| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The drawer absorbed the indicator's device actions but not the settings
window, so there was no path to the full KDE Connect app. Add an Open app
button beside Refresh, launching kdeconnect-app through execDetached so the
window outlives a quickshell reload.
|
|
The hand-rolled buttons only reacted to hover, so a press gave no visual
confirmation until the action's outcome appeared. kdeconnect made this
obvious: its row buttons shell out to kdeconnect-cli, and nothing on
screen acknowledged the click.
Button.qml now fills solid accent while the mouse is held (red for
danger), with the label flipped to base for contrast. TransportButton
gets the same treatment, and the window-switcher close button lightens
on press. Toggles, tiles and tabs are left alone: their own state change
is already the feedback.
|
|
breaktimer.sh's pause and resume arms write the state file whether or not
a daemon is running, so statusctl presentation set while the daemon was
stopped recorded a phantom paused/running state and the tile then showed
Paused or a frozen countdown for a daemon that does not exist. Gate both
verbs on btRunning, the singleton's running-or-paused derived state, so a
verb is only sent when a daemon actually exists.
|
|
|
|
The one rule worth writing down is which side writes: the daemon owns its
files and the shell only calls verbs.
|
|
Below every mode in the priority chain, so an active mode still owns the line
and breaktimer replaces only the idle 'All clear'.
It does not count toward activeCount. A running breaktimer is not a mode the
user switched on, and accenting the tile for an ordinary working day would
read as a mode stuck on.
|
|
Into the status module rather than a module of its own: break state is one
more thing the desktop is doing, which is what that page already is, and
Status.qml was already where breaktimer.sh is called from.
The paused state shows no number. The countdown is frozen while paused and a
frozen number reads as a bug, so the row says what it is instead.
|
|
The daemon already publishes these three files under XDG_RUNTIME_DIR and
waybar already consumes them, so a second consumer costs three FileViews and
invents no interface.
Read only, deliberately. The daemon loop rewrites state and phase on every
transition, so a second writer would race it, which is why presentation mode
has always called a verb rather than writing the file.
The countdown does not interpolate between the daemon's five second writes: a
local clock would drift against it and keep ticking while the daemon is frozen
outside the work window.
|
|
Nine tasks over two repos. The daemon half gains a sourced config file and a
config verb that prints the effective settings, which is what makes the
config testable without running a thirty minute work block. The shell half
reads the daemon's three runtime files, adds a row and a tile line.
Three assumptions were probed rather than trusted while writing it, each
with a precedent for biting here: padStart exists in this QML engine
(matchAll does not), a Connections resync fires correctly on a singleton
property, and the config source line leaves unnamed variables alone.
The qs verification steps redirect rather than pipe. Piped output is lost
when timeout kills the process, which reads as a silent success, and that
failure happened twice while probing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
|
The daemon already publishes its phase, state and countdown as files under
XDG_RUNTIME_DIR, and the status module already shells out to it for
presentation mode. Reading those files makes the traffic two-way for the
cost of three FileViews, with no new interface to invent.
Into status/ rather than a module of its own: breaktimer state is one more
thing the desktop is doing, which is what that page already is, and
Status.qml is already where breaktimer.sh is called from.
The config file is the other half, and stands alone. Every tunable is a
shell variable in a tracked script, so the installed copy and the committed
one have already drifted over three sound paths pointing into a home
directory. One sourced file fixes that with no parser and no format.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
|
Hovering a cursor card called `hyprctl setcursor` to switch the real
pointer, and reverted it on unhover. That looked read-only and was not:
Hyprland's setcursor handler also writes the theme to
org.gnome.desktop.interface cursor-theme, so every hover persisted a
setting the user had not confirmed. Closing the panel with the pointer
over a card left that theme set, surviving both the shell and the
session.
A `dconf watch /org/gnome/desktop/interface/` during a hover across
three cards showed the key alternating between each hovered theme and
the current one, one write per hover and one per revert. `strings` on
the Hyprland binary carries both `org.gnome.desktop.interface` and
`cursor-theme`; `hyprctl` carries neither, so the write is the
compositor's, not the caller's.
The revert could not be made reliable either. It ran through
`execDetached`, which gives no ordering, so a fast card-to-card move
could land the revert after the new set. Guarding it by lifecycle did
not help: `Component.onDestruction` fires too late to be certain, and
the same spawn race applied to Apply, which cleared the hover before
`applyCursor` and so raced the old theme against the new one.
None of that is fixable while the preview exists, because the side
effect belongs to the compositor. The four rendered shapes on each card
were always the real preview; the live switch was a bonus that could not
be had without writing the setting. Remove it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
|
The OSD showed only "Output"/"Input", not which device was in use, so
switching between, say, the analogue output and a USB DAC gave no
visible confirmation beyond the volume itself.
The name falls back description -> nickname -> name, the same order
SoundPage.devLabel uses, and elides. The panel height grows to fit the
new line.
|
|
dnd and presentation were the only modes, and presentation was the sole
holder of the Wayland idle inhibitor. That inhibitor is the only thing
that stops hypridle's loginctl lock-session at the 600s timeout, so
screen locking is on by default and there was no way to turn it off on
its own.
nolock is stored as "auto-lock disabled" so the file keeps the registry's
"missing means off" rule, and the row inverts it so the switch reads as
the default: lock on. Presentation now drives nolock the same way it
drives dnd, holding the prior value in memory and restoring it on exit,
so it needs no second inhibitor and the inhibitor has a single holder.
The page sets the disable permanently or for a number of minutes. The
deadline and its timer live in the singleton rather than the row because
the drawer page is destroyed on close while the effect has to outlive
it. The timer is shell-lifetime, the same class of limit as
dndBeforePresentation: a restart during a timed disable leaves the lock
off until it is toggled, which is safe because the user opted into the
disable.
The mode governs idle-triggered locking only. The SUPER+l bind and the
lock on suspend are separate paths and stay live.
|
|
statusctl watch reported class "down" for a missing state file and waybar's presentation rule drew a warning for it. That fires on every fresh session, before anything has written a file, which is the normal off state the registry's own invariant describes. The state was meant to flag a dead registry, but a crashed shell leaves a stale present file, not a missing one, so it never caught that case. get already read a missing file as 0; watch now matches it and the down state is gone.
|
|
Clicking a cursor card called applyCursor straight away, so the pointer,
gsettings, environment.lua and both GTK settings files all changed on a
single click with nothing to confirm. The wallpaper tab already stages a
pick and commits it from an Apply button; the cursor grid now does the
same, with a Reset beside it.
The hover preview is unchanged: it still switches the real cursor and
reverts on unhover, which is what makes staging readable, you see the
theme before committing to it. That preview is also why Apply clears
liveCursorCursor before calling applyCursor. Moving the mouse off the
card to reach the button fires the unhover revert, which would otherwise
set the old theme back over the one just applied.
A staged card takes the solid accent border and the current one drops to
a half-alpha border, so both stay visible at once.
Icon themes and cursor size still apply on click. Neither has a hover
preview to confirm against.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
|
The chat/internet-chat aliases in the icon theme all resolve to an empty
outlined monitor; dialog-messages is the filled speech bubble.
|
|
Desktop 2 is the editor and desktop 3 the terminal; the map had them the
other way round, matching waybar's stylesheet before it was corrected.
|
|
The card's desktop line showed the workspace number. waybar draws the same
eight desktops as CSS background images keyed by number, because its
format-icons takes text rather than paths; those names now live in the
Windows singleton and are resolved through the active icon theme, so a
theme switch follows the drawer and the bar and the switcher agree. A named
workspace, or one past eight, falls back to its number, and so does an icon
the active theme cannot resolve.
|
|
Cursor previews were extracted only on hover, so the row stayed blank until
each card was visited. They are now batched on open: one python pass extracts
four shapes per theme (arrow, hand, text, horizontal resize) through the
theme's aliases and every storage form it may use (hyprcursor .hlc, a shape
directory SVG, or a legacy Xcursor binary via xcur2png).
The card rows are Flows so they wrap instead of running off the right edge,
and the tab scrolls. Cards widened and names wrap to two lines. A size
selector (16-64) applies the pointer size live, to gsettings, the GTK
settings.ini files and environment.lua.
gsettings alone does not reliably switch icon/cursor here: both gtk-3.0 and
gtk-4.0 settings.ini carry the names, so apply now rewrites
gtk-icon-theme-name / gtk-cursor-theme-name / gtk-cursor-theme-size there too.
The symlinked `default` cursor alias is skipped in the scan and the current
value resolved through it, so the real theme carries the current marker.
The tab row is centred against the panel so it reads as drawer chrome rather
than the first line of a tab's content.
|
|
pkill only sends SIGTERM, and hyprsunset binds Hyprland's CTM manager
exclusively, so the replacement started immediately after pkill raced the
old process and exited with 'A CTM manager is already running'. Saving from
the Sunset tab therefore killed the night-light daemon instead of restarting
it. Bound the wait to 5s (50 x 0.1s) so a wedged process cannot hang the
restart.
|
|
xcur2png wrote its config file into the qs CWD, which is the repo root, so
hovering a legacy Xcursor theme left a stray left_ptr.conf in the tree. Point
-c at the per-theme cache dir instead.
Hyprsunset.save() could run before refresh()'s two cat processes landed and
write a header-only hyprsunset.conf, blanking the location. Guard on the same
confLoaded/appLoaded flags Hypridle already uses.
The idle save had no acknowledgement, which the spec requires. Add a notice
after the restart and show it in the tab, and switch Hypridle's save guard to
a loaded flag so a config whose first block is a listener is not mistaken for
an unloaded one.
classify() used else if, so a directory carrying both index.theme and cursors/
(Adwaita) was listed as a cursor theme only and never reached the icon picker.
Make the two tests independent; selftest gains the dual-purpose case.
|
|
The five-tab shell's paragraph still said theme opened "the other tab",
and the widened top border left the body and bottom border 6 columns
short. Widen lines 7-14 to match and name the Theme tab.
|
|
Records the IPC verbs, the shared hyprsunset.conf format, the preserved
hypridle comments, and the preview mechanisms. The hardcoded icon theme in
udt, waybar and rofi is tracked as a TODO in those repos.
|
|
Icon themes preview four glyphs each from the GTK lookup; cursor themes
preview an extracted left_ptr and change the real cursor on hover, reverted
on leave and on close. Selection applies both.
|
|
The spec calls for a seconds field shown as m:ss, but the first pass
rendered the brief's '9m 30s' form. Zero-pad the seconds so 570 reads
9:30 and 600 reads 10:00.
|
|
Rows for each listener with an enable toggle and a timeout field, the
command shown read-only. Save rewrites the file and restarts hypridle.
|
|
Tab bar and Tab-key cycle become five entries, the Loader picks inline or
file tabs, and show() refreshes the model. The tab edits the profiles
in place, fetches sun times and previews through the daemon.
|
|
Lists themes from ~/.icons, ~/.local/share/icons and /usr/share/icons,
distinguishing cursor themes by cursors/ or manifest.hl. Icon previews
come from one GTK lookup process for every theme; cursor previews extract
left_ptr from a hyprcursor .hlc via unzip or from an Xcursor theme via
xcur2png. Application sets gsettings and the Qt configs, and rewrites the
cursor env in environment.lua.
|
|
save() could run before refresh()'s async cat completed, when prefix and
tail are empty and listeners is empty, so it would write an empty file and
restart hypridle with no listeners. The real config always has a general
block, so an empty prefix means the load has not happened or failed.
|
|
Parses the general block and comments before the first listener and keeps
them verbatim; only timeouts and the enabled flag are editable, commands
are copied. Disabled listeners are written commented out, and the parser
matches the closing brace at line start so the dpms command's inline brace
does not truncate the block.
|
|
appConfText() hardcoded path = ~/.config/hyprsunset-qt/sun.json, so any
saveSettings() -- from save() or detect() -- silently reset a user-customized
cache path while fetchSun() kept writing to the parsed location. Mirror
sunset-qt's settings.py: hold the raw unexpanded string in cachePathRaw for
serialization and derive the expanded cachePath for file access, so a no-op
save is byte-identical and the two stay consistent.
|
|
Parses and serializes ~/.config/hypr/hyprsunset.conf in the exact format
hyprsunset-qt writes, so both editors share the file. Location lives in
hyprsunset-qt's own INI; fetching mirrors its curl calls and browser
User-Agent. Selftest round-trips a fixture byte-for-byte.
|
|
The note pointed at waybar/scripts/notifications.py and its module by path.
Both moved to ~/bin/archive/ with the rest of the stock waybar theme, whose
scripts had no caller left once this shell took over notifications.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
|
unified-desktop-theme moved its generated files out of ~/.cache/wal, which
pywal used to own and nothing writes any more. The palette udt-accent
generates is now at ~/.cache/udt/udt-palette.qml.
shared/Theme.qml is the only code change; every component reaches it through
a symlink. The module READMEs cite the path, so they move with it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
|
Seven tasks: the hyprsunset model, the hypridle model, the icon/cursor
model, then the tab bar and one task per tab, finishing with docs and the
cross-repo unhardcode TODOs. Models are built and self-tested first so
every commit loads. Records the brace-in-command parsing trap and the
sunset-qt byte-for-byte serializer shape.
|
|
Design for three new appearance tabs: a hyprsunset-qt-equivalent profile
editor, a hypridle timeout/enable editor that preserves the file's comments,
and an icon/cursor selector with live previews. Records the preview
mechanisms probed on this box (GTK icon lookup, hyprcursor .hlc unzip,
xcur2png) and scopes the hardcoded-theme propagation out as follow-ups.
|
|
The sanitizer matched the first \bsrc anywhere in the tag, so a
data-src="file:///tmp/decoy" before a remote src won the match and kept
the whole tag while Qt, which ignores data-src, fetched the remote
image. Scan every src assignment in the tag and keep it only if all of
them are local. Also set cache: false on the balloon preview, since the
daemon overwrites the same path on a replace and QQuickPixmapCache keys
on URL, so a replaced notification could show the previous image.
|
|
The http(s)-only regex let protocol-relative //host, ftp, data and
entity-encoded schemes through, and the comment overstated what it
removed. Replace it with a deny-by-default allowlist: an <img> is
removed unless its src, entities decoded first, is a file: URL or a
single leading slash. Protocol-relative //host is rejected while a
/absolute/path is kept.
|
|
A notification is untrusted input. Inline <img> now renders only for
local sources; an http(s) source is removed before the RichText body is
shown, so a remote sender cannot make the shell fetch a URL. The row and
the balloon share the one sanitizer in the Notify singleton.
|
|
The daemon now publishes an image path; the balloon shows it below the
text, scaled to the balloon width with a 240px cap. A daemon without the
field leaves it hidden, so the renderer and the daemon can ship in
either order.
|
|
Two plans, one per subsystem: the notifyd daemon (hint parsing, PNG encode, theme-name resolution, the image field and its cleanup) and the quickshell renderers (the balloon preview and remote inline source stripping). Either can ship first; the renderer tolerates a daemon without the field.
|
|
The daemon drops every image hint today, so a screenshot and an application image never arrive. libnotify 0.8.8 sends the content image via -i/--icon (the image-path hint) while -n/--app-icon is app_icon. The spec adds the image contract field, daemon hint handling and theme-name resolution, the balloon preview, and a local-only inline image policy.
|
|
Each row shows when the notification appeared ("Sep 15 14:32", from created), sits on its own rounded card with space between rows instead of reading as one list, and uses the balloon's 16/18/16 sizing rather than 13/11. The close target grows to a 32px hit area in the row and the balloon, since a 20px box was hard to hit.
|
|
The per-plan ledger, briefs and reports under .superpowers/ are scratch, not source. Committing the ignore rule keeps the tree clean for every clone and worktree instead of only this one, where it lived in the local info/exclude.
|
|
Mirror the balloon's right-click close-all on the drawer row: the spec says
the gestures are identical in both forms, and a right-click on a row
previously did nothing. The right-button branch precedes the live guard so it
behaves the same on history rows.
Stop importing the Status singleton into the notification shell. Referencing
it instantiated it, and its onPresentationChanged writes status.dnd and runs
breaktimer.sh, so a read-only consumer was writing state and shelling out on
every presentation toggle, and doubled the dndBeforePresentation race. Notify
now reads "/run/user/<uid>/status.dnd" directly through a FileView, the same
convention as the notifyd files; a missing file means off.
AGENTS.md named Drawer.qml as the drawer's reserved space; it is
desktop/NotificationList.qml.
|
|
The daemon is a separate process and the files are the interface; suppression
lives in the balloon shell so the drawer can list what DND held back; and the
drawer flag is what stops a notification appearing as both a balloon and a
row.
|
|
A switch and a free-text minutes field driving notify-snooze.sh. Snooze is a
file the balloon shell reads, so the row only writes it; the switch follows
the file, including a snooze that ends while the page is open.
|
|
show() is the IPC deep-link path and set only page/open, so if the history
view was open, historyLoader stayed active alongside pageLoader and painted
over the requested module page. Clear history in show(), as close() already
does.
|
|
Reachable only from the reserved space's History button, so it is a drawer
view, not a module and not a tile. Rows are inert; clear-all empties the
ring, since the daemon has no per-id history deletion.
|
|
onOpenChanged does not fire for the initial value, so a restart while the
drawer was open left notifyd/drawer at "1" and every balloon stayed
suppressed until a drawer open/close cycle. Write the starting state from
Component.onCompleted so the file always reflects the drawer.
|
