nandi/frq · a400a00

nandi/frqpublic⑂ Fork 0

⑂ a400a00

Follow the xorg renames, and stop wrapping nix in distrobox

The xorg package set is deprecated; the four X libs the runtime loader
path needs are now top-level attributes.

CLAUDE.md's nix rule was written from the host. Sessions run inside the
Arch container already, so the wrapper it prescribed does not resolve.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

nandi committed 2026-09-02T17:46:21-07:00 Browse files

a400a00 parent: 270bee0

modified CLAUDE.md +8 -9 Viewed

@@ -2,19 +2,18 @@
2	2
3	3	## Nix
4	4
5		-`nix` is not installed on the host. It lives in the Arch distrobox, so run every
6		-nix command through that:
	5	+You are already running inside the Arch distrobox, where `nix` lives, so run
	6	+nix commands directly — do not wrap them in `distrobox enter`:
7	7
8	8	```bash
9		-distrobox enter arch -- bash -lc 'cd <this directory> && nix build .#frq'
	9	+nix build .#frq
10	10	```
11	11
12		-The worktree path is the same inside the container as outside, so `cd "$PWD"`
13		-works. A remote builder (`eu.nixbuild.net`) is already configured there; the
14		-Android outputs want `--store ssh-ng://eu.nixbuild.net --eval-store auto`, for
15		-the reason nix/android.nix gives at the top.
	12	+A remote builder (`eu.nixbuild.net`) is already configured here; the Android
	13	+outputs want `--store ssh-ng://eu.nixbuild.net --eval-store auto`, for the
	14	+reason nix/android.nix gives at the top.
16	15
17		-One thing that container is not representative of: `/etc/localtime` is a
18		-regular file there rather than a symlink, so anything that reads the zone out
	16	+One thing this container is not representative of: `/etc/localtime` is a
	17	+regular file here rather than a symlink, so anything that reads the zone out
19	18	of its path sees nothing. That is a real deployment shape, not an artefact —
20	19	frq.clock handles it.

  ## Nix
--`nix` is not installed on the host. It lives in the Arch distrobox, so run every
++You are already running inside the Arch distrobox, where `nix` lives, so run
--nix command through that:
++nix commands directly — do not wrap them in `distrobox enter`:
  ```bash
--distrobox enter arch -- bash -lc 'cd <this directory> && nix build .#frq'
++nix build .#frq
  ```
--The worktree path is the same inside the container as outside, so `cd "$PWD"`
++A remote builder (`eu.nixbuild.net`) is already configured here; the Android
--works. A remote builder (`eu.nixbuild.net`) is already configured there; the
++outputs want `--store ssh-ng://eu.nixbuild.net --eval-store auto`, for the
--Android outputs want `--store ssh-ng://eu.nixbuild.net --eval-store auto`, for
++reason nix/android.nix gives at the top.
--the reason nix/android.nix gives at the top.
--One thing that container is *not* representative of: `/etc/localtime` is a
++One thing this container is *not* representative of: `/etc/localtime` is a
--regular file there rather than a symlink, so anything that reads the zone out
++regular file here rather than a symlink, so anything that reads the zone out
  of its path sees nothing. That is a real deployment shape, not an artefact —
  frq.clock handles it.

modified flake.nix +5 -5 Viewed

@@ -108,7 +108,7 @@
108	108
109	109	# egui reaches for these with dlopen rather than linking them, so being
110	110	# in the cdylib's buildInputs is not enough — whatever starts frq has to
111		- # put them on the loader path itself. Without libX11 here, vidya reports
	111	+ # put them on the loader path itself. Without libx11 here, vidya reports
112	112	# "X11 unavailable", falls back to Wayland, and winit refuses to build a
113	113	# second event loop after the failed first one.
114	114	#
@@ -120,10 +120,10 @@
120	120	libGL
121	121	libxkbcommon
122	122	wayland
123		- xorg.libX11
124		- xorg.libXcursor
125		- xorg.libXi
126		- xorg.libXrandr
	123	+ libx11
	124	+ libxcursor
	125	+ libxi
	126	+ libxrandr
127	127	vulkan-loader
128	128	];
129	129	in

        # egui reaches for these with dlopen rather than linking them, so being
        # in the cdylib's buildInputs is not enough — whatever starts frq has to
--      # put them on the loader path itself. Without libX11 here, vidya reports
++      # put them on the loader path itself. Without libx11 here, vidya reports
        # "X11 unavailable", falls back to Wayland, and winit refuses to build a
        # second event loop after the failed first one.
        #
          libGL
          libxkbcommon
          wayland
--        xorg.libX11
++        libx11
--        xorg.libXcursor
++        libxcursor
--        xorg.libXi
++        libxi
--        xorg.libXrandr
++        libxrandr
          vulkan-loader
        ];
      in