| 27 May 2021 |
 _hc | that works | 15:25:01 |
 proletarius101 | In reply to @eighthave:matrix.org
proletarius101: uniq FYI the production buildserver host is Debian/buster on baremetal. The production buildserver guest is stretch. This file should be updated to buster https://gitlab.com/fdroid/fdroid-bootstrap-buildserver/-/blob/master/Vagrantfile ah, right | 15:25:57 |
 uniq | proletarius101 _hc I've published a rebuild of basebox-stretch64 on vagrant image registry and made a MR with the checksums for makebuildserver | 15:29:30 |
| uniq | fdroidserver!928 | 15:29:37 |
| proletarius101 | how about a buster basebox? I'm working on top of the buster basebox | 15:31:44 |
| proletarius101 | of course I would also suggest to run the provision-* scripts in the build step of the base boxes. But that would be far away | 15:32:37 |
| _hc | proletarius101: yeah, that would be a nice thing to also get done in your buster upgrade. uniq and I discussed that before, ideally we could delete provision-apt-get-install. That means removing most of the packages, and putting those into the app's build entries, e.g. in sudo:. Then the hard requirements could be included in basebox | 15:34:35 |
| proletarius101 | In reply to @eighthave:matrix.org
proletarius101: yeah, that would be a nice thing to also get done in your buster upgrade. uniq and I discussed that before, ideally we could delete provision-apt-get-install. That means removing most of the packages, and putting those into the app's build entries, e.g. in sudo:. Then the hard requirements could be included in basebox the provision-apt-get-install script could be kept, because we could run the same script in the ci image. However, we can also do the ansible provision inside the docker image | 15:36:07 |
| _hc | ah right. Yeah, I guess the end goal here is merging basebox, provision-apt-get-install, and ci-images-base into one thing | 15:37:01 |
| uniq | I think the other way around would be better. having all provisions in fdroidserver | 15:37:37 |
| uniq | because we then could reuse that eg for gitlab ci | 15:37:53 |
| _hc | then remove basebox? Or just keep it as minimal as possible? | 15:38:04 |
| _hc | I guess we could move to the official Debian basebox | 15:38:14 |
| uniq | keep basebox as minimal as possible | 15:38:16 |
| _hc | makes sense to me | 15:38:23 |
| uniq | official boxes would be nice | 15:39:06 |
| uniq | they're still limited to 20gb | 15:39:16 |
| uniq | so our stack won't fit :/ | 15:39:34 |
| _hc | oh right... then we'd have to figure out how to extend the filesystem in provisioning | 15:39:39 |
| _hc | once https://gitlab.com/fdroid/fdroidserver/-/merge_requests/927 is merged, it might fit in 20GB | 15:40:00 |
| uniq | yeah! that would be amazing | 15:40:13 |
| uniq | I already looked into increasing diskimage size a while back and I though boostrapping custom images is easier hence I started that basebox project to begin with | 15:41:12 |
| _hc | yup its 18GB after 927 | 15:41:46 |
| _hc | doesn't leave much room for a build, but that might make provisioning easier | 15:42:12 |
| _hc | helps to stop installing all gradle and NDKs | 15:42:28 |
| uniq | Oh, about official vagrant boxes there's another thing I probably should note. We'd need to use contrib builds (eg. debian/contrib-buster64) for virtualbox to get stable support for virtualbox file sync | 15:47:56 |
| _hc | uniq: izzy toot please: @CalyxOS is now available for the #Android #emulator, providing truly #FreeSoftware system images. https://calyxos.org/news/2021/05/27/emulator-images They are included in the F-Droid #Docker CI image "registry.gitlab.com/fdroid/ci-images-client" | 16:53:23 |
 izzy | On it. | 16:55:19 |
| izzy | Missing part: @CalyxOS@??? (I cannot find a corresponding account at Mastodon) | 16:56:25 |
| izzy | Looks like they're only on Twitter. | 16:57:27 |
