qemu-binfmt-conf.sh groups arm with aarch64 and skipped it; Apple Silicon has
no AArch32 execution, so --ignore-family is needed for ARM32 programs. The
bin/sync-upstream extension, its tests and the Tests workflow edit are
reverted; QEMU updates become reviewed pins.
Chromium sandboxes through unprivileged user namespaces on Arch-family
kernels, so the setuid-root helper is unnecessary. The watcher tests now run
from check() as checksummed package sources, leaving tests/ and the shared
Tests workflow untouched.
Aurora Silicon's fork of the Asahi kernel (DisplayPort alt-mode and USB4
for external displays, VRR, ISP and AOP drivers) as an aarch64-only split
package, linux-aurora and linux-aurora-headers, pinned to an immutable
commit that was cold-boot qualified on a MacBook Pro 14" M1 Pro and 16"
M2 Max. It provides and conflicts with linux-asahi so Macs move over in
one explicit transaction, and pkgrel 9 orders it after the last build the
maralcbr fork published, so already-installed Macs upgrade on -Syu.
Metadata: skip_build like the x86 kernels (built only by its own PR),
release_ring fast with channels edge and rc, so one merge publishes the
same artifact to edge/aarch64 and rc/aarch64 and stable keeps asahi-alarm's
linux-asahi. No x86 job is emitted and no shared tooling changes.
Update both package recipes and source checksums to OWE 0.2.6. This release includes the transition-image timeout and output-selection fixes. Package builds pass on x86_64 and aarch64.
0.2.3 adds one-shot intro playback: owe intro, intro-status, and intro-stop, plus the renderer once/mute load and keep-open hold. Both PKGBUILDs take the v0.2.3 tarball; checksums verified with makepkg --verifysource.
Co-authored-by: Bjarne Oeverli <1419214+bjarneo@users.noreply.github.com>
The all-themes.lua spec references "gthelding/monokai-pro.nvim", a fork
that no longer exists on GitHub (HTTP 404). On every launch lazy.nvim
fails to clone it and reports "repository not found", and leaves a stale
lazy/monokai-pro.nvim.cloning marker so the error repeats.
Restore the canonical, actively maintained upstream
"loctvl842/monokai-pro.nvim" -- the value this line held before
80a278206 ("Match nvim themes"). It registers the same "monokai-pro"
colorscheme and still supports filter = "ristretto" plus the setup
override used by Omarchy 3.8's themes/ristretto/neovim.lua, so nothing
downstream changes except that the plugin resolves again.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011FcsS1gPeUiLSqqi4iKgzm
Both upstreams set fat LTO in their release profile (strata also
codegen-units = 1), which compiles the final crate on one thread. That
is a reasonable trade for the binary users run, and build() keeps it.
check() then compiled the same crate a second time under the same
profile for the test harness, and threw it away. On the aarch64 builds,
which run under QEMU, those two serial compiles were 106 of strata's
118 minutes.
check() now builds the harness with thin LTO and 16 codegen units in
its own target directory. The shipped binary is unchanged. Measured on
the x86_64 builder for strata's test binary: 708 s serial, 223 s with
this profile. Local run: shipped strata still 7m30s from target/, tests
3m43s from target-check/, 1998 passed.
ttfx has the same upstream profile but no check(), so nothing to do.
Marcelo Alcantara