10 Apr 2024 |
adamw | whether the compose is intended to be a 'final' one is a compose-level choice not an image-level one | 19:00:00 |
Micah Abbott | ah, i see | 19:00:00 |
adamw | but, there is. uh. an awful lot of complexity under that file. | 19:00:19 |
adamw | the pungi is connected to the...koji, and the koji is connected to the...osbuild service, and the osbuild service is connected to...god knows | 19:00:51 |
adamw | this whole area is godawful, honestly. for live images we wind up magically not displaying the warnings if the release of fedora-release is 1 or higher. for installer images not built by osbuild we pass --isfinal to pungi somehow, somewhere, but i can't remember how or where | 19:06:44 |
Micah Abbott | you know what we need...another build system 😛 | 19:07:50 |
adamw | ohhh, yeah, no, we don't ever pass --isfinal to pungi i don't think, pungi has stupid magic to pass it to lorax...i hate all of this. | 19:07:55 |
adamw | right. yeah. composes with labels starting RC- , Update- , or SecurityFix- are automatically considered 'supported' at https://pagure.io/pungi/blob/master/f/pungi/compose.py#_378 | 19:09:18 |
adamw | and compose.supported determines how https://pagure.io/pungi/blob/master/f/pungi/phases/ostree_installer.py and https://pagure.io/pungi/blob/master/f/pungi/phases/buildinstall.py#_159 set isfinal | 19:10:51 |
adamw | it's just that simple. sigh | 19:10:56 |
Simon de Vlieger | adamw: I know I haven't spoken about this publically yet but we're working pretty hard to make this much simpler this quarter. | 20:08:19 |
Simon de Vlieger | e.g. not requiring Go PR's and going through 4 projects and a release cycle to production | 20:08:46 |
Simon de Vlieger | where you (and in general distribution maintainers) will have a much simpler stack and have much more control over all of this | 20:10:32 |
Simon de Vlieger | but since i haven't really spoken to all the people who would be touched by this and their requirements yet (including you) we haven't announced anything yet :) | 20:15:19 |
Simon de Vlieger | this is a ... i feel the pain same as you :) | 20:15:56 |
11 Apr 2024 |
coremodule | pwhalen: hey pwhalen, are we going to have another compose once/if we decide to ship 20240410 as the milestone? | 16:22:16 |
coremodule | just wondering if we use that opportunity to fix this: https://pagure.io/fedora-iot/issue/57 | 16:23:34 |
pwhalen | hey coremodule we need to do another "official" compose to change streams, f40 is currently on devel, we need to do another to change to 'stable' which we typically do when we sign off on the release. For the bug, we need to have a new osbuild deployed to fix it :(
| 16:25:21 |
pwhalen | So my hope is we can do that before shipping, but I haven't had any confirmation. | 16:25:35 |
pwhalen | I'm working on some of those missing tests now as well. fwiw | 16:26:21 |
coremodule | awesome, thank you! I will relay that at the go/no-go. Personally I'm fine with that as long as it we can fix it before Tuesday, the testing part for me is OK. | 16:36:25 |
Simon de Vlieger | I have an open PR for this upstream and I'll try to push it through to production tomorrow. | 21:20:05 |
15 Apr 2024 |
adamw | f40 blocker review meeting starting in 5 minutes over in #blocker-review:fedoraproject.org | 15:55:10 |
adamw | hum, zezere test in openqa is failing on rawhide | 18:54:52 |
adamw | https://openqa.fedoraproject.org/tests/2571834 | 18:54:57 |
adamw | it seems like maybe a webui bug but i'm not sure? we click on "Claim Unowned Devices" here - https://openqa.fedoraproject.org/tests/2571834#step/_iot_zezere_remote/6 | 18:55:11 |
adamw | then we wind up at this screen - https://openqa.fedoraproject.org/tests/2571834#step/_iot_zezere_remote/8 | 18:55:21 |
adamw | which...well, the window title is now "Claim unowned devices", and the page has text "Unowned devices:" in it, but the "tab row" still seems to indicate we're on Home | 18:56:03 |
adamw | is the webUI bugged? | 18:56:06 |
adamw | oh, hmm, i see the 'tab bar' has been broken for a while apparently, looking at the reference needle it's trying to match | 18:57:07 |