| 14 Jan 2025 |
 hexa | * Gaétan Lepage: load 66 on the darwin builder with 10 cores | 17:30:53 |
| hexa | 4 → /tmp/nix-build-python3.12-optimistix-0.0.10.drv-0
4 → /tmp/nix-build-python3.12-torch-geometric-2.6.1.drv-0
3 → /tmp/nix-build-python3.12-torchsde-0.2.6.drv-0
1 → /tmp/nix-build-gdal-3.10.1.drv-0
3 → /tmp/nix-build-python3.12-captum-0.7.0.drv-0
15 → /tmp/nix-build-python3.12-tensordict-0.5.0.drv-0
27 → /tmp/nix-build-python3.12-torch-2.5.1.drv-2
28 → /tmp/nix-build-python3.13-torch-2.5.1.drv-3
27 → /tmp/nix-build-python3.13-torch-2.5.1.drv-1
14 → /tmp/__KMP_REGISTERED_LIB_68029
27 → /tmp/nix-build-python3.12-torch-2.5.1.drv-3
| 17:32:14 |
| hexa | that is not a realistic workload for this machine | 17:32:31 |
 Gaétan Lepage | Killed it | 18:02:57 |
| Gaétan Lepage | torch depends on google-auth-oauthlib` ??? | 18:03:34 |
| Gaétan Lepage | * torch depends on google-auth-oauthlib ??? | 18:03:49 |
 liberodark | You are the darwin killer ^^ | 18:21:42 |
| 15 Jan 2025 |
| hexa | tensorboard. | 01:03:22 |
| hexa | while waiting for the build environment for '/nix/store/ji27lc8qdxvk4cxrgdvyxm8pyl3psy77-servo-0-unstable-2025-01-14.drv' to initialize (failed with exit code 1, previous messages: sandbox initialization failed: data object length 65601 exceeds maximum (65535)|failed to configure sandbox: data object length 65601 exceeds maximum (65535))
| 01:10:57 |
| hexa | okay.gif | 01:11:09 |
 emily | known limitation of the sandboxing API that AFAIK nobody knows a good way to work around (other than making it less granular) :( | 01:12:22 |
| hexa | --option sandbox false/relaxed yield the same issue | 01:13:11 |
| hexa | * --option sandbox false/relaxed yield the same issue | 01:13:17 |
| emily | weird. they definitely shouldn't | 01:14:29 |
| emily | if it's nixpkgs-review you need --build-args or something | 01:14:42 |
| hexa | no, plain nix-build | 01:14:53 |
| hexa | https://github.com/NixOS/nixpkgs/pull/373903 | 01:16:34 |
| emily | ++ lib.optionals stdenv.hostPlatform.isDarwin [ darwin.apple_sdk.frameworks.AppKit ]; can be dropped entirely fwiw (it has no effect on the build) | 01:17:36 |
| emily | not sure why you'd get sandbox size issues even with the sandbox off though | 01:17:43 |
| hexa | are you getting it as well? 😄 | 01:20:37 |
| hexa | ok, so --option sandbox does not seem to propagate to the remote builder | 01:24:59 |
| hexa | it works when passing it on the remote builder itself | 01:25:12 |
| emily | ah, I was busy downloading a bunch of crates to test :) | 01:30:16 |
| emily | it might be a ? option you can use on the remote builder URL | 01:30:23 |
| emily | dunno, I don't have the builders set as remote builders since I don't trust them | 01:30:34 |
| hexa | I use the comunity builders through shell aliases 🙂 | 01:53:33 |
 Perchun Pak [don't ping; dm instead] | if I recognize this issue correctly, it happens to be on huge symlinkJoins in nixpkgs-review, and they work around it by using pkgs.buildEnv
https://github.com/Mic92/nixpkgs-review/blob/b3ccb2adef4b173a1f81fb258492a8a185a224ad/nixpkgs_review/nix/review-shell.nix#L40 | 10:20:02 |
| Perchun Pak [don't ping; dm instead] | not sure what is causing this issue on your package | 10:20:18 |
| Perchun Pak [don't ping; dm instead] | * not sure what is causing this issue in your package | 10:20:23 |
|  Paul Meyer (katexochen) joined the room. | 10:23:07 |
