Commit graph

3 commits

Author SHA1 Message Date
Sarthak
0780a3abf5 readme at the root, and fix what would break on a fresh machine
The repo landing page showed .config, .themes and .gtkrc-2.0 and nothing else,
so there was no sign the bootstrap existed. It has always been at
.config/rice/install.sh - that is just where it lands on disk. Root README now
carries the curl one-liner, a map of what is tracked, and a note that this is a
bare repo over $HOME so nobody goes looking for a wrapper folder again.

Things that would not have survived a replication:

  - polybar's click and scroll handlers called $HOME/bin/system-info.sh and
    $HOME/bin/volume.sh. That directory does not exist; the scripts are in
    .config/scripts. The exec lines already pointed there, so only the
    interactive bits were dead.
  - bspwmrc hardcoded DisplayPort-0, so no desktops on any other machine.
    Takes the xrandr primary now, falling back to the first monitor bspc sees.
  - install.sh picked conflicting files out of git's checkout error with a
    regex that required a dot inside the name, so .bashrc and .xinitrc were
    never matched and never backed up - and .bashrc always exists on a fresh
    Arch install, which is precisely when the checkout aborts.
  - bspwmrc sourced ~/.env and merged ~/.Xresources unconditionally; neither
    exists. Guarded.

packages.txt was missing programs whose configs this repo already carries:
fish, neovim, yazi, kitty, plus xrdb/xsetroot/xsettingsd/wmname that bspwmrc
launches, gtk3 for the control panel, and wireplumber for the wpctl calls.

Newly tracked: .xinitrc (the thing that actually starts bspwm), .bashrc (which
holds the dotfiles alias), greenclip, flameshot, kitty, Thunar, mimeapps and
user-dirs. Dropped six editor backups that had been committed by accident.
Skipped QtProject.conf - it is file-dialog history, not configuration.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011beBgzmH72WFn8wiWqRZPf
2026-09-04 10:09:16 +05:30
Sarthak
e2dfb1a67c gtk: force the dark variant, autostart particles
GTK4 ignores gtk-application-prefer-dark-theme, and did not follow the
portal's prefer-dark either, so Adwaita stayed light while the custom css
darkened only some surfaces - that is why pavucontrol had dark-grey labels
on dark backgrounds and a white content pane. GTK_THEME=Adwaita:dark fixes
GTK3 and GTK4 together.

Also recolour the stale ~/.themes Terminal-Dark leftovers, point gtk-4.0 at
a theme and icon set that exist, and start the particle layer from bspwmrc.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Cn3dHfbgNYdbZvzb1b3GQt
2026-09-04 00:33:39 +05:30
Sarthak
f341989335 Sakura Line: retheme the whole desktop from the wallpaper
Palette sampled from bspwm/wallpaper/wallpaper.jpg and applied per-role
(not per-hex, so terminal green stays green and nvim's LineNr/Comment
stay recessive) across st, kitty, polybar, rofi, dunst, nvim, qt5ct/qt6ct
and GTK 2/3/4.

Also:
- rofi drun centred, icons enabled
- polybar translucent (true ARGB) and rounded; picom corner-radius, with
  the desktop layer and fullscreen windows excluded
- fix qt5ct color_scheme_path pointing at a nonexistent /home/carrie
- export QT_QPA_PLATFORMTHEME so qt6ct is actually consulted
- point GTK at an icon theme that exists on this system
- add .config/rice: one-shot install.sh (--dry-run supported), package
  list, vendored st config.h (gitignored upstream) and the particles
  wallpaper-layer source, so a clone reproduces the rice

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Cn3dHfbgNYdbZvzb1b3GQt
2026-09-04 00:19:36 +05:30