| 29 Jul 2025 |
 emily | I might be wrong about the work it'd take to split things up! but my prediction is that using the normal closure mechanisms and just splitting up packages where relevant is not going to look more complex than what we do instead right now | 17:45:50 |
 ElvishJerricco | which leaves me believing this^ | 17:45:53 |
| emily | there aren't that many things in initrd | 17:45:57 |
| emily | and if we have checks for initrd size (which seems good in general to stop it growing accidentally for other reasons) then it seems like the burden would be appropriately distributed to maintainers of core packages with passthru.tests | 17:46:56 |
| ElvishJerricco | I'm just not convinced it's worth the effort, or sustainable | 17:47:15 |
| emily | well, that's why I was asking if you have any examples in mind of things that would make it hard :) | 17:47:42 |
| ElvishJerricco | if someone wants to proof-of-concept it, that would be cool | 17:47:43 |
| emily | I'd be open to poking at it, but it'd of course be nice to start with the hardest thing first to not waste effort on easier stuff | 17:48:09 |
| ElvishJerricco | I don't think any specific things are difficult. It's just a lot of things and its a burden that remains forever | 17:48:15 |
| emily | I think that probably putting a program in a separate output doesn't incur repeated ongoing maintenance costs in most cases | 17:48:46 |
| emily | same way that make-initrd-ng itself doesn't much | 17:48:53 |
| emily | people do closure size minimization work for other reasons, so it's not like it would be burden solely for initrd | 17:49:18 |
| ElvishJerricco | Well if you want to give it a try, it shouldn't be too hard to swap makeInitrdNG for makeInitrd and start slashing away | 17:50:32 |
| ElvishJerricco | (oof, a stock systemd initrd is 21M. not great, but acceptable) | 17:52:20 |
| emily | yeah I might | 18:04:03 |
| emily | even just looking at the diff of the closure would probably make it obvious where to start | 18:04:16 |
| emily | (I thought makeInitrd screws up the store paths or something though) | 18:05:04 |
| ElvishJerricco | no, makeInitrd just does plain ole closures | 18:05:41 |
| ElvishJerricco | it's the extraUtils derivation that does the screwing with binaries so the closure is small | 18:05:58 |
| emily | ah | 18:07:49 |
 Marcel | Is this a crime?
storePaths =
let
closure = pkgs.closureInfo { rootPaths = [ initrdCfg.package ]; };
closurePaths = lib.filter (path: path != "") (lib.splitString "\n" (builtins.readFile "${closure}/store-paths")) ;
in
closurePaths ++
[
config.environment.etc."ifstate/config.initrd.yml".source
];
now i only have to work on reducing the closure size.
| 19:57:37 |
| Marcel | * Is this a crime?
storePaths =
let
closure = pkgs.closureInfo { rootPaths = [ initrdCfg.package ]; };
closurePaths = lib.filter (path: path != "") (lib.splitString "\n" (builtins.readFile "${closure}/store-paths")) ;
in
closurePaths ++
[
config.environment.etc."ifstate/config.initrd.yml".source
];
now i only have to work on reducing the closure size.
| 19:58:08 |
| emily | IFD | 19:58:51 |
| emily | so criminal we have a name and a flag for the specific type of crime :) | 19:59:17 |
| emily | it would break on Hydra | 19:59:19 |
| Marcel | ahh ok | 19:59:56 |
| Marcel | at least my vm test works now ;D | 20:00:47 |
| ElvishJerricco | Marcel: yea if you want to do closureinfo stuff to get full closures into systemd initrd I was thinking of something like this:
{
boot.initrd.systemd.storePaths = [
(runCommand "x"
{
info = closureInfo { rootPaths = [ hello ]; };
}
''
mkdir $out
cat "$info"/store-paths | while read path; do
ln -s "$path" "$out/$(basename "$path")"
done
''
)
];
}
| 20:03:53 |
| ElvishJerricco | * Marcel: yea if you want to do closureinfo stuff to get full closures into systemd initrd I was thinking of something like this:
{
boot.initrd.systemd.storePaths = [
(runCommand "x"
{
info = closureInfo { rootPaths = [ hello ]; };
}
''
mkdir $out
cat "$info"/store-paths | while read path; do
ln -s "$path" "$out/$(basename "$path")"
done
''
)
];
}
| 20:03:57 |
| ElvishJerricco | but it'd probably be better to just have a make-initrd-ng feature for this | 20:04:21 |
