| 10 Jan 2023 |
|  Winter (she/her) joined the room. | 04:58:13 |
| Winter (she/her) | I know r13y is out of date, but does anyone know what was causing the two issues there (cpython and cbindgen), assuming they're fixed that is? The Python one just looks like it's swapping some ds for fs in the hexdump in the resulting pycs somehow (even though the ASCII characters, when applicable, stayed the same...?) while the cbindgen one looks a bit more complex (reordering of .text)? | 05:01:04 |
| Winter (she/her) | * I know r13y is out of date, but does anyone know what was causing the two issues there (cpython and cbindgen), assuming they're fixed that is? The Python one just looks like it's swapping some ds for fs in the hexdump in the resulting pycs somehow (even though the ASCII characters, when applicable, stayed the same...?) while the cbindgen one looks a bit more complex (reordering of .text). | 05:01:11 |
| Winter (she/her) | * I know r13y is out of date, but does anyone know what was causing the two issues there (cpython and cbindgen), assuming they're fixed that is? The Python one just looks like it's swapping some ds for fs in the hexdump in the resulting pycs somehow (even though the ASCII characters, when applicable, stayed the same...?) while the cbindgen one looks a bit more complex (reordering of .text). | 05:01:23 |
 raboof | a more recent report is at https://arnout.engelen.eu/nixos-iso-minimal-r13y/ | 07:10:38 |
| raboof | I can't find the issue for cbindgen right now (on mobile) but I don't think I've seen that for a while, that might be fixed. For python, I think there are some issues that are fixed in python 3.11 (update expected after 23.05), but there might be remaining problems | 07:15:51 |
| raboof | ah, the cbindgen issue is https://github.com/NixOS/nixpkgs/issues/188988 - we haven't really explained that one, so that might still be in there, but it's hard to trigger intentionally | 07:47:52 |
| raboof | python is tracked in https://github.com/NixOS/nixpkgs/issues/188978 | 07:51:43 |
 Artturin | https://github.com/NixOS/nixpkgs/pull/210102 | 20:59:55 |
| 11 Jan 2023 |
|  afontain left the room. | 19:44:28 |
| 12 Jan 2023 |
|  srid changed their profile picture. | 06:36:34 |
| 14 Jan 2023 |
| srid changed their profile picture. | 02:09:35 |
| 15 Jan 2023 |
|  @motiejus:jakstys.lt joined the room. | 09:33:06 |
| @motiejus:jakstys.lt left the room. | 09:51:07 |
|  @moke:matrix.creeper-it.de joined the room. | 13:56:49 |
| 18 Jan 2023 |
| raboof | ran the iso-gnome report again: https://arnout.engelen.eu/nixos-iso-gnome-r13y/ . nice batch of things to look into. (note that this is the build closure, so it contains not only the things that are in the gnome iso, but also the things that are needed to build the gnome iso) | 08:59:43 |
| 19 Jan 2023 |
|  Salfurium joined the room. | 22:32:21 |
| 20 Jan 2023 |
 tpw_rules | https://github.com/NixOS/nixpkgs/blob/7cf8f9bcee843560e454d66392c3e28ad850c893/nixos/modules/virtualisation/qemu-vm.nix#L228 this is not at all reproducible despite wanting to be. e.g. how the disk cache is flushed and systemd writing random seeds to the image | 05:55:29 |
| tpw_rules | (mostly putting it here so i don't forget about it, don't know when i'll have time to address it if someone else wishes to) | 05:56:08 |
 @rick:matrix.ciphernetics.nl | I suggest opening an issue on nixpkgs to keep track of long-term things :) | 06:58:10 |
 atemu12 | I also don't think that's a very high priority, is it? It's the VM for tests, it doesn't need to be bit-perfect. | 08:43:26 |
| raboof | In reply to @raboof:matrix.org
ran the iso-gnome report again: https://arnout.engelen.eu/nixos-iso-gnome-r13y/ . nice batch of things to look into. (note that this is the build closure, so it contains not only the things that are in the gnome iso, but also the things that are needed to build the gnome iso) hmm, a lot of them are duplicates of https://github.com/NixOS/nixpkgs/issues/151347 . I wonder if there's any easy way to convert derivation paths back to nixpkgs attribute paths (for those that have one), should probably be doable | 11:59:38 |
| raboof | In reply to @raboof:matrix.org
ran the iso-gnome report again: https://arnout.engelen.eu/nixos-iso-gnome-r13y/ . nice batch of things to look into. (note that this is the build closure, so it contains not only the things that are in the gnome iso, but also the things that are needed to build the gnome iso) * hmm, a lot of them are instances of https://github.com/NixOS/nixpkgs/issues/151347 . I wonder if there's any easy way to convert derivation paths back to nixpkgs attribute paths (for those that have one), should probably be doable | 11:59:59 |
| atemu12 | raboof: Not that I'm aware of. | 12:41:06 |
| atemu12 | You could take a look at the drvs and filter out everything that has GHC in it/uses it to compile | 12:41:31 |
 @trofi:matrix.org | If you have a nixpkgs tree it should be easy. nix-env -qaP will print you both .drv and attribute. | 13:11:54 |
| raboof | Neat! Had to add --drv-path to get the full path. Unfortunately it doesn't recurse into haskellPackages... | 13:53:36 |
| @trofi:matrix.org | yeah, to fix the recursion you need to pass packages-config.nix that exposes hidden attributes: pkgs/top-level/make-tarball.nix: nix-env -f . -I nixpkgs=$src -qa --meta --json --arg config 'import ${./packages-config.nix}' "''${opts[@]}" >> tmp | 13:59:18 |
| @trofi:matrix.org | https://github.com/NixOS/nixpkgs/blob/master/pkgs/top-level/make-tarball.nix#L67 | 14:00:04 |
| @trofi:matrix.org | I use a similar hack locally to map .drv files back to attribute names and to repology package names: https://github.com/trofi/nix-olde/ | 14:01:19 |
