| 8 May 2022 |
|  Gaelan Steele joined the room. | 22:56:44 |
 tomberek | Janne Heß: I added a bit. There is another thing, which is to try and get the eval failures down too. | 23:19:37 |
| 9 May 2022 |
 Vladimír Čunát | In reply to @janne.hess:helsinki-systems.de
I will be posting it tomorrow morning my time (~10h from now). Vladimír Čunát is the staging cycle done by then? staging-next is merged | 05:34:59 |
| Vladimír Čunát | In reply to @janne.hess:helsinki-systems.de
I will be posting it tomorrow morning my time (~10h from now). Vladimír Čunát is the staging cycle done by then? * staging-next is merged now | 05:35:05 |
 Janne Heß | Alright, starting ZHF in a couple of minutes ;) | 09:00:51 |
| Vladimír Čunát | Evals are fresh as well now, only for trunk-combined it's catching up caching builds from staging-next. | 09:04:19 |
| Vladimír Čunát | * Evals are fresh as well now, only for trunk-combined it's still catching up caching builds from staging-next. | 09:04:25 |
| Janne Heß | I'm still waiting for zh.fail to update since there seems to be a new fully finished eval | 09:04:42 |
| Janne Heß | ah I had a discussion with h'exa (not pinging) about the detection of failed builds on the page. Do you think we should mark build with a status of Cancelled as non-failed? | 09:05:58 |
| Vladimír Čunát | I'd treat it the same as queued. | 09:06:41 |
| Vladimír Čunát | A state somewhere between failure and success. | 09:07:05 |
| Janne Heß | Hm, does that make the graph go up or not? :D | 09:07:56 |
| Janne Heß | I currently skip all evals which have queued != 0 | 09:08:17 |
| Vladimír Čunát | Hehe... you might estimate from the last status of the same job (skipping over queued and cancelled). | 09:08:39 |
| Vladimír Čunát | But I understand if that's considered too complex to do now. | 09:08:55 |
| Janne Heß | And potentiall to heavy for Hydra :/ If you cancel a whole eval, I'd have to crawl every build twice (or three times?) (the aborted build and the "last previous success/fail") | 09:09:43 |
| Vladimír Čunát | Right, if there are too many of cancelled/queued, you probably want to skip that eval anyway. | 09:15:06 |
| Vladimír Čunát | * Right, if there are too many of cancelled+queued, you probably want to skip that eval anyway (for now). | 09:15:20 |
| Vladimír Čunát | But often it doesn't seem optimal to wait because of a couple straggling builds. | 09:18:20 |
| Vladimír Čunát | * But often it doesn't seem optimal to wait because of a couple straggler builds. | 09:18:55 |
| Vladimír Čunát | * But often it doesn't seem optimal to wait just because of a couple straggler builds. | 09:20:04 |
| Janne Heß | So it begins | 10:04:09 |
| Vladimír Čunát | https://github.com/NixOS/nixpkgs/issues/172160 | 10:05:28 |
| Janne Heß | ah I should cross-post this to discourse | 10:05:55 |
|  Robert Hensing (roberth) joined the room. | 10:25:45 |
| Janne Heß | So Robert Hensing (roberth) grahamc (he/him): I wanted to finish up the discussion about the initial tag for the upcoming release. Graham said there might be tools that rely on the tag being named that way, do you think its enough to announce the upcoming change now and hope everybody relying on this fixes their scripts? I really understand Robert's motivation here and I think it would be a relevant improvement and make this more clear | 12:38:38 |
 jonringer | In reply to @janne.hess:helsinki-systems.de
So Robert Hensing (roberth) grahamc (he/him): I wanted to finish up the discussion about the initial tag for the upcoming release. Graham said there might be tools that rely on the tag being named that way, do you think its enough to announce the upcoming change now and hope everybody relying on this fixes their scripts? I really understand Robert's motivation here and I think it would be a relevant improvement and make this more clear The main thing is git describe | 16:13:16 |
| jonringer | [09:14:21] jon@nixos ~/projects/nixpkgs (release-21.11)
$ git describe
21.11-3617-g1f8e77dfaea
| 16:14:35 |
| jonringer | actually, for hydra this might not be too much of a concern:
version = fileContents ../.version;
versionSuffix =
(if stableBranch then "." else "beta") + "${toString (nixpkgs.revCount - 291486)}.${nixpkgs.shortRev}";
And this gets updated as part of the release process
| 16:17:25 |
| Janne Heß | couldn't we just … not tag at all? | 16:19:55 |
