PRs basecamp/omarchy#6429 and #6430. Both target `quattro` — upstream `master`
is still Omarchy 3 and has no default/hypr/apps/davinci-resolve.lua at all.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Rebased onto the Omarchy 4 update (4857667), which the branch these commits
came from predated. Kept both sides:
- the O4 fixes stay (per-package pacman installs, gtk2 dropped, wrapper
unsets QT_STYLE_OVERRIDE/QT_QPA_PLATFORMTHEME)
- the full-opacity block stays alongside the new rules, noted as redundant
on Omarchy >= basecamp/omarchy#6382 but still needed on older 4.x builds
README renumbered accordingly and the "002 not fully verified" note dropped,
since both fixes are now verified end-to-end.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Two Resolve problems on Omarchy 4.x turned out to be bugs in Omarchy's own
Hyprland rules rather than in this installer, so document them as
ready-to-raise upstream patches:
001 the status bar covers Resolve's menu bar
002 Resolve's dialogs trap the pointer until Resolve is killed
Each file is self-contained: symptom, root cause with captured command
output, the exact patch in Omarchy's house style, verification performed,
how to re-verify, and reviewer caveats. Both fixes are verified end-to-end
on Hyprland 0.56.1 / Omarchy 4.0.0.r1440.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- Remove gtk2 from deps (dropped from Arch repos; Resolve 21 no longer
needs it) and install packages individually so one missing target
can't abort the whole pacman transaction
- Append a resolve-full-opacity rule to ~/.config/hypr/hyprland.lua on
Omarchy 4, overriding the new global 0.985/0.96 window translucency
(wrong for colour-critical grading); idempotent, skipped pre-4
- Unset Omarchy's global QT_STYLE_OVERRIDE/QT_QPA_PLATFORMTHEME in the
wrapper (Resolve's bundled Qt lacks the kvantum/gtk3 plugins)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Without this, wireplumber's auto-default algorithm promotes aloop to
default whenever Resolve plays audio (because aloop becomes RUNNING),
and the bridge's "send to default sink" output ends up back in aloop —
renders complete, but monitor audio loops onto itself with nothing
reaching real hardware.
Writes 51-resolve-aloop-no-default.conf with priority.session = 0,
priority.driver = 0, and node.dont-fallback / node.disable-fallback
to cover wireplumber 0.4 and 0.5+. Restart order changed to
wireplumber-first so the rule is in place when the aloop nodes
republish.
Resolve aborts on first launch on systems without a DeckLink card (default
Local.Audio.Type = DeckLink), and Resolve's audio engine retries forever
enumerating raw ALSA hardware when PipeWire owns every real card — the
"render queue stuck in progress with no output" hang.
Patches default-config.dat (and any existing user config.dat) to ALSA, then
loads snd-aloop as a virtual ALSA card PipeWire ignores, persists it via
/etc/modules-load.d/, and writes a PipeWire loopback bridge so monitor
audio remains audible during playback. Skip with RESOLVE_NO_ALOOP=1.
Ported from the Intel Arc installer where the fix was originally debugged.
