| 16 Aug 2025 |
 emily | that links to things like the Buildkite login | 14:13:20 |
| emily | so that the flow can be log in → see what you can get access to | 14:13:25 |
| emily | rather than trying to go backwards from Buildkite to however you get it to redirect you to Lix SSO | 14:13:41 |
 helle (just a stray cat girl) | potentially or at least make that the quicklinks page if we cannot get it out of the SSO that info | 14:13:55 |
| emily | I just assumed I would not have access to the Buildkite at all | 14:13:56 |
| emily | yeah I was just thinking that having a big login link on the homepage that takes you to a landing page on success would be a natural way to expose it | 14:14:20 |
| emily | since realistically if it was buried in dev docs I wouldn't have found it there either :p | 14:14:34 |
| helle (just a stray cat girl) | mhm, well, not a huge fan of a login link big on the front page for a project where to use you do not actually need to log in | 14:14:56 |
| emily | fair | 14:15:21 |
| emily | but where else are you going to hook up the Lix Pro billing flow | 14:15:58 |
| helle (just a stray cat girl) | you are describing exactly WHY I do not like have those logins that prominent | 14:17:00 |
| helle (just a stray cat girl) | :P | 14:17:03 |
| emily | the free tier is very generous – a whole 100 derivation builds per month (and those don't even count successful cache hits! you get 500 of those) | 14:17:25 |
| emily | more seriously, perhaps just adding quick links to the Gerrit top menu may be easy? | 14:18:02 |
 aloisw | Not to my knowledge, they are supposed to send it to you by email (that's where I got it from) but maybe that actually requires signing in once so that they can't be abused as a spam source. | 14:26:27 |
 niko ⚡️ | Okay now I'm back to square one - the failure just not being there on older nixpkgs. So naturally I'm bisecting nixpkgs now to see what changed | 21:36:44 |
| 17 Aug 2025 |
| niko ⚡️ | I had to take a break and I'm back to bisecting since now that I'm close to the offending change I have to skip 90% of the commits unless I wanted to rebuild the whole world. Does anyone have a script or something to only bisect over merge commits or only bisect over commits that had been built by hydra? That'd be invaluable as it always looks like this for me whenever I'm trying to bisect nixpkgs 😭 | 09:04:45 |
| aloisw | --first-parent is the usual recommendation to avoid going into staging. | 09:09:10 |
| niko ⚡️ | Yep that made it way better, thanks. Not sure why nobody mentions this option when googling for "bisect merge commits only" or whatever | 10:31:18 |
| niko ⚡️ | Of course the regression is at a staging merge that's so awesome, but at least I have my suspects now, let's see if I'm correct | 10:53:43 |
| niko ⚡️ | One of the regressions started happening after bumping bash, I don't know if I'm gonna update my config this weekend. I still have two other ones to identify | 11:34:03 |
| niko ⚡️ | Yeah so bash 5.3 breaks functional-user-envs, now I need to figure out why functional-remote-store and functional-nix-shell-basic started failing after that staging merge | 11:50:06 |
| helle (just a stray cat girl) | would it be in your skillset to move these to functional2? (this may depend on some f2 features not quite finished) | 12:30:22 |
| helle (just a stray cat girl) | else you can probably file a bug and poke me or Rutile (Commentator2.0) feel free to ping to actually do the move | 12:30:44 |
| helle (just a stray cat girl) | no use maintaining f1 if we have this sort of breakage | 12:30:55 |
| niko ⚡️ | Sure, I'll try. Also the other failure is from bumping sqlite or one of its dependencies it seems like | 12:33:21 |
| niko ⚡️ | Overriding sqlite from 3.50.2 to 3.50.1 resolved the other two failures | 12:33:54 |
| niko ⚡️ | Which is kinda scary because I can't see why it'd be sqlite at fault | 12:34:49 |
| niko ⚡️ | Someone more knowledgable and with more time than me should look into this I have no clue if it's a genuine regression or yet another false-negative because of auto-allocate-uids | 12:49:17 |
| niko ⚡️ | * Someone more knowledgeable and with more time than me should look into this I have no clue if it's a genuine regression or yet another false-negative because of auto-allocate-uids | 12:49:25 |
