| 15 Nov 2021 |
 moritz.hedtke | How reproducible is e.g. the KDE installation image? Is there any data on that? | 17:31:13 |
 @rick:matrix.ciphernetics.nl | Nope | 17:54:46 |
| @rick:matrix.ciphernetics.nl | (i think?) | 17:54:55 |
| @rick:matrix.ciphernetics.nl | There was a job for the gnome_iso, but it kept running out of space | 17:55:16 |
| 16 Nov 2021 |
 siraben | In reply to @toonn:matrix.org
Pretty cool. Is the next step basing stdenvs on Mes? : ) Yes please 🙏 | 02:45:51 |
| siraben | I'd love to look into that if anyone would like to collaborate, especially if they're familiar with how stdenv works | 02:46:30 |
 @grahamc:nixos.org | it'd be cool if someone extended the r13y website generator to make an index of various builds so we could highlight the reproducibility of the gnome iso as well as the minimal ISO on the homepage | 14:30:52 |
 tomberek | ref: https://r13y.com/iso_gnome/index.html any other overall builds that would be good to add? | 15:40:01 |
|  @etrigan63:matrix.org changed their profile picture. | 19:24:28 |
| 17 Nov 2021 |
 afontain | eh, /nix/store/0mn8d6cl3pv7py19q0k3lxnysvsvh5yi-attachment.cgi?id=359589.drv | 09:39:57 |
| afontain | (from https://r13y.com/iso_gnome/index.html) | 09:40:32 |
| moritz.hedtke | https://github.com/NixOS/nixpkgs/blob/a0dbe47318bbab7559ffbfa7c4872a517833409f/pkgs/development/libraries/cogl/default.nix | 11:59:00 |
| moritz.hedtke | Don't know why they specifically are unchecked paths | 11:59:14 |
 @ninjatrappeur:alternativebit.fr | In reply to@afontain:gnugen.ch
(from https://r13y.com/iso_gnome/index.html) I was so confused to see pandoc landing on the minimal image. It's actually the gnome iso. We probably should change the page title, it's misleading. | 12:15:12 |
| @ninjatrappeur:alternativebit.fr | The binaries generated by GHC are currently not binary-stable. More info on https://gitlab.haskell.org/ghc/ghc/-/issues/12935 | 12:16:42 |
| @ninjatrappeur:alternativebit.fr | * The binaries generated by GHC are currently not binary-stable. More info on https://gitlab.haskell.org/ghc/ghc/-/issues/12935 .
TL;DR: the unique temporary names generated during compilation are not necessarily stable. Looking at diffoscope, we're exactly hitting that issue
Meaning don't expect | 12:17:52 |
| @ninjatrappeur:alternativebit.fr | In reply to@afontain:gnugen.ch
(from https://r13y.com/iso_gnome/index.html) * I was so confused to see pandoc landing on the minimal image. It's actually the gnome iso. We probably should change the page title, it's misleading.
[edit]: We're already aware of that, I should have backlogged the chat a bit more. Sorry about that. | 12:19:13 |
 j-k |
In openSUSE, we also made it deterministic using non-parallel builds for now
| 13:16:02 |
| j-k | although compiling pandoc with 1 thread doesnt sound fun | 13:16:23 |
|  towesfeb joined the room. | 21:06:02 |
| 18 Nov 2021 |
 r-burns | In reply to @siraben:matrix.org
I'd love to look into that if anyone would like to collaborate, especially if they're familiar with how stdenv works im very interested in this, would love to collab :) | 18:45:26 |
| 19 Nov 2021 |
 baloo | work just had a problem with file ordering, and two builder not behaving the same. | 23:52:16 |
| baloo | ended a python import loop that only happened if one file was loaded before the other | 23:52:37 |
| baloo | but ... I guess disorderfs could find a couple of those | 23:52:50 |
| baloo | a doing import b; def bar(): b.foo() where b was doing from a import bar; def foo(): | 23:54:53 |
| baloo | because the checkPhase will load every file to look for unittest.TestCase, it will load every file, depending on the order they're loaded, it will fail | 23:55:53 |
| baloo | * ended up being a python import loop that only happened if one file was loaded before the other | 23:56:36 |
| 20 Nov 2021 |
 Sandro 🐧 | Same filesystem? | 02:27:30 |
|  nykw joined the room. | 11:05:37 |
| baloo | yup | 23:22:44 |
