[01:00] <SteveA> OMG a pony!
[01:01] <ddaa> SteveA: jamesh: lifeless: meeting about to start
[01:01] <ddaa> once mpool finds the way
[01:02] <ddaa> Meeting starts
[01:02] <ddaa> All things bazaar, yadda yadda yadd
[01:02] <ddaa> * roll call
[01:02] <ddaa> * production status
[01:02] <ddaa> Current focus
[01:02] <ddaa> * Smart server
[01:02] <ddaa> * SFTP advertising (jamesh)
[01:02] <ddaa> * vcs-import knits (ddaa)
[01:03] <ddaa> * supermirror branch browser
[01:03] <ddaa> * private branches
[01:03] <ddaa> * dyson
[01:03] <ddaa> * empty hosted branches
[01:03] <ddaa> Usual end
[01:03] <ddaa> * other meeting actions
[01:03] <ddaa> * critical bugs
[01:03] <ddaa> * pending sysadmin tasks
[01:03] <ddaa> * any other business
[01:03] <ddaa> https://launchpad.canonical.com/BazaarMeetingAgenda
[01:03] <ddaa> * Roll call
[01:03] <ddaa> SteveA is on a pony
[01:03] <ddaa> I'm here
[01:03] <SteveA> hi
[01:03] <jamesh> hi
[01:03] <SteveA> what's the time limit on this meeting/
[01:03] <SteveA> ?
[01:04] <ddaa> Normally until 11:45
[01:04] <ddaa> lifeless: pingus
[01:04] <ddaa> Let's move forward.
[01:04] <ddaa> Production status
[01:05] <ddaa> https://launchpad.net/products/launchpad-bazaar/+bug/53825
[01:05] <ddaa> Still out there
[01:05] <ddaa> Talked with lifeless about how to fix it and updated bug description
[01:05] <ddaa> Would be nice for somebody to do the BzrError -> oops thing soon
[01:06] <SteveA> jamesh will be kinda occupied for the next 2 weeks
[01:06] <jamesh> we need to finish off the plan for doing error reporting from scripts
[01:06] <ddaa> Okay, then I'm happy to keep on spamming launchpad-error-reports for the time being
[01:06] <SteveA> jamesh: that can go on the agenda for next week's infra meeting
[01:06] <jamesh> there is a spec on the wiki about some of the work that needs to be done
[01:07] <mpool> ddaa: lifeless is ill, gives his apologies
[01:07] <jamesh> we should probably move on though, given that we aren't discussing details
[01:07] <ddaa> Last time I spoke to him, he was going to a birthday party ;)
[01:07] <ddaa> * sftp hosting advert
[01:08] <SteveA> birthday parties can do that to you
[01:08] <ddaa> jamesh still supposed to blog about team branches
[01:08] <ddaa> I take it nothing will happen there in the near future unless somebody else gets to do it
[01:08] <jamesh> I've written part of the article.  Will post it to the launchpad list when I've fleshed it out
[01:08] <ddaa> Cool
[01:09] <ddaa> Looking forward to seeing that this week
[01:09] <ddaa> * vcs-imports knits
[01:09] <ddaa> Still waiting for spec review from lifeless and SteveA
[01:09] <ddaa> * supermirror branch browser
[01:10] <ddaa> spiv said it was focusing on that last week and should give us a delivery estimate soon, but is on leave
[01:10] <ddaa> * private branches
[01:10] <ddaa> I think there was no progress there
[01:11] <ddaa> still waiting for lifeless reply on spec review
[01:11] <ddaa> * dyson
[01:11] <ddaa> last week's action:  jamesh to talk to sabdfl to loosen DB constraint, re grass bug.
[01:11] <ddaa> https://launchpad.net/products/launchpad/+bug/53698
[01:11] <jamesh> I talked with him and added a comment to the bug
[01:12] <jamesh> I also talked with Keybuk about whether there was a particular use case for recursively checking for matching tarballs
[01:12] <ddaa> I think that's quite a useful thing to do
[01:12] <ddaa> but maybe not by default
[01:13] <ddaa> Is that on somebody's business for the next couple of weeks?
[01:13] <jamesh> yep.  In the grass case it probably isn't desirable, but it is for the other examples you posted
[01:14] <mpool> what is grass (in this context)?
[01:14] <jamesh> mpool: a GIS package
[01:14] <jamesh> geographical information systems
[01:14] <ddaa> mpool: see the bug, it was just a package with weird release names
[01:15] <mpool> ok
[01:15] <ddaa> that tripped a db integrity check
[01:15] <SteveA> from "Ge" the god of the earth, and "graphical" to do with drawings.
[01:15] <ddaa> So, it looks like nobody will be working on that in the near future.
[01:16] <ddaa> * empty hosted branches
[01:16] <ddaa> there's a bug in bzr about pushing to an existing directory and overwrite contents
[01:16] <ddaa> mpool: are you tracking that?
[01:16] <mpool> ddaa: i'm aware of it
[01:16] <jamesh> is this the case where we get an empty branch dir with no .bzr subdir
[01:16] <ddaa> I had some discussion with john in the bug, but I do not have the reference handp
[01:16] <jamesh> and bzr won't push to it?
[01:17] <ddaa> jamesh: yes
[01:17] <mpool> right
[01:17] <mpool> do you think it's a high priority?
[01:17] <ddaa> Yes, not critical, but pretty important.
[01:18] <ddaa> Since it's very easy to get into a situation where you get a busted bzr branch on the sftp server
[01:19] <ddaa> * Other meeting actions
[01:19] <ddaa> well, this part of the agenda is out of date
[01:19] <ddaa> mpool: you said you planned to do some nice spec braindumping
[01:20] <ddaa> branch viewing, reviews, email subscription, and so on
[01:20] <mpool> ddaa: most of that time has gone into supermirror-smartserver, and the svn roundtripping thing
[01:20] <mpool> i'll upload the smartserver spec
[01:20] <ddaa> Well spent time.
[01:20] <mpool> this is more of a continuing task for me than a particular action
[01:21] <ddaa> ack, will drop the action reminder from the agenda
[01:21] <mpool> we should talk about roundtripping after the formal completion of this meeting, perhaps verbally
[01:21] <ddaa> Time permitting, this meeting is followed by the launchpad weekly meeting
[01:22] <ddaa> * Critical bugs
[01:22] <ddaa> I started work on https://launchpad.net/products/launchpad-bazaar/+bug/37897
[01:22] <ddaa> Plan to fix by making importd upload and retrieve cscvs source branches on a remote sftp
[01:23] <ddaa> That change will also enable a number of other nice things in the medium and long term
[01:24] <ddaa> And it's needed for correctness because the failure mode described in bug 37897 has changed with importd-bzr-native and I did not think of that in advance
[01:24] <ddaa> I had preimpl call with jamesh about that before meeting
[01:24] <ddaa> Will abuse bzrlib's SFTPTransport
[01:25] <mpool> (ok the empty-directory bug is https://launchpad.net/products/bzr/+bug/30576, have marked it for 0.10)
[01:25] <ddaa> and use that as a paramiko wrapper
[01:25] <ddaa> bzrlib.transport.get_transport is the best thing since sliced bananas methink
[01:25] <jamesh> I wouldn't describe it as an abuse
[01:26] <ddaa> Ok, creative reuse.
[01:26] <ddaa> I may get around doing https://launchpad.net/products/launchpad-bazaar/+bug/51130 soon
[01:26] <ddaa> I'm tired of delegating, and I want some fun too.
[01:26] <ddaa> * Pending sysadmin tasks
[01:27] <ddaa> Anybody?
[01:27] <mpool> sysadmins have been beautiful
[01:27] <mpool> we're getting a dedicated little machine for benchmarks
[01:27] <SteveA> cool
[01:27] <ddaa> Yeah, elmo got a haircut and a shave, he's cute like that.
[01:28] <ddaa> mpool: are we getting a win32 machine for running regression tests?
[01:28] <mpool> that would be nice too but i haven't asked for it yet
[01:28] <mpool> are you interested in setting it up?
[01:28] <mpool> hey, you could use buildbot? :-)
[01:28] <ddaa> Yeah buildbot is a nice tool for that.
[01:29] <ddaa> It's what it was designed for.
[01:29] <ddaa> But I'm entirely unwilling to touch a win32 machine
[01:29] <ddaa> j-a-meinel seems the most familiar with that part of bzr
[01:29] <SteveA> off topic
[01:29] <ddaa> right
[01:29] <SteveA> time is short
[01:30] <ddaa> * Other business
[01:30] <SteveA> and so is joe pesci
[01:30] <ddaa> SteveA: how's our new hire?
[01:30] <SteveA> things are progressing well
[01:30] <SteveA> we're scheduling further interviews with martin and with me
[01:31] <mpool> ddaa: we will probably hire not a bzr community advocate, but rather one person across lp and bzr
[01:31] <ddaa> Makes mucho sense.
[01:31] <mpool> so, readjusting the role and criteria for that
[01:32] <ddaa> but I was specifically interested in the fate of the prospective new launchpad-bazaar guy
[01:32] <ddaa> Anybody has other business to discuss?
[01:33] <SteveA> I want to confirm a couple of things
[01:33] <SteveA> get an idea of whether they have happened, or when they might happen
[01:33] <SteveA> is that okay ddaa?
[01:33] <SteveA> how long is the meeting left to run?
[01:33] <ddaa> Absolutely, I'm done with the agenda.
[01:33] <ddaa> We have as much time as you are confortable with.
[01:33] <ddaa> Nominally, 11 mins left.
[01:34] <SteveA> ok
[01:34] <SteveA> 1. bzr native
[01:34] <SteveA> is it in RF?
[01:34] <ddaa> Nah, the last bit is/was waiting for review by jamesh
[01:34] <SteveA> ok
[01:35] <SteveA> 2. python import.  when can we do it?
[01:35] <jamesh> ddaa: I'll try to get to it soon.  pending-reviews also lists your david/cscvs/bzr-native branch as not merged
[01:35] <ddaa> I'm not too hot about doing the python import before we have support for svn renames.
[01:35] <SteveA> why?
[01:36] <ddaa> because it's high profile, and we do not want to give the impression that bzr does not do renames
[01:37] <ddaa> and because it's not something we can fix post hoc without breaking compatibility
[01:37] <mpool> ddaa: this is the kind of thing i was getting into in my mail
[01:37] <mpool> about the priorities for various types of work
[01:37] <ddaa> well, we do do it post hoc without breaking compat
[01:37] <SteveA> ok
[01:37] <SteveA> getting python imported is an important goal for me
[01:37] <ddaa> but what we have already converted will have bad quality
[01:37] <mpool> ddaa: is "we do not want to give the impression that bzr does not do renames" really the real reason?
[01:38] <SteveA> so I'd like a bug opened for that, and a description in there of the things we should do first to enable that
[01:38] <mpool> i mean, "bzr does not convert svn renames" is a much less serious shortcoming
[01:38] <ddaa> jamesh: cscvs/bzr-native needs a fix, thanks for reminding me
[01:39] <ddaa> mpool: SteveA: if everybody is happy with running a python import without renames, I can start it today
[01:39] <mpool> well, is everyone happy with it?
[01:39] <SteveA> I don't understand the thing about renames
[01:40] <ddaa> Will mail you two about it.
[01:40] <ddaa> It's not a difficult decision to make,
[01:40] <SteveA> ok
[01:40] <jamesh> SteveA: if a file gets renamed in SVN, when it gets imported, it will look like an add and a delete, which will affect annotate info, etc
[01:41] <SteveA> so that is all that is holding up python
[01:41] <ddaa> I could find plenty of bad reasons for postponing it, but nothing really serious
[01:41] <jamesh> there aren't that many renames, so it probably won't be visible most of the time
[01:41] <SteveA> I don't think python do a lot of renaming on the mainline
[01:41] <SteveA> so just doit
[01:41] <mpool> i agree
[01:41] <ddaa> okay
[01:42] <ddaa> ACTION: ddaa to import python
[01:42] <SteveA> error reporting
[01:42] <SteveA> 3. error reporting
[01:42] <SteveA> (from buildbot) is this a problem at present?
[01:43] <ddaa> yes
[01:43] <ddaa> I just do not know how to get buildbot to tell launchpad about errors
[01:43] <SteveA> what is the 1 sentence summary of the nature of the problem?
[01:43] <SteveA> why is it important?
[01:43] <SteveA> what user or admin workflow does it negatively effect?
[01:44] <ddaa> Because I want to do something that uses: https://sodium.ubuntu.com/~jamesh/pending-reviews/david/launchpad/importd-ng/full-diff
[01:44] <SteveA> um, affect
[01:44] <SteveA> or whichever
[01:44] <ddaa> Overall, the problem is that users do not see errors.
[01:45] <SteveA> ok
[01:45] <SteveA> so, user adds their upstream svn in launchpad
[01:45] <ddaa> And that it's hard for admin to get synthetic info, though jamesh script can help punctually
[01:45] <SteveA> and then nothing happens
[01:45] <SteveA> because there is an error
[01:45] <SteveA> and the user never sees that there is a problem
[01:45] <SteveA> is that so?
[01:45] <ddaa> SteveA: nothing happens because of the requirement of operator intervention
[01:45] <ddaa> When the operator reload roomba, the import gets tested
[01:46] <SteveA> what is roomba?
[01:46] <ddaa> and if it fails, it gets marked as "test failed" and the use does not know what failed
[01:46] <SteveA> please can we rename things to have less obscure names?
[01:46] <ddaa> SteveA: importd autotest system
[01:46] <mpool> I second that
[01:46] <SteveA> let us call it "importd autotest"
[01:46] <ddaa> that was thoroughly explained in the importd documentation I wrote a few months ago
[01:46] <SteveA> yes
[01:47] <mpool> would it not be possible to have buildbot write to a plain old log file, and just look through that?
[01:47] <ddaa> also, when an import that's syncing starts failing
[01:47] <SteveA> and antidisestablishmentarianism is thoroughly explained in Hobbe's tract of 1643
[01:47] <SteveA> but that doesn't mean it is a good name for something
[01:47] <ddaa> the only way for the user to notice is to see that the branch has gone out of sync
[01:48] <ddaa> Sorry no way to make a one sentence summary of the issues with the import process.
[01:48] <SteveA> ddaa: by "that" do you mean "the thing called roomba"?
[01:48] <SteveA> or do you mean "the problem with error reporting"
 that was thoroughly explained in the importd documentation I wrote a few months ago
[01:48] <SteveA> beware of using pronouns on irc
[01:48] <ddaa> that = "an import that's syncing starts failing"
[01:48] <SteveA> much possibility of misunderstanding
[01:49] <SteveA> I see
[01:49] <ddaa> an import that's syncing is run by hoover BTW
[01:49] <SteveA> I'm sorry.  I thought you were saying:
[01:49] <SteveA> "roomba is documented somewhere, so it is a good enough name"
[01:49] <mpool> can we back up for a sec
[01:49] <mpool> is the problem that "the system won't automatically tell users what's wrong"
[01:49] <mpool> or "it's too hard for david to find out what's wrong?"
[01:49] <ddaa> yes, but not only
[01:49] <mpool> or a bit of both?
[01:50] <ddaa> it's easy for me to find what's wrong usually, at least when the answer is easy
[01:50] <ddaa> the problem is that "the system does tell users that something is wrong"
[01:51] <ddaa> and that "the system relies on ddaa to babysit it so new user entered data turns into an import"
[01:51] <mpool> ok..
[01:51] <mpool> and would it be enough to just tell the user broadly what happened
[01:51] <ddaa> The hard be for me is to collate failures to get a sense of what are the important problems.
[01:51] <mpool> like a one-sentence description
[01:51] <ddaa> Nope.
[01:51] <mpool> "failed because of an error while converitng", "couldn't contact server", etc
[01:52] <mpool> what kind of thing do you want to tell them?
[01:52] <mpool> actually we're running out of time
[01:52] <mpool> should we continue or defer it?
[01:52] <ddaa> Please ask me on ML
[01:52] <mpool> ok
[01:52] <mpool> steve, anything else
[01:52] <ddaa> that's a complicated issue and I would have to do some background research to give good answers
[01:54] <ddaa> Meeting closed then
[01:54] <ddaa> Need to take a rest break before the launchpad meeting.
[01:54] <mpool> sure