symlinks resolve as soon as the file
exists at that path.
test/shell.d/config-test.sh already asserts this install and fails without it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Jm1hEGtGRUrfqxMbTi1fWe
Stable-track counterparts of the -dev packages. Source is pinned to a
single shared upstream commit with a real sha256 (pacman 7 validates
checksums on commit-pinned git sources), versioned per the attached-rcN
convention so vercmp orders rc1 < rc2 < final. Seeded at 4.0.0rc1 from
the quattro tip because no upstream 4.x tag exists yet and the packaging
requires the quattro tree (v3.8.x has no shell/ or etc/).
Both packages stay off the fast ring: RCs publish to edge only; stable
receives finals via bin/repo migrate.
Marcelo Alcantara