tlater / dotfiles Goto Github PK

Whenever StumpWM is launched after an update, it will fail to find any modules loaded with load-module. This causes it to fail loading.

The workaround is to manually load swm-gaps after launch using the StumpWM eval: (load-module "swm-gaps").

After this, for some reason modules will load fine... I suspect there is some caching going on on the lisp end, but the workaround seems to only work when the load is invoked interactively.

/sys/devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:00/PNP0C09:00/ACPI0008:00/iio:device0/in_illuminance_raw

Since #299 I'm using sway, which uses a different xdg-desktop-portal implementation than Hyprland.

Apparently, nvidia produces images in an unsupported color format, which causes xdg-desktop-portal-wlr to crash.

This is fixed in emersion/xdg-desktop-portal-wlr#191, however xdg-desktop-portal-wlr has not had a release since.

Even if I use xdpwlr directly from git, it might not actually work, and I'd need a patch for wlroots. It might work for the things I care about, though, considering the Hyprland portal worked fine, so I should try that.

While it's cool to pull in dev tools for specific projects into the projects and not elsewhere, I need to write a flake file to do that first. I'd like nixfmt to be available for that.

Also possibly watch https://github.com/nix-community/rnix-lsp for future inclusion.

For some modes, when asked to render a buffer they will complain that they're missing a syntax table and are therefore unable to load.

As it turns out, this is because js-mode isn't loaded previously.

Currently I know of these modes being affected:

• js2-mode
• mhtml-mode

Docs are currently in debauchee/barrier-wiki#8.

Emacs unconditionally cleans up its socket when it shuts down, but systemd expects the socket to be kept alive.

Hence, we need to teach emacs not to do that.

Seems there's a variable server--external-socket-initialized that will detect if systemd passed the socket to us.

There's also server-sentinel in server.el:396, which cleans up the socket file. Presumably all that needs to be done is to make it check whether an external process created the socket, and if it did, not clean it.

I've made the basic config work again in 7a152b8, however theming requires more effort.

See the 1.7.0 release

Likely feasible

• ~/.python_history
  • Need to write a python shell setup script as documented here
• ~/.esd_auth
  • Need to disable the module-esound-protcol-unix.so pulseaudio module
  • This is set system-wide in /etc/pulse/default.pa, and I'm not
    sure how to override module loads
• ~/.mailcap
  • Only used by mutt currently, need to see if its location can
    realistically be changed
• ~/.mbsyncrc
  • Set by home-manager, should be easy to change
• ~/.stumpish_history
  • Need to see if this can be changed
• ~/.Xresources ~/.xprofile ~/.xsession
  • All created and used by home-manager, though they are default
    paths, it should be possible to change
• ~/.lyrics
  • Probably ncmpcpp? Should be easy to fix.
• ~/.bash_history
  • Not sure what calls bash, but this should be easy
• ~/.ipython

Less feasible

• ~/.zcompdump
  • Fixed in #48
  • Not actually fixed
  • Fixed again in #76
• ~/.mozilla
  • Infuriating issue open for 18 years
• ~/.pki
  • Also caused by Firefox. Seriously, I need a different browser. Maybe I will switch to chrome after all. Or maybe patch Firefox.
• ~/.stumpwm.d
  • Seems hard to change since it's defined both in a function which would need to be overridden, but isn't on time,
    and in a variable
  • Go upstream with this one!
• ~/.java
  • I set _JAVA_OPTIONS, not sure why this is created anyway. Seems it's a bug
• ~/.profile
  • Fairly standard generic "env variable" definition point, unlikely to be able to change this one
• ~/.nix*
  • Said to be obsolete as of 5 years ago, but flakes never happened, did they?
  • See also the upstream issue