| 21 Jan 2022 |
 @grahamc:nixos.org | oh cool | 20:28:53 |
 das_j | It actually works | 20:29:10 |
| @grahamc:nixos.org | nice! | 20:29:13 |
| das_j | I hate hashrefs so much | 20:29:14 |
| @grahamc:nixos.org | I mean lists and list flatenning and all of that is generally awful | 20:29:39 |
| das_j | I'm always like %{%whatever and perl is always like "why would you do it this ugly way?" | 20:29:45 |
| @grahamc:nixos.org | it is why I write (and ask for) small functions with tests | 20:29:49 |
| das_j | In reply to @grahamc:nixos.org
it is why I write (and ask for) small functions with tests oh no :D | 20:30:13 |
| das_j | I didn't write one :/ | 20:30:20 |
| @grahamc:nixos.org | well... I can help ... | 20:30:27 |
| das_j | not sure how testable it is… most stuff comes from the framework and the weird plugin | 20:30:44 |
| @grahamc:nixos.org | yeah | 20:30:50 |
| das_j | there's already a PR if you want to take a look at the implementation… I really don't know how I would hook a test framework in it | 20:31:05 |
 ma27 | grahamc (he/him): which LDAP fix? | 21:58:08 |
| ma27 | btw anything missing to merge https://github.com/NixOS/hydra/pull/1091 ? :) | 21:58:29 |
| @grahamc:nixos.org | ma27: https://github.com/NixOS/hydra/pull/1132 | 22:12:55 |
| 22 Jan 2022 |
 nf | is there a way to filter a list of jobs by failure mode? (dependency failed/timeout/etc) | 12:20:06 |
| 23 Jan 2022 |
|  @cgroves:matrix.org joined the room. | 22:09:24 |
| 24 Jan 2022 |
 vcunat | In reply to @ncfavier:matrix.org
is there a way to filter a list of jobs by failure mode? (dependency failed/timeout/etc) No, I don't think so. | 11:31:32 |
| vcunat | There's some JSON API, I think, so perhaps there it might be easy to post-process. | 11:32:43 |
| das_j | You can use the pretty documentation by opening https://petstore.swagger.io and pasting https://raw.githubusercontent.com/NixOS/hydra/master/hydra-api.yaml into the Explore bar on top | 11:34:01 |
| @grahamc:nixos.org | das_j: I'm looking at the runcommand log PR, and I think in general it looks pretty good. Obviously I'm inclined to follow up with some tests, but I'm wondering about the naming of the log files. I'm somewhat inclined to give them either a UUID stored in the DB, or use the runcommandlog record's ID. What do you think? | 14:24:57 |
| @grahamc:nixos.org | I'd want to use that ID for accessing the log's contents instead of passing in a filename to the controller, and also for determining where to write it on disk | 14:25:22 |
| das_j | Yeah that makes a lot more sense, my approach was just the one with the least resistance ;) | 14:25:58 |
| @grahamc:nixos.org | a UUID is nice because it gives something to do the directory sharding by | 14:27:00 |
| @grahamc:nixos.org | but they're not as nice to query by | 14:28:19 |
| @grahamc:nixos.org | I'm not sure we need any security properties here around URL enumeration | 14:28:46 |
| @grahamc:nixos.org | though if we assume postgres >=13 (released in 2020) we get UUID generation in the database for cheap migrations: gen_random_uuid | 14:32:53 |
| das_j | I think it's fair to expect everyone to upgrade to 13. I'm still too lazy for the 14 upgrade : | 14:33:30 |
| das_j | * I think it's fair to expect everyone to upgrade to 13. I'm still too lazy for the 14 upgrade :( | 14:33:32 |
