| 1 Feb 2023 |
K900 | No, but I don't think it's a big stretch to assume that the channel version matches the system | 14:12:21 |
Arson Copperflame | In reply to @nzbr:nzbr.de But that'd only give us the channel version, not what's actually installed. nixos-wsl-version should return the version that's currently running. Can a module that comes from a channel know the channel's git rev? Then we could just drop the last part of the version number (for the module at least) and put a version file into the repo that we bump after every release | 14:12:36 |
Arson Copperflame | In reply to @k900:0upti.me No, but I don't think it's a big stretch to assume that the channel version matches the system It is, if the bug that gets reported is related to running nixos-rebuild for example | 14:14:03 |
Sandro 🐧 | In reply to @k900:0upti.me And have a separate stable branch but none us run stable, right? So that would be a bit shooting into the dark. | 14:15:12 |
Arson Copperflame | stable as in nixos-stable or as in a release version of nixos-wsl? | 14:15:55 |
Sandro 🐧 | nixos-stable | 14:16:50 |
K900 | As in nixos-stable, yes | 14:16:55 |
Arson Copperflame | I'm running NixOS 22.11 stable | 14:17:07 |
Arson Copperflame | with a few select packages from unstable | 14:17:22 |
Sandro 🐧 | good to know that, I run nixos-unstable with ~60 patches from all over the place | 14:23:50 |
Arson Copperflame | main is supposed to run on the latest stable nixos version (or the one that's referenced in flake.nix to be exact). That's what the tests check against. Ideally it'll run on unstable as well, but I've never tested for that specifically | 14:24:52 |
Arson Copperflame | nixpkgs seems to add a .version-suffix file to their channel tarballs to give lib access to the commit hash | 15:21:47 |
Arson Copperflame | We could just do that and push to github pages, no changes to how nixos-wsl-version works required | 15:27:19 |
Arson Copperflame | * We could just do that and push the tarballs to github pages, no changes to how nixos-wsl-version works required | 15:27:30 |
Sandro 🐧 | In reply to @nzbr:nzbr.de We could just do that and push the tarballs to github pages, no changes to how nixos-wsl-version works required we want to push tarballs to releases | 15:31:28 |
Sandro 🐧 | putting them in git will only make it slow and the repo big | 15:31:44 |
Arson Copperflame | In reply to @sandro:supersandro.de we want to push tarballs to releases Not the installer tarballs, the channel sources | 15:31:52 |
Sandro 🐧 | In reply to @nzbr:nzbr.de
main is supposed to run on the latest stable nixos version (or the one that's referenced in flake.nix to be exact). That's what the tests check against. Ideally it'll run on unstable as well, but I've never tested for that specifically I don't know, everything I do is based on unstable.. | 15:32:17 |
Arson Copperflame | And from what I understand, the new and improved (tm) github-pages doesn't require pushing to the repo | 15:32:23 |
Sandro 🐧 | didn't read anything about that yet | 15:35:14 |
Arson Copperflame | https://github.com/marketplace/actions/upload-github-pages-artifact | 15:35:48 |
| @plutoisnotblue:matrix.org removed their display name plutoisnotblue. | 23:35:52 |
| @plutoisnotblue:matrix.org left the room. | 23:36:25 |
| 2 Feb 2023 |
Sandro 🐧 |
not contain any symbolic or hard links
Oh no
| 12:33:11 |
K900 | That's fine though | 12:33:59 |
Arson Copperflame | That's fine, because our repo doesn't contain any (I think). The action I linked does --hard-dereference when creating the tarball anyway | 12:34:00 |
K900 | We can just put a tar file in the tar file | 12:34:05 |
Arson Copperflame | And we'd just upload the sources, not the installer tarball | 12:34:24 |
K900 | I've merged the Rust path thing | 12:36:34 |
K900 | After testing it on like five machines | 12:36:41 |