17:58:08 <spwhitton> #startmeeting
17:58:08 <MeetBot> Meeting started Tue Aug  8 17:58:08 2023 UTC.  The chair is spwhitton. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:58:08 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:58:13 <spwhitton> #topic Roll Call
17:58:14 <spwhitton> Sean Whitton
17:58:17 <Emperor> Matthew Vernon
17:58:20 <roehling> Timo Röhling
17:58:20 <Myon> Christoph Berg
17:58:25 <tumbleweed> Stefano Rivera
17:59:07 <spwhitton> Don't think we received any apologies.  Let's leave it 5m.
17:59:29 <helmut> Helmut Grohne
18:00:00 <spwhitton> smcv: ping
18:00:03 <spwhitton> mjg59: ping
18:02:23 <mjg59> Matthew Garrett
18:03:02 <spwhitton> #topic Review of previous meeting's AIs
18:03:11 <spwhitton> Emperor: you filed your bug, thank you -- any follow-up needed?
18:03:15 <spwhitton> I asusme not, just asking to be sure.
18:03:35 <Emperor> I don't think so; I don't know if the release-notes will take the suggestion or not.
18:03:53 <spwhitton> okay.  well it was just an aid, not a necessity.  thanks again.
18:03:58 <spwhitton> I updated the slides.
18:04:02 <Emperor> For the minutes, the bug is #1041072
18:04:05 <spwhitton> tumbleweed: how are we doing with all being co-presernters on the session?
18:04:20 <spwhitton> #link https://bugs.debian.org/1041072
18:04:26 <Myon> I'll not be there, missed the registration deadline :-/
18:04:29 * smcv is here, sorry I'm late
18:04:33 <spwhitton> hello smcv
18:04:47 <spwhitton> to be clear, ideally we are all co-presernters even if we'r enot attending
18:04:55 <Myon> ok with me
18:04:56 <Emperor> spwhitton: I got an email about the talk, which suggests the system thinks I'm a co-presenter
18:05:05 <tumbleweed> spwhitton: checking...
18:05:30 <helmut> registered speakers: Stefano Rivera & Sean Whitton & Matthew Vernon & Timo Röhling & Helmut Grohne
18:05:58 <tumbleweed> mjg59, Myon: Please sign up on https://debconf23.debconf.org/ and I can add you
18:06:01 <spwhitton> okay, can smcv, mjg59 and Myon go and SSO login there
18:06:12 <spwhitton> just logging in ought to be sufficient.
18:06:13 <tumbleweed> smcv is added
18:06:17 <mjg59> tumbleweed: On it
18:06:25 <Myon> I'm logged in
18:06:33 <spwhitton> #topic Any remaining DebConf23 presentation arrangements
18:06:42 <spwhitton> we haven't talked much about this
18:06:46 <mjg59> tumbleweed: Ok should be there
18:06:49 <spwhitton> but, it's a BoF, so that's not in itself a problem
18:06:56 <tumbleweed> everybody added, thanks
18:07:00 <spwhitton> nice, thanks.
18:07:04 <Myon> ta
18:07:06 <spwhitton> tumbleweed: have you had a chance to look at the slides?
18:07:23 <tumbleweed> spwhitton: had not, will do
18:07:27 <Myon> "WARNING: Talk proposal submitted, but speaker hasn't registered to attend. Register now!"
18:07:39 <tumbleweed> :)
18:07:45 <spwhitton> we don't actually have another meeting before debconf
18:08:01 <spwhitton> so ... any remaining tasks?
18:08:13 <spwhitton> tumbleweed: do you feel ready? :)
18:08:30 <tumbleweed> sure
18:09:23 <spwhitton> okay then.  well, post mail or here on IRC if you find any issues after reviewing the slides.
18:09:31 <helmut> we have a meeting during debconf, no? do we delay or reschedule that?
18:09:47 <spwhitton> good question.  thoughts?
18:09:54 <tumbleweed> OK, skimmed them
18:10:11 <Myon> you'll likely have better things to do there than talking to me via IRC
18:10:18 <tumbleweed> that seems straightforward enough. I'll do some skimming and thinking about anything else to cover
18:10:29 <tumbleweed> (skimming the related bugs I mean)
18:10:42 <spwhitton> tumbleweed: cool.  it is down as a BoF, so, presentation should be ~10m or so
18:10:43 <Myon> so move (or skip?) it
18:10:54 <tumbleweed> spwhitton: right
18:10:58 <spwhitton> I suggest we skip it unless we get bugs filed, in which case we move it
18:11:05 <helmut> +1
18:11:06 <Myon> ack
18:11:16 <spwhitton> thank you again Stefano.
18:11:20 <spwhitton> #topic merged-/usr
18:11:37 <spwhitton> helmut: I think things are quiet but maybe you have things to discuss?
18:11:53 <helmut> not much has happened here. it mostly seems like a one-man show at this time and I had a nice vacation :)
18:12:07 <helmut> one thing seems worthy reporting though.
18:12:32 <helmut> if you move an empty directory from / to /usr as part of a package upgrade (i.e. no move between packages), it gets lost reliably
18:12:52 <smcv> (I have a thing to add when helmut is finished, if he doesn't cover it)
18:13:34 <helmut> https://subdivi.de/~helmut/dumat.yaml has been improved a bit and is updated 4 times a day. it now associates bugs usertagged fileconflict for instance
18:13:57 <helmut> I've also spent quite some time filing bugs (for file conflicts and for empty directories that should be deleted)
18:14:04 <helmut> and that's about it. smcv?
18:14:12 <spwhitton> helmut: have you been getting constructive responses to your filings?
18:14:49 <helmut> not universally, but mostly. some people argued about severities when they couldn't reproduce my unpack errors
18:15:06 <spwhitton> well, they're mainly for tracking so I guess severities don't matter too much?
18:15:20 <helmut> I tend to send file conflict bugs without patches at rc severity
18:15:30 <helmut> that tends to annoy some people
18:15:41 <spwhitton> yeah.  well good that most responses have been good.
18:15:54 <spwhitton> smcv: please go ahead
18:16:07 <smcv> helmut: I'm still seeing unsupported upgrades listed in that file, like bullseye to unstable. would it be possible to have a filtered output that only covers the supported/supportable scenarios?
18:16:42 <helmut> smcv: I consider that details. can we defer that outside the meeting rather than taking everyones time for this aspect?
18:16:52 <smcv> of course
18:16:57 <smcv> so the thing I wanted to mention
18:17:02 <helmut> smcv: do you have time after the meeting?
18:17:38 <helmut> smcv: do note though that "bullseye to unstable" often is a version range
18:17:39 <smcv> I got a MR merged into debootstrap to teach it to /usr-merge all suites >= trixie even for the buildd profile, as we discussed before
18:17:59 <spwhitton> coolio.  a stable update is required to switch over the buildds, right?
18:18:31 <smcv> which Luca NMU'd, and promptly had hostile-NMU'd away by a developer who no longer has upload rights as a result
18:18:39 <spwhitton> oh, right,
18:18:45 <spwhitton> where were things left?
18:18:45 <helmut> spwhitton: do note that DSA does not upgrade to bookworm yet, so we need an oldstable update
18:18:56 <smcv> things were left with Luca's NMU being reinstated
18:18:58 <spwhitton> helmut: ah good point.
18:19:24 <spwhitton> Hideki has traditionally been the debootstrap maintainer, but he seems inactive
18:19:29 <smcv> so now we have merged-/usr for sufficiently new suites' buildds, but only if either DSA upgrades to a suitable version, or "larger" infra replicates the change
18:19:37 <helmut> to be precise, luca quickly reverted the revert
18:19:42 <smcv> right
18:20:02 <spwhitton> smcv: do you think something should happen imminently?  have you asked DSA if they have plans to upgrade?
18:20:40 <helmut> From my pov, we can leave things as is for now. We want buildds merged once we lift the moratorium
18:20:41 <smcv> I have not, but I'm getting concerned that trixie was meant to be the point at which maintainers can start relying on merged-/usr, and that kinda conflicts with official buildds not having it
18:21:13 <spwhitton> smcv: could we action you to find out DSA's plans, at least?
18:21:20 <spwhitton> and, do you agree with what helmut just said?
18:21:22 <helmut> ok, rough sketch of my plans may help here maybe
18:21:35 <smcv> "we can rely on merged-/usr now[1]"    [1] but if it makes a package FTBFS then it's still RC
18:21:41 <smcv> seems not entirely helpful
18:21:44 <helmut> at this time, I'm ironing out issues in the detector and filing patches for actionable things
18:22:16 <helmut> I also need to update dep17 with the consensus achieved thus far and propose a concrete combination of mitigations
18:22:47 <helmut> then, I'll arm dumat to automatically file rc bugs and once that has happend and we have /usr-merged buildds, we can lift the moratorium and mass migrate files
18:23:31 <helmut> that bug filing part totally is not implemented at all yet and I want to be really careful not to file lots of bogus rc bugs
18:23:36 <spwhitton> helmut: is there any reason for smcv not to pursue the buildd chang ein the meantime?
18:24:31 <helmut> spwhitton: there just isn't that much urgency yet, but I'm ok with moving forward on that side
18:24:33 <smcv> it's not just buildds I'm concerned about, but also autopkgtest, piuparts, all similar things
18:24:54 <spwhitton> if smcv has the time to work on it, then doing it in parallel seems like a good idea
18:24:59 <spwhitton> to avoid random hold ups later
18:25:00 <helmut> +1
18:25:18 <smcv> last time we said "too soon", do we have consensus that it is not too soon now?
18:25:52 <helmut> I think I said too soon, because we didn't have consensus on the primary mitigation. We do have that consensus now.
18:26:26 <smcv> oh yeah you weren't sure whether you needed to change the bootstrapping protocol, which would have also required a stable update
18:26:31 <Myon> is that mitigation summarized anywhere? (sorry I didn't follow closer)
18:26:38 <helmut> My last consensus call said that we do not want to rely on dpkg changes as the primary vehicle for finishing the transition and that generally resulted in agreement
18:26:54 <Myon> nod
18:26:59 <spwhitton> yes.
18:26:59 <helmut> smcv: Can we piggy back my debootstrap change on the stable update?
18:27:28 <helmut> https://salsa.debian.org/installer-team/debootstrap/-/merge_requests/96
18:27:44 <helmut> With current consensus, we also need that MR in stable
18:28:39 <smcv> and oldstable if DSA haven't upgraded yet?
18:28:53 <helmut> Unsure about process here. If Hideki is MIA here, should I merge and NMU it?
18:28:56 <helmut> smcv: yes
18:29:26 <helmut> so yeah, can we get that change into unstable and backport both changes at once rather than having to pester SRM four times?
18:29:28 <spwhitton> helmut: just use DELAYED etc. in the usual way I think
18:29:46 <spwhitton> shall I action you to do that?
18:30:04 <helmut> I'd appreciate more eyeballs on that change though. I'll probably have to highlight it to d-devel
18:30:07 <spwhitton> smcv: and can I action you to prepare stable or oldstable updates after helmut's update reaches unstable?
18:30:23 <spwhitton> it's not really TC business but having action items to check up on next meeting may help jog our memories.
18:30:27 <smcv> spwhitton: ok
18:30:32 <helmut> please give that change at least 3 weeks in trixie before migrating to stable.
18:30:50 <spwhitton> #action helmut to get all changes that need to go to stable debootstrap into sid
18:30:53 <helmut> I hope it doesn't break anything, but I kinda fear otherwise
18:31:04 <spwhitton> #action smcv to prepare (old)stable update for debootstrap once helmut's upload has been in sid for three weeks
18:31:10 <spwhitton> okay.  shall we move on?
18:31:18 <spwhitton> thanks to both of you.
18:31:31 <helmut> sure, we've lost ourselves in details anyway and this all is design and implementation work rather than ctte matter
18:32:26 <spwhitton> I find all this quite helpful for retaining enough of this in my head.
18:32:29 <spwhitton> #topic Any Other Business
18:32:32 <spwhitton> is anyone aware of anything?
18:32:48 <Emperor> not me
18:33:02 <helmut> the tzdata vs UTC one Myon mentioned seems to have resolved itself peacefully
18:33:19 <Myon> yes
18:33:43 <roehling> same for the cmake/meson/python install layout
18:33:49 <spwhitton> great!
18:33:57 <Myon> I didn't expect it to become an issue, just wanted to ask for more opinions while it wasn't clear yet
18:34:57 <Myon> (I spent most of the morning on debugging the fallout, only to realize that installing tzdata-legacy from *unstable* instead of from trixie made the remaining PostgreSQL testsuite problems go away)
18:35:11 <smcv> the cmake/meson/python install layout I have proposed a debhelper patch, interested people please take a look
18:35:36 <smcv> likewise for meson packages regressing by compiling bytecode when previously they didn't
18:36:06 <roehling> I had a look at the dh patch and it looked fine, but my Perl skills are very limited
18:36:24 <spwhitton> #endmeeting