[09:27] vorlon, hey, was there any reason you did file bug #1853646 without doing the actual removals? [09:27] bug 1853646 in zope.lifecycleevent (Ubuntu) "RM: zope.lifecycleevent, lazr.lifecycle: python2-only, Ubuntu-specific, build-depends on removed van.pydeb, no other reverse-dependencies" [Undecided,New] https://launchpad.net/bugs/1853646 [10:47] anyone want to port code-of-conduct-signing-assistant to PyQt5 so it can be kept? [10:47] it fails to start anyway since bionic, where Qt4Webkit went away [10:48] LP: #1855255 [10:48] Launchpad bug 1855255 in code-of-conduct-signing-assistant (Ubuntu) "Port to python3/pyqt5 or remove from focal archive" [Undecided,New] https://launchpad.net/bugs/1855255 === ricab is now known as ricab|lunch === ricab|lunch is now known as ricab === Son_Goku is now known as Conan_Kudo === Conan_Kudo is now known as Son_Goku [15:57] cjwatson: have you seen by any chance when attempting to push a new git repo to launchpad? "fatal: remote error: Path translation timed out." [16:00] coreycb: Yes, it's a known problem when the parent repository has lots of refs. I have an MP in review to (probably) fix this, but we can work around it if you tell us the repo you're trying to push [16:02] cjwatson: great, thanks. I'd like to create lp:~ubuntu-server-dev/ubuntu/+source/migrate and lp:~ubuntu-server-dev/ubuntu/+source/python-cryptography if possible. [16:03] seb128: LP: #1853646> mostly because there were a bunch more to file and I left one open so I could find it again and use it as a template :) [16:03] Launchpad bug 1853646 in zope.lifecycleevent (Ubuntu) "RM: zope.lifecycleevent, lazr.lifecycle: python2-only, Ubuntu-specific, build-depends on removed van.pydeb, no other reverse-dependencies" [Undecided,Fix released] https://launchpad.net/bugs/1853646 [16:03] coreycb: There's no sysadmin vanguard around right now, so might have to wait a bit ... [16:04] cjwatson: ok [16:04] I'll ask when I next spot one, and let you know [16:04] thanks [16:13] bryce: o/ I have a simplestreams upload prepared that I need sponsorship for, and powersj pointed me in your direction. The debdiff is https://paste.ubuntu.com/p/J35wNpBDwQ/. Is there a better way for me to communicate this to you (git-ubuntu &c.)? [16:15] kanashiro: https://lists.ubuntu.com/archives/ubuntu-devel/2019-April/040622.html is the background I was referring to on Rails issues in proposed migration [16:16] Odd_Bloke, good morning! yes I'll help with sponsorship [16:17] Thanks! [16:17] Odd_Bloke, we do usually use git ubuntu for stuff but I can work with debdiffs. Unless you want to learn git ubuntu I can just take care of that end of things [16:18] bryce: I'd be happy to learn git-ubuntu. [16:21] I suggest that you take it incrementally. For a start, if you already have a debdiff, you can just clone the 'ubuntu/devel' branch from https://code.launchpad.net/ubuntu/+source/simplestreams, apply your changes, and then submit those back as a merge proposal. [16:23] There is currently no good solution for supplying the upstream tarball - you'll need to get that to your sponsor out of band (same as the situation with a debdiff) [16:24] rbasak, thanks for the link, fair points you raised there [16:25] Odd_Bloke, ok let's dig into that post team meeting. I'm reading through the debdiff presently. [16:29] bryce: rbasak: What's the appropriate commit message format? [16:29] We don't have any specific rule, but we usually use the same format as what would go into debian/changelog since then we can use a tool called reconstruct-changelog to write debian/changelog for us [16:30] But nothing is mandated [16:30] Odd_Bloke, what you're written looks fine, the trick is to have it as its own git commit [16:44] bryce: So my single git commit message would be "* New upstream snapshot.\n - tools/j2signed: ...\n - ..."? [16:49] Odd_Bloke, essentially, but indent it 2 spaces on the left, just like you cut and paste from the d/changelog [16:52] bryce: https://code.launchpad.net/~daniel-thewatkins/ubuntu/+source/simplestreams/+git/simplestreams/+merge/376423 <-- how's that? [16:54] Odd_Bloke, sorry distracted by the meeting :-) [16:54] Odd_Bloke, ok so the main thing is to make the changelog a separate commit from the rest of the changes [16:54] bryce: I think we should be clear that making the changelog a separate commit is a convention we've established in our squad and isn't a git-ubuntu requirement. [16:55] so "git commit debian/changelog -m 'changelog'; git commit -a " * New upstream snapshot\n..." [16:55] rbasak, er, I thought it was a requirement [16:55] It's not :) [16:56] rbasak, in any case, for teaching purposes best to give just one "right" way to do it [16:56] bryce: I'm torn. I appreciate that point in general, but my intention for git-ubuntu is for all Ubuntu developers to be able to use it regardless of their specific preferences and workflows. I don't want to put people off by adding things they have to do. [16:57] So I want to be liberal in what we accept. [16:57] If it follows usual git conventions and results in a functional upload, I think we should accept it. [16:58] FWIW, if I'm coming to you for guidance, being told that there are many ways I can do it is more offputting than having specific instructions. (_I_ understand you're having a more general conversation, but bear in mind that most people _wouldn't_ know that. :p) [16:58] bryce: Updated the MP, how does it look now? [16:59] Odd_Bloke, yes looks great [16:59] Cool. :) [17:01] next, for the review process, I download your branch to my local checkout of simplestreams, in an lxc container, run the autopkgtests, and if all looks good I will approve and upload [17:01] bryce: https://people.canonical.com/~dwatkins/simplestreams_0.1.0-30-g3cc8988a.orig.tar.gz <-- that's my orig tarball if you need it [17:01] I'm not attached to it though, so feel free to generate your own. :p [17:02] Odd_Bloke, there's some optional steps that people sometimes do, like stick it in a PPA to verify it builds, or add it to Bileto to demonstrate the autopkgtests will pass. But these are just to help reviewers, and are separate from git ubuntu itself. [17:02] Odd_Bloke: what, you're saying upstream didn't even release a tarball? :-P [17:03] rbasak: You're saying it isn't obvious that people should fetch it from my people.canonical.com account?? ^_^ [17:04] heh [17:04] Odd_Bloke, so yeah this would probably be a case where setting up a PPA for the reviewer would be super helpful, but I'll work from that tarball [17:05] bryce: I'm absolutely happy to do what you would generally expect, to develop "muscle" memory. [17:06] bryce: To which end, I've just uploaded to https://launchpad.net/~daniel-thewatkins/+archive/ubuntu/streams [17:06] Odd_Bloke, ok, so the next step in that case would be to run 'git ubuntu build-source' which should spit out a .changes file you can dput into a ppa [17:07] there is also a 'git ubuntu build' which will locally build the binaries, assuming you have needed dependencies installed [17:07] I usually run git ubuntu build first as a sanity check, then git ubuntu build-source [17:08] ppa looks good [17:09] I hit a couple of different errors when using `build-source`: https://paste.ubuntu.com/p/bCtsPgxzBH/ [17:09] yep it's complaining about the lack of a tarball [17:09] (The source package I uploaded to the PPA is the one that our uss-tableflip tooling produces.) [17:09] and then complaining that a debian directory is already inside the tarball [17:10] bug 1734657 [17:10] bug 1734657 in usd-importer "collision with debian dir on build-source - FileExistsError: [Errno 17] File exists: 'debian'" [High,Triaged] https://launchpad.net/bugs/1734657 [17:10] I think there's a couple ways to work around that, but easiest if you're generating the snapshot yourself is to just exclude it from the tarball [17:11] I guess that's maybe OK for simplestreams, but in general we shouldn't be modifying the orig tarball like that [17:12] rbasak, thus my caveat "if you're generating the snapshot yourself" [17:13] Even then though I don't think it's OK unless you _are_ upstream. [17:13] Because you'd be uploading an orig tarball that isn't actually a snapshot of what it claims to be [17:15] bryce: OK, in lieu of working out how to modify our snapshot script (or repacking the tarball manually), is the PPA as-is sufficient? [17:15] Odd_Bloke, I think so, yes [17:15] OK, cool. [17:15] I've repro'd the debian dir issue here locally [17:17] it occurs for me on the master branch too [17:17] bryce: see the bug I linked above [17:22] rbasak, so what's the workaround? [17:23] bryce: to not use git ubuntu build-source - it's fundamentally broken in this edge case and needs fixing. [17:23] I'm sorry to say that's my general opinion of build and build-source - they are broken in too many edge cases for me to recommend anyone try using them. [17:24] (right now) [17:24] I'm hoping this can all be fixed soon :) [17:24] So I guess in answer to your question: arrange the orig tarball manually and use dpkg-buildpackage -S [17:25] The first error Odd_Bloke found is expected behaviour but could defeinitely do with a more helpful error message. [17:25] (no possible way to get an orig tarball -> no source package build possible) [17:27] ok... so "exclude debian from the tarball" [17:27] No, because that results in an incorrect tarball. You don't want to upload that. [17:28] If upstream put a debian directory in their release tarball, it should be present in the orig tarball in the Debian/Ubuntu upload. [17:28] dpkg can deal with it [17:28] The upload shouldn't be compromised because of a git-ubuntu bug. [17:31] I'm getting called for breakfast. rbasak feel free to carry it from here else I'll continue in a bit. [17:48] bdmurray, i'm about to migrate ubuntu-release-upgrader to git [17:48] bdmurray, could you please review https://code.launchpad.net/~rbalint/ubuntu-release-upgrader/ubuntu-release-upgrader/+merge/376424 before the switch? [17:50] rbalint: Is there anything about the dist-upgrader tarball creation we need to be concerned about? [18:19] rbasak, who should I get in touch to fix mysql 8 related issues? === ricab_ is now known as ricab [19:38] Odd_Bloke, package sponsored, sorry for the delay. [19:39] kanashiro: can you pastebin the error in here? Ping me or Skuggen here. [19:41] cjwatson: around to let me pick your brain? [19:49] bryce: Thanks! [19:54] rbasak, sphinxsearch implements mysql support so you can connect to it using a client like you do with a regular mysql database. The problem is that I can't connect to it using ruby-mysql2 (ruby-riddle tries to do it) after mysql 8, there is no error log [19:54] according to upstream notes this bug was fixed in v3.1.1 (http://sphinxsearch.com/docs/sphinx3.html#version-3.1.1-17-oct-2018) and we have 2.2.11 in the archive [19:55] manticore is a sphinx fork and they fixed this issue in this commit: https://github.com/manticoresoftware/manticoresearch/commit/ed98fdd7230c1d7298e1000fd10812338c6ba698 [19:55] since this is a fork the solution might be similar (or not), need to check [20:02] rbasak: If I wanted to request the addition of a package (simplestreams) to a packageset (ubuntu-cloud), how would I go about doing so? An email to devel-permissions@l.u.c? (And would you consider that a reasonable request?) [20:09] Odd_Bloke, do you mean add a package to the cloud-image Ubuntu seeds? [20:12] if it is, part of the process may involve creating a branch of https://code.launchpad.net/~ubuntu-core-dev/ubuntu-seeds/ and doing a MP. Robie would know the actual procedure better than me though. [20:21] bryce: I mean a packageset change so that ubuntu-cloud uploaders (such as myself :) could upload simplestreams: https://people.canonical.com/~ubuntu-archive/packagesets/focal/ubuntu-cloud [20:24] Odd_Bloke, gotcha, sorry [20:26] bryce: :) [21:10] bdmurray, related to the git conversion? [21:15] Odd_Bloke: yes, email devel-permissions@ please. I haven't thought about in detail but superficially that sounds fine. [21:22] Ack, thanks!