| 21 Feb 2021 |
 jochensp | cdesai has set up a build server already, we only need someone to implement a secure way to transfer the apks to Ciaran for signing | 21:26:06 |
| jochensp | or we implement HSM support and someone gets those keys | 21:26:25 |
| jochensp | or we implement reproducible builds and add more signatures | 21:26:49 |
| jochensp | there are so many ideas, but we need someone to implement them | 21:27:05 |
 @freenode_FstplttnSchntzl:matrix.org | Maybe toot a sentence on Mastodon, if someone would be interested? | 21:28:12 |
 @SylvieLorxu:matrix.org | rsync? | 21:28:32 |
| jochensp | Sylvia: rsync could be part of the solution, sure. We need someone to implement and test it | 21:30:44 |
| @SylvieLorxu:matrix.org | Where are the unsigned APKs currently stored? Are checksums also available? | 21:31:48 |
| jochensp | Sylvia: which apks exactly? | 21:33:21 |
| @SylvieLorxu:matrix.org | The unsigned APKs of the apps that are build that Ciaran needs to sign. I would expect they are already stored in some directory or something? And a (preferably signed) checksum file that we can use to determine that the transfer was 100% correctly done when we do it? | 21:34:13 |
| jochensp | not sure what you are talking about, I would propose you install a build server for yourself and start with that | 21:36:01 |
| @SylvieLorxu:matrix.org | There are several issues that make it impossible to run a buildserver on Fedora currently otherwise I would have done so months ago already. The Docker image isn't working either so... yeah | 21:37:09 |
| @SylvieLorxu:matrix.org | I can't run a build server locally, I wish I could | 21:37:21 |
| jochensp | you could install a Debian VM | 21:37:30 |
| jochensp | ..or send MRs to make it work on Fedora :) | 21:37:49 |
| @SylvieLorxu:matrix.org | I am very willing to help make it run on Fedora (although I would rather focus on the Docker image to make it run everywhere) but I don't know the software well enough to be able to myself | 21:38:28 |
| @SylvieLorxu:matrix.org | And I consistently am low on RAM so running a full VM is problematic on my system too | 21:38:52 |
| jochensp | fixing Docker is fine as well | 21:39:24 |
| @SylvieLorxu:matrix.org | I have opened GitLab issues that show my problem but as I don't know fdroidserver well I didn't manage to fix those myself | 21:39:57 |
| @SylvieLorxu:matrix.org | And nobody responded so... yeah | 21:40:04 |
| jochensp | as I said, we are lacking contributors | 21:40:17 |
| @SylvieLorxu:matrix.org | In this specific case, yes | 21:40:33 |
 @freenode_izzy:matrix.org | Is there some specific reason behind our "quoting rules" in YAML? One day I go nuts. For example, "commit: <hash>" (no quotes), but "commit: '1.2'" (needs quotes), and then "commit: 1.2.3" (no quotes again). Similar for versionName. Both are always strings, so why are quotes needed at all? | 22:30:45 |
| @SylvieLorxu:matrix.org | I looked into the code a bit and there's actually 2 stages | 22:36:47 |
| @SylvieLorxu:matrix.org | Stage one is running yamllint to check if it is valid yaml | 22:37:05 |
| @SylvieLorxu:matrix.org | Stage two is running fdroid rewritemeta and seeing if the output would be the exact same | 22:37:21 |
| @SylvieLorxu:matrix.org | rewritemeta uses ruamel.yaml to write YAML | 22:37:39 |
| @SylvieLorxu:matrix.org | So it is really really strict | 22:37:50 |
| @SylvieLorxu:matrix.org | Even if your YAML is completely valid, if ruamel.yaml happens to write it down differently (probably optimized to output the smallest possible valid YAML) lint will fail | 22:38:27 |
| @freenode_izzy:matrix.org | So it's actually the Ruamel rewrite messing that up? Guess that would be hard to fix then :( Confuses the heck out of a lot of people. | 22:39:53 |
