| 28 Feb 2022 |
 toonn | This is an evaluation that I want to compare to staging-next rather than previous evals of the jobset, https://hydra.nixos.org/jobset/nixpkgs/toonn-wip | 15:12:07 |
| toonn | Ah, darn, I meant to link a specific eval of that, just pick the most recent one. | 15:12:28 |
 @grahamc:nixos.org | cool, I think I see one of the problems | 15:14:19 |
| @grahamc:nixos.org | you click the compare link and it takes you to https://hydra.nixos.org/eval/1746179?compare=staging-next&full=0 which is fixed in time | 15:14:31 |
| @grahamc:nixos.org | and so you have to do that clicky-around process to make this comparison | 15:14:47 |
| @grahamc:nixos.org | what if you could get a link that was like: https://hydra.nixos.org/jobset/nixpkgs/toonn-wip/latest-eval?compare=staging-next | 15:15:09 |
| toonn | grahamc: The link doesn't quite work for me. I already do that. But when I focus on something else for a while or something the urls in the history are close to useless because I don't know eval IDs by heart. That's why I'd love to have a way to set a little bit of state for an evaluation so that future me or others I link the eval to always end up with a useful comparison. | 15:20:00 |
| @grahamc:nixos.org | yeah | 15:20:20 |
| @grahamc:nixos.org | I understand, but I don't mean remember eval IDs -- I mean that URL would always be fresh with the most recent eval | 15:20:47 |
| @grahamc:nixos.org | I'm hesitant to add a "default comparison" field to the database because I think that risks getting very very complicated | 15:21:08 |
| @grahamc:nixos.org | so I wonder if there is a way to make some stable URL which gets you what you're looking for every time | 15:21:30 |
| toonn | That's probably a useful feature but not what I'm after. | 15:21:26 |
| toonn | I want to be able to visit past evals as well and get the intended comparisons. | 15:22:01 |
| @grahamc:nixos.org | oh I see | 15:22:14 |
| @grahamc:nixos.org | can you write up a detailed ticket and open an issue? | 15:22:27 |
| @grahamc:nixos.org | I think this could easily make sense in a PR model | 15:22:50 |
| toonn | Hmm, I think we could do this stateless if there was an endpoint that served a comparison to the closest branch or something. So if I base something on a staging-next commit I get a comparison to a staging-next eval of that commit. | 15:23:43 |
| toonn | Not every commit has an eval of course. Not sure how to work around that. Eval with fewest additional commits on top could be surprising. | 15:24:41 |
 hexa | I wish we had a more up to date hydra on unstable by now | 16:28:25 |
| hexa | but all PRs seem to be stalled | 16:28:45 |
| toonn | git.io stopped? : / | 18:05:48 |
| toonn | grahamc: I just opened https://github.com/NixOS/hydra/issues/1168 for this feature request. Let me know if there's anything I can clear up further. | 18:08:57 |
 cole-h | In reply to @toonn:matrix.org
git.io stopped? : / Hm, shame. | 18:21:23 |
| toonn | cole-h: https://github.blog/changelog/2022-01-11-git-io-no-longer-accepts-new-urls/ | 18:23:21 |
| toonn | Yeah, it really is a shame. The whole point of git.io was the *lack* of features, which made it trustworthy by design. No more "Hmm 🤔, if I click this shortened URL, might I end up somewhere dodgy?" | 18:24:32 |
| toonn | I guess the lack of brand recognition made them prefer the full URLs. | 18:25:36 |
| @grahamc:nixos.org | thanks! | 19:35:40 |
| toonn | np | 20:31:35 |
|  @chris:mkaito.net joined the room. | 23:24:10 |
| @chris:mkaito.net changed their display name from chris to mkaito. | 23:30:46 |
