Skip to content

Sync upstream quattro through 2fca7fe0 - #95

Merged
mark-groves merged 35 commits into
quattrofrom
cursor/upstream-changes-pr-753f
Oct 26, 2026
Merged

mark-groves merged 35 commits into
quattrofrom
cursor/upstream-changes-pr-753f

Conversation

@cursor

@cursor cursor Bot commented Oct 25, 2026

Copy link
Copy Markdown

Upstream quattro moved from 4fa93e9f (24 Oct, already on this fork via #93) to 2fca7fe0 (25 Oct, Update translations omacom#16604). This replays 35 of the 36 first-parent commits. Two merge commits are included as their net result: scratchpad kept separate from the Quake console (omacom#7579) and the power-profiles migration finishing without a session (omacom#16565).

What changed

Desktop and shell

Power profiles

Updates and agents

Setup, theme, and translations

Kept on the fork

Upstream omacom#16568 would share the one update authorization with AUR builds and the boot check. This fork does not take that. AUR, Grok Bot, Cursor, and the boot check still run cold, each behind the no-update wrapper, after the shared authorization ends. Face, Grok Bot, and Cursor stay in the menu. Cursor stays the official tarball, not a mise tool. English drops agents.helpStartToSeeLimits and the power-profile menu strings. The notification panel, snooze, and the notifications settings object stay out.

The translation manifest conflict on omacom#16574 was resolved by taking upstream's manifest. Catalogs still omit agents.helpStartToSeeLimits.

Tests

Passed: update-sequence, update-hook-security, update-boot-verify, i18n, launch-browser, hyprland-qconsole, keybindings-menu, hyprland-floating-workspaces, hyprland-default-config, monitor-recovery, platform-packages, platform-setup, first-run, agent-account, omasnap, default-browser-lookup, omarchy-kernel, agent-usage-shared, default-apps, clipboard, and menu.

./test/shell was 19 of 479 failed. The same five fail on quattro: hyprland-window-close-all (pidwait rejects the pidfile), lock-face (Loader assertion), bin-style (face and Cursor installers), factory-reset-accessory-trust (ESP_MNT unbound), and setup-security-face (/run/lock not writable). The rest are the usual sandbox gaps: no mise, socat, or pacman-conf, updatedb pruning, systemd-tmpfiles --inline, and Asia/Chungking.

Open in Web View Automation 

ryanrhughes and others added 30 commits October 25, 2026 23:03
Keep scratchpad separate from Quake console
* Launch a browser again when the configured one is gone

With ~/.config/mimeapps.list naming a browser that is no longer installed,
xdg-settings and xdg-mime both answer omarchy-browser.desktop. Discarding
the routing entry left omarchy-cmd-default-browser empty, so
omarchy-launch-browser passed uwsm-app an empty command and every link
ended in "Command not found: \"\"".

Only accept an xdg answer once its command is installed, and fall back to
the first installed https handler in mimeinfo.cache, which covers any
browser rather than a fixed list. Read Exec= from the same data dirs the
resolver searches, including the Nix profile the launcher always read,
and refuse an empty command with a notification that says how to set a
default browser.

Fixes omacom#15814

* Read the web app's Exec= from the data dirs the resolver searched

omarchy-cmd-default-browser can now settle on any installed https handler in
the XDG data dirs, but omarchy-launch-webapp still read Exec= from only
~/.local, ~/.nix-profile and /usr. A Chromium-family browser it accepted from
/usr/local/share or another XDG data dir resolved, then launched uwsm-app with
--app=... as the command. Read the same dirs in the same order as
omarchy-launch-browser now does.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Codex Medium <noreply@openai.com>
Co-Authored-By: muse-spark-1.3-contributor High <noreply@meta.com>

* Guard the web app launcher against a missing browser

omarchy-launch-webapp fell back to chromium.desktop when no default browser
resolved, and with that entry gone it handed uwsm-app an empty command and
"--app=..." as the command name. Reuse the browser launcher's guard and
notification instead, so both paths fail with an actionable message.

* Tell web app users they need a Chromium-based browser

omarchy-launch-webapp maps any non-Chromium default to chromium.desktop,
so with Firefox set and no Chromium the empty-command guard claimed no
default browser was found and told the user to set one again. Say that web
apps need a Chromium-based browser and how to install Chromium instead.

* Drop the Nix profile lookup and trim the browser tests

Both launchers and the resolver read only the XDG data dirs now. A Nix
profile that should be searched belongs on XDG_DATA_DIRS, which
home-manager already sets.

The tests keep the cases that fail on quattro: the mimeinfo.cache
fallback, an uninstalled xdg-settings answer, Exec= from XDG_DATA_DIRS,
and the two launcher guards. The Nix cases, the resolver exit check the
launcher guard already covers, and the duplicate web app data-dir case
are gone, and each guard checks its notification once.

---------

Co-authored-by: Vitally Tezhe <vitally@tezhe.eu>
Co-authored-by: Omarchybot <317366263+omarchybot@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Codex Medium <noreply@openai.com>
Co-authored-by: muse-spark-1.3-contributor High <noreply@meta.com>
Co-authored-by: Ryan Hughes <ryan@heyoodle.com>
omacom#16397)

Hyprland 0.56 unloads a plugin as soon as a config load stops asking for it, and loading or unloading hyprbars costs a cascade of config reloads: four on the way in, three on the way out. Because the plugin was only asked for while something floated, every Super + L cycle through floating paid both, stalling the compositor each time. Once the plugin is in, it is now asked for on every config load, and draws nothing until a floating workspace turns its options on.

Co-authored-by: Omarchybot <317366263+omarchybot@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…tions (omacom#16550)

* Translate the first-run, git setup, weather and input-method notifications

The first-run invitations, the omarchy-setup-git wizard, the weather
status and the input-method failure toast sent English literals rather
than going through omarchy-i18n. The extractor and the key-reference
check now cover install/ too, so a literal there fails i18n-check.

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

* Cover install/ in the i18n check and extraction, and stop matching English weather text

i18n-check skipped pull requests that only touched install/, and the sync
workflow could not extract or publish install/ changes, nor flag them for
human review. The weather sample compared against the English
unavailable text, which is now translated; it checks the exit status.

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

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
… ARM laptops (omacom#14737)

* Switch to omarchy-power-profiles-daemon for working power profiles on ARM laptops

Stock power-profiles-daemon has no driver for cppc_cpufreq, so ARM laptops
only got placeholder balanced/power-saver profiles that changed nothing, and
no performance profile for omarchy-powerprofiles-set to pick on AC. Omarchy's
build adds a CPPC driver and fixes the powerprofilesctl shebang itself, so the
installer and Quattro upgrade no longer sed it.

A migration swaps the package on existing installs, restarts the daemon and
reapplies the AC/battery profile.

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

* Leave a stopped or masked power-profiles-daemon alone in the swap

Setting a profile reaches the daemon over D-Bus, which starts a daemon
the user disabled, and fails when it's masked. That failure stopped
every later migration for the update.

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

* Keep the powerprofilesctl shebang repair in the quattro upgrade

The upgrade installs whatever base list the channel's omarchy package
carries, and until a stable release has this change that list still
picks the stock daemon. The repair is a no-op on the Omarchy build.

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

* Drop the power-profiles-daemon migration test

AGENTS.md and agents/skills/migrations.md now say not to add a test for
a migration: it runs once per machine and is frozen once it ships.

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

* Note why the base list uses Omarchy's power-profiles-daemon

AGENTS.md asks for the reason next to any replacement package.

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

* Let omarchy-reinstall-pkgs replace the stock power-profiles-daemon

Without --ask 4, pacman --noconfirm declines the conflict and the
reinstall aborts on a machine the swap migration hasn't reached.

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

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Ryan Hughes <ryan@heyoodle.com>
Co-authored-by: Codex XHigh <noreply@openai.com>
Its condition had no status function, so the implicit success() failed
whenever extract was skipped upstream, which is nearly every run: no
translation has been produced since the workflow landed.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Finish system updates when mise tool downloads fail

* Keep one test for a failed mise update

The mise+boot case repeats update-boot-verify-test.sh, since the boot gate never reads the mise result, and the new hook-security block repeats that file's success-path checks: the code between mise and AUR is the same either way. The interactive pass adds nothing over -y.

* Note that a failed mise update no longer stops omarchy update

---------

Co-authored-by: Ryan Hughes <ryan@heyoodle.com>
…16506)

open_json read a successful body through _read_by, bounded by three times the socket timeout, but an HTTP error status read its body with error.read(), which only bounds each wait. A provider answering 429 or 500 and dripping its body held the collector, and the updater's wait held every later refresh behind it. Past the deadline the body is left out and the status still answers.

Co-authored-by: Omarchybot <317366263+omarchybot@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Codex Medium <noreply@openai.com>
…onds (omacom#16562)

* Take the low battery warning down once the battery is no longer low

Critical toasts never expire on the new notification engine and are restored on every shell start, so the warning stayed on screen after plugging in, after recovering, and across restarts. The battery service now closes it in process whenever a check finds the battery on AC or above the threshold, and again whenever a matching toast is shown while that holds: one restored at startup, released by a held deck, or delivered after the charger went in. It matches the sender and icon rather than the translated title, so it works in any language, and it spawns nothing. After a restart while still low, the restored warning gives way to the fresh one rather than stacking under it.

The notifications service gains the two pieces that needs: dismissWhere(app, appIcon), which reaches rows held behind an open deck or still waiting to be inserted as well as those on screen, and a shown signal for every row on its way to the screen, including a replacement folded into an insert still queued. dismissWhere drops a matching row's queued replacement, since one landing after the close would bring the card back. History asked for with showHistory is neither announced nor dismissed. A user's clone of the notifications plugin from before this change has neither piece, and the battery service carries on without them.

Fixes omacom#8813
Fixes omacom#14639

Co-Authored-By: Diogo Ferreira <diogo@underdev.org>
Co-Authored-By: Codex Medium <noreply@openai.com>
Co-Authored-By: Muse Spark 1.3 High <noreply@meta.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Write the low battery notification docs on one line

Omarchy's AGENTS.md asks for full lines in docs/; the surrounding bullets predate that.

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

* Revert "Write the low battery notification docs on one line"

This reverts commit dc7552a.

* Revert "Take the low battery warning down once the battery is no longer low"

This reverts commit 65fb295.

* Send the low battery warning as a normal toast that goes after 30 seconds

As a critical toast it never expired on the new notification engine and came back on every shell start, so it stayed on screen after plugging in, after recovering, and across restarts on AC. It already asks for 30 seconds; only critical urgency overrode that. As an omarchy-action toast it still gets through Do Not Disturb.

Fixes omacom#8813
Fixes omacom#14639

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

---------

Co-authored-by: omarchybot <omarchybot@users.noreply.github.com>
Co-authored-by: Diogo Ferreira <diogo@underdev.org>
Co-authored-by: Codex Medium <noreply@openai.com>
Co-authored-by: Muse Spark 1.3 High <noreply@meta.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…on-ssh

Let the power-profiles-daemon migration finish without an active session
…d a packaged install (omacom#16570)

- base-test resolves ROOT physically, so omarchy-plymouth-set, which refuses a non-canonical source root, accepts a checkout reached through a symlink.
- base-test unsets the agent account homes (CLAUDE_CONFIG_DIR, CODEX_HOME, GROK_HOME and the OMARCHY_AGENT_*_HOME pins), which leaked into agent-account-copy when the suite ran inside an agent session and made a fake running CLI look like it used another home.
- omarchy-kernel-test expects the root half the command actually picks: the packaged /usr/bin/omarchy-kernel where one is installed, else the checkout.
- platform-setup-test stubs every first-run script the repo ships, so first run completes now that git.sh is one of them.

Co-authored-by: David Heinemeier Hansson <david@hey.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
A remote herdr SSHing in at boot would otherwise start the server with a
tty environment, leaving its panes without WAYLAND_DISPLAY so agents can't
paste clipboard images.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Auto-merge never took without a required check, and the next run
rebuilds i18n-sync from quattro, so an open sync pull request lost its
translations to it (omacom#16569). collect has already run the structural
i18n test on these files, so land merges straight away.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Describe the screenshot toast with Edit and Open buttons

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

* Announce saved screenshots with Open and Edit buttons

Pass Omasnap --notify from the screenshot and clipboard-image commands, so a
saved shot gets the recording-style toast instead of a timed preview.

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

* List Open before Edit in the screenshot docs

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

* Add catalog strings for Omasnap's screenshot toast

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

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
)

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Replace Osaka Jade with Jade Horse

Jade Horse brings three upscaled Year of the Horse wallpapers and an
omarchy wordmark recolored from the shared one, so the glyphs land on
the same pixels as every other theme. The unlock images are repainted
from Osaka Jade's in the new palette. A migration moves anyone still on
the bundled Osaka Jade over, since refreshing a theme that no longer
exists fails.

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

* Drop the Osaka Jade migration so existing themes stay as they are

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

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A PopupCard stays mapped while it fades out, and until now it took input the whole time. When a tray app quits from its own menu and was the only tray icon, the menu's anchor collapses to the empty tray and the fading card sits on the bar, over the widgets beside the tray. With its focus grab already released, a desktop click does not dismiss it, so if the fade does not finish those widgets stay dead until the bar is rebuilt.

Give the card an empty input region once it is closed, so it only ever takes input while open.

Co-authored-by: Omarchybot <317366263+omarchybot@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…omacom#13749)

When the internal panel is off and the last external monitor is unplugged,
Hyprland adds a FALLBACK placeholder output. The external-active check
counted it as an external display, so the internal panel was never
recovered.

Fixes omacom#13741

Co-authored-by: dhia <dhia.salhi@ensi-uma.tn>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
omarchybot and others added 5 commits October 25, 2026 23:05
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
…ters (omacom#16580)

* Ignore the full stop after a URL when checking translated manual chapters

A bare URL that ends a sentence picked up its period, so languages that
write the stop as 。 or । or move the URL mid-sentence failed "URLs
changed", leaving 39-backgrounds and 48-security untranslated in 13
languages.

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

* End a URL at its first non-ASCII character in the manual check

The pattern still took a translated full stop written right after the
URL (。 । ۔) as part of it, so those chapters kept failing.

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

* Strip only sentence-ending punctuation from URLs in the manual check

Cutting a URL at its first non-ASCII character let a changed link
destination such as https://localsend.org/ダウンロード pass as the
original. Keep the whole URL and remove only the full stop or comma
that ends the sentence, in either script.

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

* Strip only the punctuation the published link drops

marked, which publishes the manual, leaves trailing ASCII punctuation out of a bare autolink but keeps 。, । or ۔ in it, and keeps everything inside an explicit (destination). Stripping those too let a translation through whose published link 404s, so the check now mirrors the renderer: ASCII only, bare URLs only, and a URL ends at a backtick.

The two chapters that kept failing put a bare URL against the sentence's full stop, which a language writing 。 or । attaches to the link. 39 now uses a markdown link and 48 puts its placeholder URL in a code span, so neither depends on how a language ends a sentence.

Co-Authored-By: Codex Medium <noreply@openai.com>
Co-Authored-By: Muse <noreply@muse.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Codex Medium <noreply@openai.com>
Co-authored-by: Muse <noreply@muse.dev>
…macom#16560)

* Translate the About box headings, keys, age, uptime and update date

fastfetch has no translations, so omarchy-launch-about runs a copy of
the config with the headings (re-centred by display width) and keys in
the current language, and the OS age, uptime and update rows print
through omarchy-i18n, the date in the language's locale.

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

* Fall back to the untranslated About config when jq cannot read it

fastfetch accepts JSONC but jq does not, so a comment or trailing comma
left an empty copy that the About screen then ran. Publish the copy
only when the conversion succeeds.

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

* Run the untranslated About config when there is no runtime directory

The launcher read XDG_RUNTIME_DIR at the top level, so sourcing it under
set -u without one, as launch-about-test.sh does on a bare runner,
stopped on an unbound variable.

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

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
@cursor
cursor Bot marked this pull request as ready for review October 25, 2026 23:10
@coderabbitai

coderabbitai Bot commented Oct 25, 2026 •

Copy link
Copy Markdown

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

Run configuration
  • Source mode: Auto
  • Repository: Unavailable
  • Branch: Unavailable
  • File: No configuration file from this repository applied
  • Commit: Unavailable
  • Layers: Organization UI
  • Review profile: CHILL
  • Plan: Advanced
  • Run ID: 36ba8fca-a4c5-436c-a7c6-10bf32bd8c73

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

  • 🔍 Trigger review
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@mark-groves
mark-groves merged commit 0267c3f into quattro Oct 26, 2026
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.

7 participants