| 26 Jul 2025 |
 emily | not sure if Lix has any declared tiers | 12:59:11 |
 Qyriad | Most of the time that a Lix build for a platform has broken it has been because of a downstream Nixpkgs breakage | 13:00:28 |
| Qyriad | though I think sometimes that might've been through dependencies CppNix didn't have | 13:01:24 |
 aloisw | There have also been a couple of instances where 32-bit specific bugs have been introduced in Lix. | 13:02:24 |
| emily | building is one thing, running is another :) | 13:03:35 |
| emily | (though at least most 32 bit bugs can be caught statically with sufficiently aggressive checking) | 13:03:56 |
| aloisw | The instances I'm thinking of were actually all build failures. I do not know whether it actually runs. | 13:04:44 |
| aloisw | I could try running a 32-bit Lix on my system, but probably it will OOM regularly. | 13:05:17 |
| emily | In reply to @xokdvium:matrix.org
Well you can build it cross (and hydra does build it). (the thing about building Nix cross is that it means you're building the system cross so you probably have no need for Nix on the system) | 13:05:45 |
| emily | (but I'm all for deconfusing platform stuff) | 13:05:52 |
| emily | In reply to @aloisw:julia0815.de
I could try running a 32-bit Lix on my system, but probably it will OOM regularly. running on an AArch64 box that has 32 bit mode is probably viable | 13:06:29 |
| emily | like, "might even bootstrap a basic NixOS system if you're lucky and send some patches" viable | 13:06:48 |
| aloisw | I'm actually thinking of x86_64 running an x86_32 Lix. | 13:07:05 |
 Sergei Zimmerman (xokdvium) | Not necessarily. For bootstrapping PowerPC OPNA has mentioned first cross-compiling an installer and then moving toward a native build. | 13:07:14 |
| aloisw | nix-repl> pkgs.pkgsi686Linux.lixPackageSets.git.lix
«derivation /nix/store/4h894hnlpdnz841s99lbcy7yggjpkps7-lix-x86_64-unknown-linux-gnu-2.94.0-pre-20250725_57b1c289b592.drv»
Shit, I must be doing something wrong. | 13:09:35 |
| emily | In reply to @xokdvium:matrix.org
Not necessarily. For bootstrapping PowerPC OPNA has mentioned first cross-compiling an installer and then moving toward a native build. right | 13:10:05 |
| emily | with 32 bit it's a lot harder (because bootstrapping on that much RAM and CPU is hard to impossible depending) | 13:10:33 |
| emily | though you can probably get away with -j1 --cores 1 for a good while longer. as long as you don't want a full desktop system | 13:11:04 |
| emily | even the subset of i686-linux we build on fast hardware causes a good bit of grief | 13:11:34 |
| emily | e.g. jq upstream is broken on 32 bit | 13:11:45 |
| emily | In reply to @aloisw:julia0815.de
nix-repl> pkgs.pkgsi686Linux.lixPackageSets.git.lix
«derivation /nix/store/4h894hnlpdnz841s99lbcy7yggjpkps7-lix-x86_64-unknown-linux-gnu-2.94.0-pre-20250725_57b1c289b592.drv»
Shit, I must be doing something wrong. I think that's correct. | 13:12:05 |
| emily | we put the build platform in the derivation names for cross, I think. | 13:12:15 |
| emily | I'm not sure why. also I could be wrong. | 13:12:21 |
| emily | nix derivation show it, perhaps. | 13:12:31 |
| aloisw | In reply to @emilazy:matrix.org
even the subset of i686-linux we build on fast hardware causes a good bit of grief Well I assume "fast hardware" does not matter, it's either 32-bit issue or runtime OOM usually? | 13:12:38 |
| aloisw | In reply to @emilazy:matrix.org
we put the build platform in the derivation names for cross, I think. No, first of all pkgsi686Linux should not be cross and second the "host" (in autotools speak) platform is in the name. | 13:13:23 |
| aloisw | In other words this is a cross build from i686-linux to x86_64-linux. Caused by me setting both localSystem and crossSystem to x86_64-linux and pkgsi686Linux only overriding the former. | 13:14:44 |
| emily | right ok, sorry | 13:17:01 |
| emily | the variant sets override very weirdly | 13:17:11 |
| emily | (especially if you try to compose them) | 13:17:35 |
