| 8 Nov 2018 |
 Marcus (old) | Anyone knows why this is still failing? https://f-droid.org/wiki/page/org.jwz.xscreensaver/lastbuild_54000 | 14:31:45 |
| Marcus (old) | it worked locally for me at some point | 14:32:14 |
| Marcus (old) | locally meaning inside a buildserver as well | 14:32:22 |
 Mathijs | In reply to @mathijs:matrix.vgorcum.com
though we could also use the debian repo of matrix.org actually, using the matrix.org packages is probably smart | 14:32:35 |
 Coffee | In reply to @eighthave:matrix.org
oh, meeting time today also Oh, I thought we were already in the meeting 😁 | 14:42:33 |
| Coffee | matrix.org has an ansible setup too, right? | 14:43:15 |
| Mathijs | slavi has an extensive ansible setup | 14:44:12 |
| Mathijs | but putting containers in VMs seems a bit redundant to me | 14:44:30 |
| Mathijs | I think matrix.org has an ansible setup as well, but I'm not sure how well that works | 14:45:11 |
 Nico |
<@eighthave:matrix.org> oh, meeting time today also
Oh, I thought we were already in the meeting 😁
Me too :P
<@mathijs:matrix.vgorcum.com> though we could also use the debian repo of matrix.org
actually, using the matrix.org packages is probably smart
Yes, that's how I do it. And how I would do it here. But if we find something cool with Ansible, we can consider this too. | 15:05:39 |
| Mathijs | downside of the package is that is uses python 2 for now | 15:06:08 |
| Mathijs | and python 3 helps a bit in speed and ram usage | 15:06:37 |
| Nico | But it's planned to switch to Python 3 in future, right? I don't think we'll have problems with it at the beginning. | 15:12:01 |
| Mathijs | yeah, they're obviously planning on it | 15:12:59 |
| Coffee | I have anoher topic -
I just finished writing an index diff compressor, for archives of the index. | 15:16:35 |
| Coffee | I think it would be nice if we could run that in a small VM and publicly host an index archive. | 15:17:46 |
| Coffee | And maybe diffs will eventually be something for the repo itself. | 15:18:15 |
| Coffee | So the client doesn't have to download the entire index over a metered connection. | 15:18:43 |
| Nico | Good idea for hosting an index archive!
Regarding using diffs in the client, I remember _hc being against it because it would give servers more information about users, i.e. having privacy implications. For the same reason the client downloads the index with all languages and then cuts off the ones not needed. | 15:24:55 |
| Marcus (old) | Coffee: link? | 15:24:59 |
| Marcus (old) | And more technical details. | 15:25:19 |
| Marcus (old) | :-) | 15:25:21 |
| Coffee | I'll upload it soon. | 15:25:42 |
| Coffee | It's a bit of a hack right now - I'm abusing the fact that the index contains no newline characters at all. | 15:26:32 |
| Coffee | Which causes pain for most normal diff tools. | 15:27:01 |
| Coffee | So I replace all spaces with newlines, append an extra newline, and then generate ed diffs between that. | 15:27:28 |
| Coffee | The ed script is then recompressed with the original jar manifest files. | 15:28:05 |
| Coffee | It won't verify in that form, but if you perform the reverse operation, it will verify again. | 15:28:36 |
| Coffee | I could just go with a binary differ, but I thought this might be easier to eventually implement in the client. | 15:31:18 |
| Mathijs | In reply to @nico:dorfbrunnen.eu
But it's planned to switch to Python 3 in future, right? I don't think we'll have problems with it at the beginning. they they'll switch to py3 'soon' (that is at version 0.34) | 15:32:31 |
