[00:00] <anon^_^> gpg --keyserver hkps://keyserver.ubuntu.com:443 --recv-keys 7EF33F027E9E4869F46F77E34E72F77D7D158F33
[00:00] <anon^_^> gpg: key 4E72F77D7D158F33: public key "Launchpad PPA for Micah Lee" imported
[00:00] <anon^_^> gpg: Total number processed: 1
[00:00] <anon^_^> gpg:               imported: 1
[00:01] <anon^_^> however, apt-key fails
[00:01] <anon^_^> sudo add-apt-repository ppa:micahflee/ppa
[00:01] <anon^_^> .
[00:01] <anon^_^> Executing: /tmp/apt-key-gpghome.XNiyRBLgEY/gpg.1.sh --keyserver hkps://keyserver.ubuntu.com:443 --recv-keys 7EF33F027E9E4869F46F77E34E72F77D7D158F33
[00:01] <anon^_^> gpg: WARNING: Tor is not properly configured
[00:01] <anon^_^> gpg: keyserver receive failed: Permission denied
[00:02] <rbasak> anon^_^: please stop spamming the channel. You said you had a question. Please just ask it. We don't need a bug report spammed into the channel - we have the bug tracker to record a bug if that's what you want to do.
[00:02] <anon^_^> it's a critical bug, i don't have launchpad account
[00:03] <rbasak> It doesn't sound like it meets our definition of "Critical": https://wiki.ubuntu.com/Bugs/Importance
[00:03] <rbasak> You can create a Launchpad account.
[00:03] <Unit193> I have 18.04, I have tor, gpg --refresh-keys works as expected.
[00:04] <anon^_^> Unit193, yes gpg --refresh-keys works, but add-apt-repository is broken
[10:29] <rbalint> hi, how do i resolve circular build dependencies in Ubuntu? Should I just upload a "stage1" source package then, after rdeps are rebuilt, the good one or there are extra steps?
[10:29] <rbalint> (i could not find the wiki page for that)
[10:30] <rbalint> i'm looking at the loop including golang-google-cloud-compute-metadata-dev
[10:56] <cjwatson> rbalint: (at least some) archive admins can inject build-deps to resolve loops
[10:57] <cjwatson> but will need clear instructions on the sequence
[11:10] <rbasak> mapreri: "Default to checking signatures while pulling a .dsc" -> regresses download of some Ubuntu packages. I'm not sure why.
[11:11] <rbasak> mapreri: some examples: hoel lyricue net-snmp orthanc-mysql pam-mysql proftpd-dfsg rsyslog
[11:12] <rbasak> mapreri: failure example: https://pastebin.ubuntu.com/p/vYH7xNm6Zq/
[11:13] <mapreri> rbasak: because it makes a lot of sense in debian, but doesn't in ubuntu, which doesn't have an equivalent to debian's debian-keyring.
[11:13] <mapreri> some other canonical employee mailed me the other day about it
[11:13] <mapreri> so I expect the verification to fail for anything that is not signed by a key in the debian-keyring, which pretty much means ~everything with a direct upload to ubuntu
[11:13] <rbasak> I see
[11:14] <mapreri> rbasak: so I'm thinking of https://paste.debian.net/1095441/ - which should keep the validation enabled on debian only (since it subclasses SourcePackage and overrides the pull_dsc() method with a verify_signature=True)
[11:15] <rbasak> mapreri: sounds good
[11:15] <Odd_Bloke> Ohh, _that_ explains why I've been seeing those failures a lot recently.  I thought I'd screwed up my GPG keyring or something.
[11:16] <rbasak> mapreri: can we upload a fix fairly quickly please?
[11:16] <Odd_Bloke> (Luckily my strategy of ignoring the problem in the hopes it would go away has paid dividends.)
[11:16] <mapreri> rbasak: I wanted to finally enable autopktest for ubuntu-dev-tools in the next upload, I was waiting on finding enough tuits to do that
[11:17] <rbasak> mapreri: I don't think a fix for this issue should wait for any other changes.
[11:17] <rbasak> It's breaking the majority of people who use the package, no?
[11:17] <mapreri> well, those tools that download source packages, at most.
[11:18] <rbasak> So most (all?) Ubuntu developers then!
[11:18] <mapreri> which are arguably the most important of them all
[11:18] <mapreri> let me see if I can do that all now…
[11:18] <rbasak> Thanks!
[11:22] <rbasak> mapreri: arguably the behaviour is wrong on Debian too - what if someone retires, their key is removed from the keyring, but their last upload remains?
[11:22] <mapreri> rbasak: well, you proposed that change in the first place
[11:23] <mapreri> and added a --no-verify-signature to pull-debian-source to override that…
[11:23] <mapreri> this was a little more than a month ago, already forgotten? :)
[11:23] <mapreri> just, your change broke the testsuite, but I noticed only after merging.  my change "Default to checking signatures while pulling a .dsc" was a way to fix the testsuite after that
[11:24] <rbasak> I noticed --no-verify-signature is attributed to me but I don't remember the detail!
[11:24] <mapreri> https://code.launchpad.net/~racb/ubuntu-dev-tools/+git/ubuntu-dev-tools/+merge/326608
[11:24] <mapreri> oh
[11:24] <mapreri> that was 2017!
[11:24] <mapreri> sorry, I looked at month and day without year :3
[11:25] <rbasak> Perhaps I had hit the Debian verification failure case already?
[11:26] <rbasak> I'm not sure it was my intention to add additional checking!
[11:31] <rbasak> mapreri: looks like it needs SourcePackage.pull() adjusting, not pull_dsc().
[11:31] <rbasak> Otherwise the default of True gets passed through.
[11:32] <mapreri> good point.  the original report I had was from `backportpackage`, which only uses pull_dsc().
[11:43] <mapreri> rbasak: could you please confirm that https://paste.debian.net/1095443/ works for you?
[11:44] <rbasak> Looking
[11:45] <Odd_Bloke> Do we have any policy about what level of logging should be written to the journal (vs. a package-specific file in /var/log)?
[11:47] <rbasak> mapreri: confirmed. Reproduced in a fresh eoan container, and I patched /usr/lib/python2.7/dist-packages/ubuntutools/archive.py directly with your patch, and pull-lp-source started working on a problem package.
[11:47] <rbasak> Incidentally it looks like pull-lp-source is still on Python 2, but I would leave that for now.
[11:50] <mapreri> most of ubuntu-dev-tools is still py2
[11:51] <mapreri> it's not a difficult task to migrate it, but requires someone to do it
[11:54] <mapreri> on that regard, I hoped ddstreet merged his fork that did over a year ago before…
[11:58] <mapreri> rbasak: uploaded, let's see how this goes.
[12:04] <rbasak> mapreri: thank you for looking at it so quickly!
[12:04] <rbasak> And sorry for the fallout. I'm not trying to punish you for looking at my MP, honest :-/
[12:06] <mapreri> don't worry for that bit :)
[12:19] <hallyn> hm, 'pull-lp-source qemu' on eoan fails on checking the signature
[12:21] <hallyn> cpaelzer: ^
[12:23] <rbasak> hallyn: see above!
[12:24] <hallyn> oh? sorry :)
[12:29] <hallyn> (finally reading the backlog) zomg
[12:36] <cpaelzer> oh
[12:37]  * cpaelzer read the backlog as well now ... oO
[12:38] <hallyn> it seems to imply that .dsc's didn't use to be verified before?  that surprises me
[12:41] <hallyn> cpaelzer: so, actually, i was wanting to lok through the source to see - should virtio-blk storage be enabled ?
[12:42] <hallyn> i'm not finding it in the manpage, at any rate, maybe it's just not there
[12:43] <cpaelzer> hallyn: I think it is enabled by default, IIRC we have no delta
[12:43] <mapreri> hallyn: it doesn't really make sense to verify a signature on a ubuntu's .dsc, because ubuntu developres can just change their gpg key whenever they want.
[12:43] <cpaelzer> let me check
[12:43] <mapreri> and they are not collected on a keyring either
[12:43] <hallyn> mapreri: well, but they do keep them on launchpad no?
[12:43] <mapreri> there is no trust path to them, etc
[12:43] <hallyn> "it's better than nothing"
[12:44] <mapreri> we still have no keyring to give to gpg
[12:44] <hallyn> hm
[12:44] <hallyn> i wonder how big it would get if we were to just rsync a tree of all the ubuntu developer's pubkeys from lp :)
[12:44] <hallyn> (as a sidestep to apt-get update if you have ubuntu-dev-tools installed)
[12:45] <cpaelzer> hallyn: yeah I see virtio-blk.o build and linked - so default-on/no-delta is correct
[12:45] <cpaelzer> maybe we miss a new manpage since you expect that one?
[12:45] <hallyn> cpaelzer: great, thanks!  i'll do some more testing
[12:45] <hallyn> i just thought checking the source would be easier
[12:46] <hallyn> but, it wasn't :)
[12:46] <hallyn> i dunno - upstream might just not stick every blockdev type in the manpage.  i just thoght they used to
[12:47] <cpaelzer> ok, let me know if you find something missing
[12:48] <hallyn> will do, thx.  (it'll be later this afternoon, unfortunately)
[12:48] <cpaelzer> as long as you give me a highlight I can collect it the other day then
[12:49] <hallyn> what i'm looking for is, having an lvm thinpool, i want to be able to dd if=/dev/zero of=/xxx bs=1G count-2;  rm /xxx, fstrim -v /, and then have the backing store be trimmed.
[12:51] <cpaelzer> so you look for discard being passed on?
[12:52] <cpaelzer> there is one discard option at -blockdev and more depending on the format (e.g. qcow2 has more)
[12:58] <hallyn> cpaelzer: well, no - apparently (according to the interwebz) virtio-blk only supports discard with linux 5.0, which I'm not using yet;  and is obsoleted  virtio-scsi does support it.
[12:59] <hallyn> but when i did 'man qemu-system-x86_64' i didn't find any reference to it. but then virtio-blk also doesn 'tshow up, so i guess it's just not ther.e  it's probably fine :)
[13:00] <cpaelzer> yeah virtio-scsi is what rharper recommended to jamespage recently for discard as well
[13:30] <rbasak> hallyn: the problem with best effort for signature is that you can't rely on that for scripting. Otherwise something will inevitably fail.
[13:31] <rbasak> It could be written to validate against the apt repository signature in the common case that the source package version being downloaded is available in an apt repository I suppose.
[13:31] <rbasak> But that would need implementing, and it'd have to be a different code path from the general path of being able to grab any version published ever.
[13:31] <rbalint> cjwatson, thanks! and is injecting ppa-built build deps preferred over doing the loop breaking in -proposed?
[13:34] <cjwatson> rbalint: I think so
[13:36] <rbalint> cjwatson, ack, will go that way then
[13:53] <rbasak> Subject: [proposed-migration] mysql-8.0 8.0.16-0ubuntu2 stuck in eoan-proposed for 13 days.
[13:53] <rbasak> I know. Please stop spamming me.
[13:53] <rbasak> Five today so far :-/
[14:01] <xnox> force-skip-test uploader/rbasak/all
[14:02] <rbasak> ?
[14:22] <Laney> rbasak: Can't immediately see from code inspection why that's happening, so I'll add a bit of debugging...
[14:22] <rbasak> Thanks!
[14:23] <Laney> this is https://git.launchpad.net/~ubuntu-release/britney/+git/britney2-ubuntu/tree/britney2/policies/email.py#n176 if you want to play along at home
[14:24] <Laney> I forgot about that exciting calculation we have there
[14:27] <Laney> At 13 days age the threshold should be 11 days, and that's what is in the cache
[14:27]  * Laney blames computers
[14:33] <Laney> it's not every run either
[15:13] <GunnarHj> Hi rbasak, do you possibly have time to consider this:
[15:13] <GunnarHj> https://lists.ubuntu.com/archives/devel-permissions/2019-August/001391.html
[15:24] <rbasak> GunnarHj: I suppose I should, as a Bangla speaker!
[15:24] <rbasak> It sounds perfectly fine to me, but I need to check if it needs one DMB member or a quorate number to approve.
[15:25] <rbasak> IIRC it should be PPU though, rather than personal-* packageset entry
[15:27] <GunnarHj> rbasak: Had no idea that you speak Bangla. :) I mentioned this to Łukasz when he sponsored the addition of the package, and he thought that the packageset would be fine.
[15:29] <rbasak> He doesn't seem to be online right now
[15:29] <GunnarHj> rbasak: Right.
[15:29] <rbasak> If you're curious, https://wiki.ubuntu.com/DeveloperMembershipBoard/KnowledgeBase#Personal_packagesets_and_glob_expansions is our documentation on that.
[15:29] <rbasak> But this is something we should be able to sort out ourselves and tell you technically how we're achieving it.
[15:29] <rbasak> Oh
[15:29] <rbasak> "This is defined as the set that the DMB has agreed that Gunnar may upload, which includes individual packages to which he has PPU, as well as glob expansions."
[15:29] <rbasak> OK, I'm wrong.
[15:30] <rbasak> I think we need to treat it as a PPU application though
[15:31] <GunnarHj> rbasak: With a voting round then?
[15:32] <rbasak> IIRC, yes, but I'm just trying to find documentation to confirm.
[15:32]  * rbasak prefers to be consistent about these things
[15:33] <rbasak> I don't see it being a problem to add you though - makes perfect sense to me.
[15:33] <rbasak> And nor do I want to make you write up a tedious application or anything.
[15:35] <rbasak> GunnarHj: I can't find any documentation on an exception so I think a vote is required. Let me add this to our meeting agenda.
[15:35] <rbasak> GunnarHj: IMHO think your email request is sufficient.
[15:36] <GunnarHj> rbasak: Ok, that works for me. It means that I'll need to bother my sponsor an extra time, but that's how it is.
[15:39] <GunnarHj> rbasak: Since you are a Bangla speaker, it would be nice it you could install the package and let me know how useful you think it is.
[15:41] <rbasak> GunnarHj: I'd be happy to give it a try, but my literacy skills are quite poor and rusty. My spelling is atrocious :-/
[15:41] <rbasak> I did say speaker, not writer or reader :-P
[15:43] <GunnarHj> rbasak: Your view would be much more valuable than mine, anyway. I can type বাংলা, but that's about it. :)
[15:43] <rbasak> :)
[15:44] <rbasak> (incidentally the Ubuntu Mono font completely screws that up)
[15:46] <GunnarHj> rbasak: Yeah, even I realize that it doesn't look right.
[15:48] <rbasak> There's a complexity that some Indian writing systems have (Bengali, Devanagari, maybe some more) that vowels appear around consonants. It was never obvious to me which way round Unicode expects them to be encoded for all cases, and it's really common to see rendering get all confused.
[15:53] <GunnarHj> rbasak: I have seen some Ask Ubuntu questions about problems with such languages in e.g. gnome-terminal. Sometimes I point them to konsole, which seems to handle it better.
[16:54] <Locutusofborg> hello, do we have some sort of way to know all the delta one particular developer introduced in Ubuntu wrt Debian?
[16:54] <Locutusofborg> I would like to know how many gcc-9 fixes I did, since now they should go in Debian (including wl-asneeded ones)
[16:54] <Locutusofborg> and do some NMUs there
[16:56] <rbasak> The changelog?
[16:56] <rbasak> Clearly that's not the answer you're looking for, but I'm not sure what I'm missing.
[16:56] <rbasak> Oh, you mean across all packages?
[16:56] <rbasak> The sponsorship miner perhaps?
[16:56] <rbasak> Also grep-merges might be helpful.
[16:57] <Locutusofborg> rbasak, grep-merges shows only outdated stuff?
[16:57] <rbasak> There's also https://launchpad.net/~/+uploaded-packages
[16:57] <rbasak> Locutusofborg: yes, good point
[16:58] <Locutusofborg> rbasak, problem, is that I keep my merges up-to-date :p
[16:58] <Locutusofborg> but yeah, I can look for gcc-9 bugs in Debian and compare with rmadison the Ubuntu version
[16:59] <Locutusofborg> starting from there maybe
[16:59] <Locutusofborg> https://bugs.debian.org/cgi-bin/pkgreport.cgi?submitter=doko@debian.org
[16:59] <rbasak> Or grep your mailbox for queue accepts?
[16:59] <Locutusofborg> cat page |grep "ftbfs with GCC-9" |cut -f 2 -d ":" |cut -f 1 -d "]" > list
[17:00] <Locutusofborg> this might work
[17:00] <Locutusofborg> rbasak, doesn't help in general
[17:00] <Locutusofborg> I'm not the only one who did fix ld-asneeded stuff
[17:00] <rbasak> I'm not sure I understand the general specification for your request
[17:00] <Locutusofborg> I want to start from *my* packages, but it is a good time to send all this kind of delta we have
[22:17] <Unit193> Is X-GNOME-Gettext-Domain a thing?
[22:29] <Unit193> (codesearch didn't really pick up anything.)
[23:49] <Unit193> https://launchpad.net/builders/lgw01-amd64-046 and a couple others seem a bit stuck.  I could build that package on an old P4 in about two minutes. :3