| 26 Jun 2023 |
 @bbigras:matrix.org | any ideas what this is about? I just ran a successful build 30 minutes ago.
'nix flake metadata' returned exit code 1:
error: cannot open connection to remote store 'daemon': error: reading from file: Connection reset by peer
(use '--show-trace' to show detailed location information)
| 17:08:10 |
| @bbigras:matrix.org | Adding hydra to trusted-users might have fixed it. not sure. hydra-queue-runner was already trusted. | 17:14:38 |
| @bbigras:matrix.org | ah I still have it. it seems to be intermittent. | 18:58:16 |
|  @janik0:matrix.org changed their profile picture. | 21:38:24 |
| 27 Jun 2023 |
|  cafkafk joined the room. | 10:11:16 |
| 28 Jun 2023 |
|  @attelis:matrix.org left the room. | 13:30:13 |
|  feamor joined the room. | 17:41:43 |
| 29 Jun 2023 |
 Sandro | When I have a package I want to test on hydra and create one gcroot for every step in between so that build progress gets cached even when the package fails, how would I do that? | 12:14:35 |
 das_j | I don't think that'S supported | 12:17:21 |
| Sandro | Anyone knows if any other CI solution has something like that? | 12:29:11 |
| Sandro | I mean I could adopt releases-small or so and plot all my packages into that 🤔 | 12:29:31 |
 toonn | I'm not sure if there is anything. But splitting a single derivation into multiple would work everywhere. I'm not entirely sure why Nixpkgs errs on the side of more monolithic derivations. | 12:32:29 |
 cransom | if you were super big on always adding gcroots, you could add a postbuildhook that would add a root for every build you had.
postBuildHook = pkgs.writeScript "nix-copy-paths" ''
export IFS=' '
mkdir -p /nix/var/nix/gcroots/nixcache
for path in $OUT_PATHS; do
nix-store --add-root /nix/var/nix/gcroots/nixcache/$(basename $path) -r $path
done
'';
| 15:48:02 |
| cransom | that's a nix postbuildhook, apart from hydra. | 15:48:15 |
| 30 Jun 2023 |
| Sandro | I would need to clean them up manually, right? | 01:48:04 |
|  fabaff joined the room. | 14:22:47 |
| 1 Jul 2023 |
|  @lotte:chir.rs changed their profile picture. | 09:47:24 |
| 3 Jul 2023 |
|  Nate joined the room. | 18:01:50 |
| 4 Jul 2023 |
|  @tomh:riotchat.de left the room. | 14:26:16 |
| 5 Jul 2023 |
|  @rimuru:gentoo.chat changed their profile picture. | 14:20:06 |
|  AtnNn joined the room. | 18:19:58 |
| 6 Jul 2023 |
|  Hayley joined the room. | 20:16:47 |
| 7 Jul 2023 |
| @lotte:chir.rs changed their profile picture. | 05:36:35 |
| 8 Jul 2023 |
|  d3routr joined the room. | 13:59:18 |
|  @easel:matrix.org joined the room. | 16:30:14 |
| 9 Jul 2023 |
|  hacker1024 joined the room. | 10:53:38 |
| hacker1024 | Hi all, I am trying to make a declarative project. I am trying to wrap my head around the declInput argument. Why do so many projects cat it? | 10:57:41 |
| 10 Jul 2023 |
| das_j | In reply to @hacker1024:matrix.org
Hi all, I am trying to make a declarative project. I am trying to wrap my head around the declInput argument. Why do so many projects cat it? Depends on what your declInput is. For me it's usually a git repo so it's not cat-table | 06:58:58 |
| das_j | declInput is what you configure in the "Declarative input type" in the project settings | 06:59:27 |
| hacker1024 | Thanks. But what is the purpose of printing it at build time? Take [this one](https://github.com/dhess/pinpon/tree/f4f984f3a45e4239914c912f489fd37110b65d45/nix/jobsets), for example - the input in spec.json is just the repository. | 07:25:21 |
