| 25 Aug 2023 |
 Julien | I want to check for the reproducibility of a lot of derivations and a would like the builds to be performed on remote machines | 09:30:52 |
| Julien | So far it seems to me that Nix is not allowing to do that: if I do nix-build --check it will build locally and if I do nix-build --check --max-jobs 0 it will use my remote builders but do not tell me anything about the reproducibility of the derivation. I suppose some of you here may have tried to distribute the jobs of checking for the reproducibility of a lot of derivations, do you have any feedback on this kind of question ? | 09:32:34 |
| Julien | I think my ideal workflow would be that the remote builder does the build step and then the coordinating nix daemon could do the comparison, but it looks like it is not working this way. | 09:34:13 |
 raboof | That sounds reasonable to me - I've never tried checking reproducibility using remote builders myself | 12:37:35 |
|  @rimuru:gentoo.chat changed their display name from rimuru to rimuru (moved to @kaya:catnip.ee). | 22:00:22 |
| @rimuru:gentoo.chat left the room. | 22:27:24 |
| 26 Aug 2023 |
|  @henrytill:matrix.org joined the room. | 23:18:19 |
| 27 Aug 2023 |
|  @uuuvn:dendrite.fail changed their display name from uuuvn to uuuvn 📎. | 15:31:17 |
|  lord_fomo joined the room. | 21:21:00 |
| 28 Aug 2023 |
 @trofi:matrix.org | I usually do Content Addressed build with a non-material change in stdenv to see if both machines produce the same hashes for larger build trees :) | 06:40:27 |
| Julien | Interesting trick, thanks | 06:42:13 |
| 30 Aug 2023 |
 j-k | https://go.dev/blog/rebuild | 08:17:25 |
| 31 Aug 2023 |
|  henrirosten joined the room. | 06:05:06 |
|  IndustrialRobot changed their display name from IndustrialRobot (she/her) to IndustrialRobot. | 10:50:52 |
|  moritz.hedtke removed their display name moritz.hedtke. | 16:13:19 |
|  Philip Taron (UTC-8) joined the room. | 21:46:27 |
| 1 Sep 2023 |
|  t4ccer joined the room. | 01:45:48 |
| 2 Sep 2023 |
|  prescientmoon joined the room. | 00:02:07 |
| 3 Sep 2023 |
|  @ek_is_anika:tchncs.de left the room. | 19:04:35 |
| 4 Sep 2023 |
|  mib 🥐 joined the room. | 08:19:57 |
| @trofi:matrix.org | Test for you intuition: does gcc build depend on filesystem type of /nix/store? And subsquently, could it fail to build on fancy ones? :) | 15:07:32 |
 raitobezarius | Trivially yes | 15:08:30 |
| raitobezarius | Case sensitiveness :p | 15:08:39 |
| raboof | this is a quantum question: after reading it my intuition changed 😆 | 15:09:03 |
| @trofi:matrix.org | Support /tmp is case ensitine and nix/store is not. Will it still fail? :) | 15:09:34 |
 atemu12 | trofi: I'd say it depends on both /nix store and build dir | 15:09:41 |
| @trofi:matrix.org | * Suppose /tmp is case ensitine and nix/store is not. Will it still fail? :) | 15:09:50 |
| @trofi:matrix.org | Yeah, apparently every time you fiddle with $out in your phases you probe the /nix/store. Spoiler: https://github.com/NixOS/nixpkgs/issues/253274 | 15:10:21 |
| atemu12 | Though most commonly used fs should be fine | 15:10:25 |
| atemu12 | Dang it 32b | 15:10:48 |
