15:03:29 <rclobus> #startmeeting
15:03:29 <MeetBot> Meeting started Tue Apr 25 15:03:29 2023 UTC.  The chair is rclobus. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:03:29 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:03:45 <rclobus> #topic Who is here today?
15:04:14 * rclobus is Roland Clobus, working on the live images for Debian
15:04:15 <lamby> hey hey
15:04:37 * lamby is Chris Lamb, working on diffoscope, strip-nondetermism (sometimes) and general reproducibility in Debian etc. etc.
15:05:52 <rclobus> Let's wait until 15:10Z to see who else chimes in, otherwise this will be a really small meeting :-)
15:05:55 * vagrantc ~= Vagrant Cascadian working on Debian and mostly the build-essential related packages lately
15:07:02 <rclobus> #info Agenda is here: https://pad.riseup.net/p/rb-irc-meetings-keep
15:08:39 <rclobus> It is now 15:10Z
15:09:15 <rclobus> Shall we do a short round instead, where each of us gets the opportunity to say something?
15:09:38 <vagrantc> sounds like a plan :)
15:09:48 <rclobus> #topic vagrantc has the stage
15:11:13 <rclobus> vagrantc: Do you have something to report for the meeting?
15:11:19 <vagrantc> i have mostly been poking at gcc, binutils, etc. in debian ... and eyeing up the fact that they appear to be unreproducible in debian but not necessarily for other distros
15:12:17 <vagrantc> with gcc, the main troublesome issue is gnu modula 2, which is an optional add-on to gcc ... and i proposed fixes for some timestamps ... but there are very pernicious other timestamps
15:12:41 <vagrantc> i suspect other distros do not embed gm2 (gnu modula 2) ...
15:13:01 <vagrantc> holger pointed out that modula 2 is a decades old language :)
15:13:52 <vagrantc> #link https://tests.reproducible-builds.org/debian/bookworm/amd64/pkg_set_build-essential.html
15:13:55 <vagrantc> so close!
15:14:22 <vagrantc> got some patches to improve the situation for linux and binutils, too ...
15:14:44 <vagrantc> should probably write up a more detailed post to rb-general ...
15:14:47 <vagrantc> that's about it :)
15:15:24 <rclobus> Thanks
15:15:32 <rclobus> #topic lamby has the stage
15:16:10 <lamby> Gosh, I can't think of much to say now.
15:16:24 <lamby> Although today I've been trying to get my head around all of these Sphinx changes
15:16:38 <lamby> I've also been working on a weird website bug, but that's not very interesting
15:17:06 <lamby> I've also been looking at some new scholarly papers that have been published which touch on reproducible builds
15:17:21 <Foxboron> They are public, want me to link?
15:17:41 <lamby> Actually the one I /want/ to link have asked us to not link it publically yet. :)
15:17:51 <lamby> (as it's in pre-print or similar)
15:17:57 <Foxboron> Uhh, From Fourne?
15:18:19 <lamby> Right, I think so.
15:18:34 <lamby> Either way, it will go in the monthly report if/when.
15:18:41 <Foxboron> Then they are publicly actually publicly available :p
15:19:25 <lamby> My plans for the rest of this week are to jump back into diffoscope; got a few easy wins I can tick off there.
15:19:30 <lamby> Anyway, that's enough for now.
15:19:30 <lamby> .
15:19:51 <rclobus> Thanks
15:20:00 <rclobus> #topic Foxboron has the stage
15:20:16 <Foxboron> Yo :) I have also been reading the mentioned paper
15:20:35 <Foxboron> I have not done any reproducible builds work lately in Arch, and I don't think we have any updates there
15:22:11 <Foxboron> .
15:22:14 <Foxboron> Thanks :)
15:22:22 <rclobus> Thank you
15:22:31 <rclobus> #topic rclobus takes the stage
15:23:02 <rclobus> For the Debian live images: they are generated using Debian hardware and are semi-reproducible
15:23:22 <rclobus> As mentioned last month, they are reproducible within the 6 hours after generating them
15:23:47 <rclobus> I haven't found the time yet to see what can be done about that, because I got distracted...
15:24:17 <rclobus> The machine on Jenkins which builds the live images and tests them for reproducibility has issues,
15:24:31 <rclobus> the upload speed is horribly low.
15:24:48 <rclobus> h01ger is in contact with osuosl to see what can be done about it.
15:25:31 <rclobus> I've also worked on getting all live images (both from Jenkins and get.debian.org) tested better, but...
15:25:36 <vagrantc> the 6 hour window is so that the builds happen with the same archive (e.g. before updates are pushed) ?
15:25:51 <rclobus> openQA is also running on that machine, so all workers are disabled for now :-(
15:26:06 <rclobus> vagrantc: Indeed, within the same DAK run.
15:27:02 <vagrantc> so, also fixable by a reliable snapshot server, i'm guessing
15:27:38 <rclobus> vagrantc: Yes, when the snapshot server is also providing non-free-firmware, I can use that again.
15:28:03 <vagrantc> i've often wondered about starting two builds in parallel, rather than in sequence to limit those sorts of issues as well
15:28:27 <rclobus> I guess it is the typical: the closer a Debian release is, the more hardware changes etc are happening :-)
15:28:32 <vagrantc> heh
15:29:13 <rclobus> vagrantc: In Jenkins I start 2 builds after each other, and in worst case need to discard both of them due to an intervening DAK run.
15:29:25 <rclobus> Overall that rarely happens.
15:30:20 <rclobus> But still, as far as I can see: the live images are still reproducible, but slightly hard to get a 1:1 identical copy for the average user.
15:30:45 <rclobus> That's it from my side, regarding the reproducible state of the live images.
15:31:16 <rclobus> #topic Any other business
15:31:48 <rclobus> Let's advertise the event in May:
15:31:51 <rclobus> #info https://wiki.debian.org/DebianEvents/de/2023/DebianReunionHamburg
15:33:13 <rclobus> Any other topics?
15:34:30 <rclobus> Then... Thank you of showing up
15:34:57 <rclobus> #endmeeting