17:59:45 <danielsh> #startmeeting Reproducible Builds Meeting #8
17:59:45 <MeetBot> Meeting started Tue Jan 31 17:59:45 2017 UTC.  The chair is danielsh. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:59:45 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:59:45 <danielsh> #topic Introductions
17:59:50 <danielsh> Who's here?
18:00:04 * h10ger Holger Levsen
18:00:06 <infinity0> hi
18:00:15 * deki waves
18:00:17 <danielsh> h10ger, you seem to have had a bit flip...
18:00:23 <h10ger> :)
18:00:39 <danielsh> We have apologies from emaste, lamby, and possibly mapreri.
18:00:41 <h10ger> h01ger cant talk atm…
18:01:08 <brett> I'm here, if multitasking, hi all.
18:01:22 <danielsh> hi brett , you're first on the agenda
18:01:22 <Diziet> Hi.  Ian Jackson.  I'm sort of here.
18:01:32 <danielsh> welcome, Diziet
18:01:35 <danielsh> let's go
18:01:36 <danielsh> #topic Action items from last meeting
18:01:39 <danielsh> > ACTION: brett to add "reporting bugs" (or similar…) to --help output of diffoscope
18:01:47 <danielsh> brett, status? ^^
18:01:54 <brett> It's done, let me pull up the bug number.
18:02:11 <brett> #852015
18:02:18 <brett> Already merged with master and released in diffoscope 70.
18:02:25 <h10ger> nice
18:02:31 <danielsh> Let's move on then
18:02:34 <danielsh> > ACTION: lamby to publish generic tools somewhere (future-proof URL) and to update release docs
18:02:46 * vagrantc waves
18:02:47 <danielsh> We have a proxied comment here: <mapreri> - tools should be in the main https://reproducible-builds.org website (or at least, this one should link directly to them, no need to host them in it)
18:02:48 <h10ger> hey vagrant. u missed the intro only
18:03:09 <danielsh> Mattia and Chris aren't here; does anyone else have comments here?
18:04:02 <danielsh> #action link to http://reproducible.alioth.debian.org/releases/ from reproducible-builds.org
18:04:09 <danielsh> > ACTION: infinity0 write code for SOURCE_PREFIX_MAP alternatives, send data to thread
18:04:31 <infinity0> i have a draft spec almost ready too
18:04:33 <danielsh> That was about comparing three alternatives that were considered at the time; the thread has moved past that
18:04:41 <h10ger> infinity0, cool
18:05:24 <danielsh> Anything else ?
18:05:42 <danielsh> #topic Upcoming events
18:05:49 <danielsh> h01ger, you mentioned fosdem ?
18:06:08 <bmwiedemann> hi.
18:06:12 <h10ger> yeah, well… (do we have a past events topic too?)
18:06:21 <danielsh> bmwiedemann, hello, notice #action above
18:06:27 <h10ger> for fosdem i was going to suggest an informal meetup somewhere
18:06:37 <danielsh> h10ger, we can discuss past events too
18:06:39 <h10ger> bmwiedemann, backlog is at http://meetbot.debian.net
18:06:43 <h10ger> +hi
18:06:57 <h10ger> danielsh, please do #save
18:07:01 <danielsh> #save
18:07:06 <h10ger> thx
18:07:15 <infinity0> i'll be at fosdem too
18:07:19 <danielsh> (bmwiedemann has been on channel all along, though)
18:07:19 <h10ger> not sure when+where to best meet at fosdem
18:07:43 <h10ger> without knowing the schedule, i'd say 1400 on saturday
18:07:50 * bmwiedemann will not be at fosdem for family reasons
18:08:32 <h10ger> place wise i would suggest the cafetaria close to the K building
18:08:47 * h10ger will send a mail about this to the list…
18:09:09 <danielsh> h10ger, past events, then ?
18:09:13 <infinity0> sounds good
18:09:14 <h10ger> i was thinking just of an informal gathering, to say hi, to connect, to discuss stuff
18:09:41 <danielsh> h10ger, what about outside people ?
18:09:47 <h10ger> who besides infinity0 will be there?
18:09:56 <h10ger> danielsh, outside people cannot join RL meetings
18:09:57 <danielsh> e.g., hold some BoF of reproducible builds to people who aren't on this meeting right now
18:10:32 <h10ger> thats why i'll send an email
18:10:53 <bmwiedemann> or give a 5 minute lightning talk about rb
18:11:02 <brett> (Also upcoming, a little further out, is LibrePlanet.)
18:11:19 <h10ger> i think at libreplanet there will be 1-2 talks
18:11:30 <brett> At least one, yeah.
18:11:35 <h10ger> spectranaut, vagrantc and/or lamby, iirc
18:11:47 * h10ger is not sitting at his usual computer as you might have figured :)
18:12:09 <brett> One of our blogs mentioned Vagrant's talk at least.
18:12:26 <vagrantc> yeah, i've got a talk there
18:12:48 <vagrantc> also have talks at scale15x and embedded linux conference
18:12:58 <vagrantc> dunno if anyone else will be around for those
18:13:16 <danielsh> #info Upcoming events: FOSDEM, LibrePlanet, scale15x, embedded linux conference
18:13:55 <h10ger> past events?
18:13:55 <danielsh> h10ger, you mentioned something about past events ?
18:13:59 <danielsh> go
18:15:03 <danielsh> h10ger, did you have anything to say about past events
18:15:04 <danielsh> ?
18:15:06 <h10ger> last week was devconf.cz where dgilmore and myself held a talk on reproducible builds and fedora, which IMHO was pretty well received. in talks after the presentation i also gathering the impression that basically everyone (i talked to) was expecting reproducible fedora to be a good thing and to happen eventually
18:15:44 <h10ger> there's a video on youtube, a link to that and the slides i've put in the 93 issue of our weekly news
18:15:58 <bmwiedemann> https://www.youtube.com/watch?v=d8oEg0p6QLE
18:16:14 <h10ger> bmwiedemann also made a pull request for SOURCE_DATE_EPOCH in rpm
18:16:20 <danielsh> #link https://www.youtube.com/watch?v=d8oEg0p6QLE devconf.cz talk last week by h01ger and dgilmore
18:16:39 <danielsh> (that's not "Past events", but okay)
18:16:50 <h10ger> (one of the rpm upstreams attended the talk too and afterwards said he would look into that patch)
18:17:13 <h10ger> EOT from me, though feel free to ask
18:17:15 <bmwiedemann> #link https://github.com/rpm-software-management/rpm/pull/141
18:17:57 <danielsh> No questions ?
18:18:06 <danielsh> #topic Any other business
18:18:14 <danielsh> For good measure let me repeat what has just been said:
18:18:15 <bmwiedemann> when that one is merged I only have one rpm patch left to cleanup and nicify
18:18:34 <danielsh> #link https://github.com/rpm-software-management/rpm/pull/141 pull request for SOURCE_DATE_EPOCH by bmwiedemann
18:18:34 <danielsh> Anything else ?
18:18:35 <h10ger> bmwiedemann, neato.
18:18:43 <h10ger> bmwiedemann, why not make a pull request for that too?
18:19:09 <danielsh> bmwiedemann, are there other rpm changes that you don't have WIP patches for yet?
18:19:27 <danielsh> Or will that second patch be the last change we want in rpm, for the time being?
18:19:32 <bmwiedemann> would like some feedback for improvement on that one first
18:19:54 <h10ger> rpm still need patches to mock/koji/zypper for .buildinfo support… (mock+koji is what fedora uses, zypper is the suse buildtool, AIUI)
18:20:19 <danielsh> zypper is to suse what apt is to debian, AIUI
18:20:26 <h10ger> koji+zypper are more or less what sbuild/pbuilder is in debian
18:20:46 <bmwiedemann> we have some equivalent to buildinfo already called buildenv and that is already supported to setup an identical build env with "osc build"
18:20:48 <h10ger> danielsh, i thought that would be yum^wdnf
18:21:12 <danielsh> h10ger, does suse even _have_ yum/dnf?  I thought that was in fedora but not in suse
18:21:18 <bmwiedemann> just no archive of rpms for the rolling release distro yet
18:21:24 * h10ger lets bmwiedemann explain
18:21:33 <bmwiedemann> only zypper+rpm in suse
18:21:44 <h10ger> so zypper is like yum/dnf?
18:21:47 <bmwiedemann> yes
18:21:51 <h10ger> k
18:22:02 <bmwiedemann> at some time we even had working zypper builds for fedora
18:22:09 <h10ger> got that (slightly) wrong in the devconf.cz slides…
18:22:46 <bmwiedemann> and there is a zypper-aptitude compat script written by me :-)
18:22:54 <danielsh> bmwiedemann, why have a separate 'buildenv' format?  I thought buildinfo was supposed to be more or less portable ?
18:23:03 <h10ger> bmwiedemann, ouch ;-)
18:23:06 <danielsh> Not with the same details on every distro, but at least the same syntax
18:23:07 <bmwiedemann> was prior art
18:23:19 * danielsh nods
18:23:45 <h10ger> bmwiedemann, do you have a wiki page or somesuch describing the status of reproducible openSUSE in one place?
18:24:13 <bmwiedemann> not yet. I'll have to write up one, but we had certain troubles with the opensuse wiki
18:24:39 <h10ger> oh, btw, today we also got the news that "heads" is now reproducibly build. heads is a tails-like payload for coreboot, IOW "tails in bios"
18:24:49 <h10ger> (link in weekly reports 93 too)
18:25:04 <h10ger> bmwiedemann, ack.
18:25:47 <danielsh> Anything else on the fedora/opensuse discussion?
18:25:54 <danielsh> Or any new topics to raise?
18:26:04 <h10ger> we have this problem with binNMUs
18:26:31 <danielsh> listening
18:27:04 <h10ger> which create files which have the same timestamp and filename as in previous uploads… which confuses rsync backups and the like. i need to reply to aurel32 to explain this to him
18:27:06 <bmwiedemann> binary Non-Maintainer-Uploads ?
18:27:32 <danielsh> bmwiedemann, yes - binary rebuilds of the same source
18:27:33 <h10ger> bmwiedemann, sort of. rebuilds of exiting source packages triggered by "anyone"
18:28:05 <h10ger> the effect can be seen on the buildd network as they dont run the latest sbuid version, as josch explained on the list
18:28:21 <danielsh> bmwiedemann, the rebuild could use newer dependencies, though
18:28:30 <h10ger> i think once sbuild is fixed we should involve the release team to get that on their radar
18:28:45 <h10ger> danielsh, thats exactly the point, usually :)
18:28:59 * danielsh nods
18:29:08 <h10ger> EOT for me, see thread on the (debian) list for more details
18:29:20 <h10ger> (feel free to ask though… :)
18:29:48 <danielsh> #info debian binnmu timestamp issue - see reproducible-builds@ debian list
18:30:15 <h10ger> (the reason for that binNMU bug is that debian/changelog doesnt get modified, thus the rebuilds are done with the original SOURCE_DATE_EPOCH value, thus the same timestamps…)
18:30:45 <danielsh> h10ger, I don't have anything intelligent to add here...
18:31:03 <danielsh> Just whether we should heads up to the release team even ahead of a fixed sbuild, but I assume you've considered that
18:31:27 <h10ger> sbuild is fixed, the fix just aint deployed on the buildd network
18:31:59 <h10ger> reaching out to the release team now means they are aware earlier of the issue which is good
18:32:30 <bmwiedemann> usually more (useful directed) communication is better
18:32:33 <danielsh> shall I #action you for this ?
18:33:08 <h10ger> yup
18:33:11 <danielsh> #nick h01ger
18:33:24 <danielsh> #action h01ger followup binnmu timestamp issue with aurel32 and/or release team heads up
18:33:29 <h10ger> :)
18:33:34 <danielsh> So I think I can reach for my gavel now.
18:33:39 <danielsh> 45s for objections ...
18:34:20 <danielsh> Thanks all.
18:34:21 <danielsh> #endmeeting