| 7 Dec 2025 |
 Gaétan Lepage | Thanks connor (burnt/out) (UTC-8)!
Actually, I was wondering whether the stubs could fix the import at build time... | 11:11:53 |
| Gaétan Lepage | Indeed (I'm quite dumb for not having made the link sooner) this issue is caused by https://github.com/NixOS/nixpkgs/pull/459416/commits/9d38d180dba19909a6058a4e16ffa1803536c4d8.
Is there a way I could disable this for this package? | 11:50:49 |
| Gaétan Lepage | Opened https://github.com/NixOS/nixpkgs/pull/468659. I think it's way better than my previous solution. | 14:24:33 |
 connor (burnt/out) (UTC-8) | Left a comment | 17:11:27 |
| Gaétan Lepage | Thanks a lot. Working on a better solution. | 17:26:06 |
| 8 Dec 2025 |
 adrian-gierakowski | is it expected for cudaSupport=true to cause nix itself to be rebuilt? https://github.com/NixOS/nixpkgs/issues/468947 | 11:22:53 |
| adrian-gierakowski | looks like so tests related deps causing this | 11:23:29 |
| adrian-gierakowski | * looks like so tests related deps causing this:
> nix why-depends /nix/store/rf0rkc1p07vgrc25b6br3f35aipipp3g-nix-2.32.4 /nix/store/awc75hb0iz3gnb4qbg3z1g9lk8jqiibi-hwloc-2.12.2-lib
/nix/store/rf0rkc1p07vgrc25b6br3f35aipipp3g-nix-2.32.4
└───/nix/store/i7chz8mn48rgmm1y8a6yy6gk6bs2dv8m-nix-2.32.4
└───/nix/store/3pvks90shqql6bvb67cc88n21r7b3sx0-nix-util-2.32.4
└───/nix/store/iq2bbr1l3ka2yfjc93gswsp43xr9a6mr-libblake3-1.8.2
└───/nix/store/bk2q02fwz5mxa4xvblcl1p89ygllm04f-onetbb-2022.3.0
└───/nix/store/awc75hb0iz3gnb4qbg3z1g9lk8jqiibi-hwloc-2.12.2-lib
| 11:26:18 |
| adrian-gierakowski | Maybe I shouldn't set cudaSupport = true systemwide, but create a separate instance of pkgs for just those package which need cudaSupport? | 12:51:02 |
 Albert Larsan | It used to rebuild at least as far back as 25.05, and still rebuilds on 25.11.
Some NixOS options change defaults based on the cudaSupport/rocmSupport values. | 12:59:11 |
| adrian-gierakowski | Thanks @albertlarsan68:albertlarsan.fr! So would you still recommend setting cudaSupport at system level? | 14:29:37 |
| Albert Larsan | That is what I do for the machines that have an Nvidia GPU, for which I eat the cost of building nix (although I have a private binary cache to which I push the closures of the updated machines before uploading the flake update to the machines) | 14:36:59 |
| 9 Dec 2025 |
| connor (burnt/out) (UTC-8) | Gaétan Lepage: SomeoneSerge (back on matrix): could one of you approve/merge https://github.com/NixOS/nixpkgs/pull/467975? | 00:09:23 |
 SomeoneSerge (back on matrix) | Robbie Buxton: have you seen https://linus.schreibt.jetzt/posts/ubuntu-images.html? Also stumbled on https://github.com/numtide/nix-vm-test while trying to remember the url | 16:35:16 |
| SomeoneSerge (back on matrix) | Nix in this case is just a tool we already had for memoization and isolation, it just happens to come with a particularly shitty scheduler | 16:36:28 |
 Robbie Buxton | No I haven’t, I’ll take a look, cheers! | 16:37:26 |
| SomeoneSerge (back on matrix) | Any other memoization solution one could build must necessarily be a version of Nix (on top of the actual Nix), likely stripped down and using heuristics instead of trying to faithfully compute the hash "of that and only that which matters"... | 16:38:47 |
| SomeoneSerge (back on matrix) | Which with the current state of affairs is probably cheaper and more efficient 🤷 | 16:39:14 |
| SomeoneSerge (back on matrix) | But mathematically still the same thing | 16:39:27 |
 tomberek | Anyone using impure derivations to explicitly have non-caching behavior? | 16:42:29 |
| SomeoneSerge (back on matrix) | No, that's not really useful? | 17:07:37 |
| 10 Dec 2025 |
| SomeoneSerge (back on matrix) | * No, that's not really useful?
Edit: we generally do want memoization (caching), as well as ability to deterministically name things in sentences "X succeeds (fails)". Just that so far we haven't included all relevant platform info in the derivation, so the equivalence classes are too big. It's one line fixable though
| 18:29:41 |
| SomeoneSerge (back on matrix) | * No, that's not really useful?
Edit: we generally do want memoization (caching), as well as ability to deterministically name things in sentences "X succeeds (fails)". Just that so far we haven't included all relevant platform info in the derivation, so the equivalence classes are too big. It's one line fixable though
Impure derivations just add sandboxing that needs to be bypassed, and a scheduler that is more of a liability
| 18:30:29 |
 Ari Lotter | hm looks like vllm fails to build because outlines has a test that imports libcuda.so on collection.. | 22:03:19 |
| SomeoneSerge (back on matrix) | https://github.com/NixOS/nixpkgs/pull/465751#issuecomment-3604113652 | 23:30:45 |
 apyh | hahaha you're already ahead of me! | 23:31:20 |
| SomeoneSerge (back on matrix) | Gaetan has been on it | 23:32:00 |
| apyh | yeah, seems like there's a clear path to patch / upstream a fix to llama cpp python to make it lazy | 23:33:22 |
| apyh | btw i posted logs about that torch nvrtc thing | 23:33:36 |
| 11 Dec 2025 |
| Gaétan Lepage | Yes, but it's not as straightforward as they initialize some top-level constants at module import time with the content of the loaded library...
I still think that it would be the best solution to this issue. | 09:19:45 |
