2020/02/11/#ubuntu-meeting.txt

=== cpaelzer__ is now known as cpaelzer
=== cpaelzer__ is now known as cpaelzer
[14:01] <doko> o/
[14:02] <didrocks> he
[14:03] <doko> cpaelzer, jamespage, joining?
[14:03] <jamespage> o/
[14:03] <jamespage> yep
[14:05] <doko> let's start with component mismatches
[14:05] <doko> https://people.canonical.com/~ubuntu-archive/component-mismatches-proposed.svg
[14:05] <doko> didrocks: can you do/forward alsa-lib?
[14:06] <didrocks> doko: I will forward it
[14:06] <cpaelzer> here
[14:06] <doko> and I didn't look at the status of ghostscript/fonts, although cpaelzer did
[14:06] <didrocks> doko: actually, it’s the kernel team who maintains alsa
[14:07] <didrocks> so I guess it should be them filing the MIR
[14:07] <doko> jamespage: cinder/python-tabulate ?
[14:07] <didrocks> (looking at alsa-lib subscriber)
[14:07] <cpaelzer> https://bugs.launchpad.net/ubuntu/+source/fonts-urw-base35/+bug/1862048 was waiting for a subscriber
[14:07] <ubottu> Launchpad bug 1862048 in ghostscript (Ubuntu) "[MIR] fonts-urw-base35" [High,New]
[14:07] <cpaelzer> didrocks: you  said you can't do so and pinged others
[14:07] <cpaelzer> has this happened?
[14:07] <didrocks> yes
[14:07] <didrocks> desktop-packages is sub
[14:07] <joeubuntu> sorry all, late to the meeting.
[14:08] <cpaelzer> Then this is ready for promotion @didrocks
[14:08] <didrocks> will do after this meeting
[14:08] <cpaelzer> I updated the bug accordingly
[14:09] <doko> I'll forward apport/terminator to foundations
[14:09] <cpaelzer> there are two more in te new MIR queue
[14:09] <doko> and doing the licensecheck ones
[14:09] <cpaelzer> https://bugs.launchpad.net/ubuntu/?field.searchtext=&orderby=-date_last_updated&field.status%3Alist=NEW&assignee_option=none&field.assignee=&field.subscriber=ubuntu-mir
[14:09] <cpaelzer> telp / amtk - has anyone context on those?
[14:10] <didrocks> I don’t but I’ll handle it
[14:10] <doko> gedit
[14:10] <didrocks> them*
[14:11] <cpaelzer> thanks didrocks, IÄll assign you ont he bugs then
[14:11] <didrocks> thx
[14:11] <cpaelzer> Lets also look at the recently modified incomplete
[14:11] <doko> so the open one is cinder/python-tabulate
[14:11] <cpaelzer> whiel I fetch the Link I think Laney wanted to talk about something
[14:11] <cpaelzer> Laney: around?
[14:12] <cpaelzer> incomplete MIRs
[14:12] <cpaelzer> https://bugs.launchpad.net/ubuntu/?field.searchtext=&orderby=-date_last_updated&field.status%3Alist=INCOMPLETE_WITH_RESPONSE&field.status%3Alist=INCOMPLETE_WITHOUT_RESPONSE&field.subscriber=ubuntu-mir
[14:12] <cpaelzer> jeepeney was done alst week and needs no further action for now
[14:12] <doko> this list is long
[14:13] <cpaelzer> they don't auto-expire
[14:13] <cpaelzer> doko: therefore - we only look at the last touched
[14:13] <cpaelzer> although when in Frankfurt on the sprint we could clear out the past if all us want to do so
[14:13] <cpaelzer> should be a quick everyone-nods-and-set-invalid pass
[14:13] <cpaelzer> ec2-instance-connect still is disliked by me and security
[14:14] <cpaelzer> but I know rbalint is working on it
[14:14] <cpaelzer> anyone haveing any other MIRish topic to discuss ?
[14:15] <didrocks> nothing here
[14:15] <doko> jamespage: still here?
[14:15] <jamespage> yep
[14:16] <jamespage> sorry - multi-tasking never helps == multi-failing
[14:16] <jamespage> I think there was one new security MIR added to the queue for joeubuntu's team
[14:16] <jamespage> in the last week - masakari
[14:17] <doko> filed https://bugs.launchpad.net/ubuntu/+source/python-tabulate/+bug/1862773
[14:17] <ubottu> Launchpad bug 1862773 in python-tabulate (Ubuntu) "[MIR] python-tabulate (dependency of cinder)" [High,Incomplete]
[14:17] <doko> joeubuntu: ^^^
[14:18] <jamespage> masakari is in the backlog on the security team trello so its in the queue
[14:18] <jamespage> thanks
[14:19] <jamespage> doko: coreycb or I will pickup completion of that MIR for tabulate
[14:19] <joeubuntu> IT's on the list, thanks doko
[14:19] <doko> ta
[14:19] <doko> anything else?
[14:19] <cpaelzer> I think we are good
[14:19] <cpaelzer> as I said Laney had some question
[14:20] <cpaelzer> but I don't know which one, only that didrocks and I said plese get to the IRC meeting to talk about it
[14:20] <cpaelzer> didrocks: do you know what it was about?
[14:21] <didrocks> I don't at all
[14:21] <didrocks> we have our desktop meeting in 10 min, so he should soon be around
[14:21] <cpaelzer> hmm later/next time then
[14:21] <cpaelzer> oh that is good
[14:22] <cpaelzer> Oh FYI I won't be here next week (PTO)
[14:22] <doko> see you, bye
[14:22] <cpaelzer> cu
[14:22] <didrocks> enjoy cpaelzer :)
[14:25] <didrocks> doko: just to be clear: I’ll let you handle alsa-lib dep with the kernel team if you don’t mind
[14:29] <doko> didrocks: to be clear, doing adminstrative work on a MIR doesn't have to be done just be me
[14:29] <doko> now https://bugs.launchpad.net/ubuntu/+source/alsa-ucm-conf/+bug/1862776
[14:29] <ubottu> Launchpad bug 1862776 in alsa-ucm-conf (Ubuntu) "[MIR] alsa-ucm-conf & alsa-topology-conf (b-d of alsa-lib)" [High,Incomplete]
[14:31] <didrocks> I don’t think filing an empty MIR really helps in moving the discussion forward with the people who need to deal with it, but probably material for discussing in Frankfurt
[14:32] <Laney> back here
[14:32] <Laney> is the meeting still going obn?
[14:32] <didrocks> it's done
[14:33] <Laney> meh
[14:33] <didrocks> but I guess you can write here, people seem still being around
[14:33] <cpaelzer> yep
[14:33] <Laney> it's not on the fridge calendar btw
[14:34] <Laney> yeah ok, so there's a project going on at the minute with some of us @ canonical
[14:34] <Laney> basically the goal is to make the oem enablement tweaks that some hardware requires, and comes with when you buy it with ubuntu pre-loaded, also available if you buy with another OS and install Ubuntu yourself
[14:35] <Laney> essentially requires defining extra packages to install for different hardware
[14:35] <cpaelzer> doko: jamespage: joeubuntu: ^^ highlight to make you come back to the meeting :-)
[14:35] <Laney> we're going to do this by including metapackages on the iso that declare what they're compatibile with, and dynamically installing them
[14:35] <Laney> now "on the iso" implies "in main"
[14:35] <Laney> and "package*s*" implies many
[14:36] <Laney> so I'm coming to you to try to work out an exception for this
[14:36] <cpaelzer> This sounds familiar, did you bring that up on the last engineering sprint already?
[14:36] <Laney> I did write up a draft here: https://wiki.ubuntu.com/MIRTeam/Exceptions/OEM
[14:36] <Laney> yes
[14:36] <Laney> the idea is that packages which fit that draft can go in without MIR
[14:36] <cpaelzer> great, the exception is what I'd have asked for
[14:36]  * cpaelzer reading ...
[14:38] <cpaelzer> so the real content is in an extra apt archive
[14:38] <cpaelzer> and this is really just abotu the meta-packages to go onto the iso
[14:38] <cpaelzer> that was unclear last time we talked, thanks for adding the link on the wiki page
[14:39] <Laney> more or less, and that part is being considered by the TB
[14:39] <Laney> the meta pkgs themselves are super trivial
[14:39] <didrocks> how many of those packages per release do you expect to see during a release life?
[14:40] <Laney> no idea
[14:40] <Laney> that all depends on how many laptops are enabled which is a bit beyond my field of view
[14:40] <Laney> but you could say several
[14:41] <didrocks> wondering as well about the burden on the SRU team, but that’s another topic (to see if the mechanism is realistic)
[14:41] <cpaelzer> Laney: I know it was only an idea back then when we talked - but is there any chance to get a linter-script for these rules ont that wiki page?
[14:41] <didrocks> anyway, if we expect several, I think we should have an automated checker for them
[14:41] <cpaelzer> the AAs could use that to verify that the package follows the rules before promotions
[14:41] <didrocks> exactly :)
[14:41] <cpaelzer> hehe, same thoughts it seems
[14:41] <Laney> explain more
[14:41] <Laney> please
[14:41] <cpaelzer> Laney: I think the definitions on the page are good
[14:42] <cpaelzer> Laney: the next step would be writing and attaching a script of some sort
[14:42] <Laney> I don't really have weeks to spend writing a script
[14:42] <Laney> or days even
[14:42] <Laney> so if that blocks this, it might do so for some time
[14:42] <cpaelzer> well, if we don't have such a thing the AAs will ahve to manually check against the defnitions ont his page
[14:43] <cpaelzer> which can work, but is error prone as we all know
[14:43] <cpaelzer> and it scales wit hthe number of packages that will go this path
[14:43] <didrocks> and can consume even more time depending on how many packages we are talking about (which we should know before starting this)
[14:43] <didrocks> I guess the OEM team can give some estimation
[14:44] <cpaelzer> Laney: IMHO it will not stall the approval of this approach to not (yet) have this script
[14:44] <cpaelzer> But once the actual "please promote on the base of this" happens
[14:44] <cpaelzer> then having one will make it fast
[14:44] <cpaelzer> and lacking that checker will make it slow and the AAs grumpy
[14:45] <Laney> ok I can put it on the list, but I have to deliver the project itself as a higher priority, hope you understand
[14:45] <cpaelzer> absolutely
[14:46] <didrocks> I think it can be seem as a broken record, but I would really like to have at least have a guess estimate of the number of packages we are talking about
[14:46] <Laney> didrocks: can you explain the background behind your request?
[14:46] <Laney> if it's 10, you prefer to review them all manually?
[14:46] <Laney> but 15 not?
[14:47] <didrocks> I guess 10 is indeed ok, but if it’s 30, the script should be mandatory before we start such a process
[14:47] <Laney> it's an interesting one
[14:47] <didrocks> as AA will likely spend more time and it will be more error-prone
[14:48] <Laney> if you decline this exception then it is the *MIR* team that gets more work
[14:48] <juliank> It's not that hard to have one package as a reference, and then run debdiff against any new ones, though
[14:48] <Laney> I'm interested because it saves me paperwork, but you should be because it saves you MIRs to review
[14:48] <juliank> And that provides a reasonable review base, I'd guess
[14:48] <cpaelzer> that is a simple approach to such a helper script
[14:48] <cpaelzer> good hint juliank
[14:48] <didrocks> I’m more thinking about MIR/AA/SRU in general, and not not caring just to deliver but giving the load to others
[14:49] <Laney> I'm saying the difference between declining until I write a script and approving without the script isn't that great
[14:49] <Laney> it's the addition of some paperwork
[14:49] <Laney> but if debdiffing some template is OK, ...
[14:49] <didrocks> which will back-pressure all parties to have a script
[14:50] <didrocks> or it will never happen because life and next projects…
[14:50] <juliank> "debdiff <first .dsc> <new .dsc> | filterdiff -p1 -x debian/changelog -x debian/modaliases" should be enough of a script
[14:50] <didrocks> yep
[14:50] <Laney> seriously?
[14:50] <joeubuntu> This sounds like a method to just get around MIRs,  will the packages that get installed be supported for 10 years?
[14:51] <Laney> it is a method to get around MIRs
[14:51] <Laney> they'll be supported for the life of the release yes
[14:51] <Laney> get around in the sense that doing MIRs will be extremely repetitive
[14:52] <joeubuntu> But the burden of support isn't any lower, shouldn't we do some validation of the packages supportability?
[14:52] <cpaelzer> joeubuntu: the packages will have no "active" content
[14:53] <cpaelzer> joeubuntu: we essentially do the validation on the template
[14:53] <cpaelzer> and then based on that provide a fast path as long as that pattern is matched
[14:53] <joeubuntu> OK, that makes sense.
[14:53] <joeubuntu> Thanks
[14:53] <cpaelzer> Laney: please correct me if your plan changed from what I barely remember from half a year ago :-)
[14:54] <cpaelzer> Laney: maybe the words in the last few messages here could be added to the wiki
[14:54] <Laney> no that's it
[14:54] <cpaelzer> it seems they help to calrify
[14:54] <Laney> so that debdiff thing, that is ok for a script for you?
[14:54] <Laney> I can upload a 'good' template somewhere and then just add that to the page ...
[14:54] <cpaelzer> yeah, if you attach the files needed and how to invoke to the wiki that should be ok
[14:55] <didrocks> I think so, just wrap it in an helper for the AA and that’s fine
[14:55] <cpaelzer> a repo somewhere and a link to it from the wiki might be even better than an attachment
[14:55] <didrocks> I would really put that in lp:ubuntu-archive-tools
[14:55] <cpaelzer> didrocks: weren't there a aa-helper repository somewhere?
[14:55] <cpaelzer> yes that is what i was looking for
[14:55] <cpaelzer> Laney: would that work for you (not stalling you too much) but giving you what you need?
[14:56] <Laney> I thought you wanted an actual static verifier, but this should be ok if you don't mind
[14:57] <cpaelzer> we are all reasonable people, in a perfect world there would be a super-duper-verifier - but this seems to work
[14:57] <cpaelzer> if it turns out it does not it will slow down the AAs and thereby your packages from promotion
[14:57] <didrocks> and all packages will have a very strong testsuite in them :p and and and… :)
[14:57] <cpaelzer> but I think this will be good
[14:57] <juliank> Oh, test suite
[14:57] <didrocks> ahah
[14:57] <cpaelzer> Laney: would you ping me once you updated the wiki/repo please?
[14:58] <juliank> It would be nice to have autopkgtest that installs it and run apt update I guess to ensure there's no typos in the .list entry
[14:58] <cpaelzer> we lack the people to "sign-off" on this atm, but I can send a mail to the MIR team then
[14:58] <Laney> cpaelzer: ok, I will do, thanks
[14:58]  * Laney ignores juliank trying to load more work onto this
[14:58] <cpaelzer> juliank: good suggestion
[14:58] <cpaelzer> lets add it only as a bonus-objective to avoid the anger of Laney
[14:59] <didrocks> heh
[14:59] <Laney> if you're being cool you'd do it with autodep8
[14:59] <Laney> and *that* would be the static verifier
[14:59]  * Laney runs
[14:59] <juliank> Oh yeah, I should write an autodep8 module for verifying packages which install sources.list entries
[14:59] <didrocks> yeah, but nothing ensures you that in the long run, the repo is still valid
[15:00] <didrocks> and have needed packages
[15:00] <didrocks> but anyway, let’s move on :)
[15:00] <didrocks> (or you need regularly running autodep8, which in the end is a jenkins-like solution with reporting and so on…)
[15:06] <Laney> funny thing but juliank is at some point in our lives going to work on something similar to that