[01:52] <thomi> wgrant: lots of actions on these pages send emails, which cause tracebacks like  http://pastebin.ubuntu.com/10419041/ - is there an easy way to get those emails to be delivered to a local account... or maybe just not sent at all ?
[01:53] <wgrant> thomi: install a postfix or similar in your container, configured for local deluvery only
[01:54] <wgrant> the emails are all sent to root@lovalhost in dev mode.
[01:54] <thomi> wgrant: ahhh, ok
[01:54] <thomi> wgrant: you recommend postfix? I on;y have experience of exim, and that was *years* ago
[01:55] <wgrant> well, postfix is whwt i know best, and debconf provides a one-click local-only option
[01:55] <wgrant> but exim will work fine too
[01:56] <thomi> nice
[01:57] <thomi> I remember exim being a PITA
[02:00] <StevenK> thomi: 3 or 4?
[02:02] <thomi> StevenK: hmm... it's been long enough that I can't remember. It would have been  ~ 15 years ago, so....
[02:02] <thomi> maybe 3?
[02:14] <thomi> wgrant: comments on placement of 'edit mailing list subscription' link? https://bugs.launchpad.net/launchpad/+bug/1425646/+attachment/4327910/+files/Screenshot%20from%202015-02-26%2015%3A11%3A55.png
[02:14] <mup> Bug #1425646: It's too hard to find the +editemails page. <trivial> <ui> <Launchpad itself:New> <https://launchpad.net/bugs/1425646>
[02:14] <thomi> not sure if it should go there, or in the global navigation panel on the RHS
[02:37] <blr> thomi: seems okay, although isn't strictly speaking 'user information'
[02:37] <thomi> blr: yeah.. OTOH, if it's over on the RHS then it's separated from the other email-related settings
[02:38] <thomi> so I'm not sure what's more usable... maybe both? no, that seems like a bad idea...
[02:38] <blr> thomi: would almost be nice to have something analgous to Latest Memberships for lists
[02:38] <blr> but that's a bit out of scope
[02:38] <thomi> yeah
[02:38] <thomi> also, you tend to end up subscribed to a lot of lists.. or at least have a lot of mailing lists in that list
[02:38] <blr> better there than on the RHS I think
[02:40] <blr> the wording of the options on the menu could use improvement, it really isn't clear what the difference between "administer" and "administer account" is without loading the views
[02:41] <blr> doesn't help that the view headings are different from the links!
[02:42] <blr> adminster => review person, adminster account => review person's account
[02:42] <thomi> blr: agreed, but ... also not in scope for what I'm looking at right now :D
[02:42] <blr> just sayin' :)
[02:42] <thomi> yeah
[02:42] <thomi> I think at least one of those aren't seen by mere mortals though
[02:42] <blr> the UI stuff is a big rabbithole
[02:45] <wgrant> Administer options aren't shown to mortals, yeah
[02:45] <wgrant> we need tor rrdesign the settings things entirely
[02:45] <wgrant> there should be a Settings link which has a submenu yo all of the settings
[02:55] <thomi> updating doctests :(
[02:56] <wgrant> 1:(
[02:59] <thomi> hmm, I think we need a link at the top of each of +editemails and +editmailinglists that points to each other
[03:00] <wgrant> sonewhere on each, yeah
[03:15] <thomi> wgrant: ever thought of upgrading ./bin/test to produce subunit v2 output? (with an appropriate option, of course)
[03:15] <thomi> combined with trv, it'd make these results a lot easier to read IMO
[03:22] <wgrant> thomi: It can do subunit v1, but I think our subunit's too old for v2.
[03:22] <thomi> wgrant: what's the barrier to upgrading it?
[03:22] <wgrant> thomi: Mostly not wanting to break things that might depend on its output, eg. buildbot.
[03:23] <thomi> wgrant: we can upgrade it and have it spit out v1 by default
[03:46] <wgrant> thomi: That link looks good. Does the page work?
[03:47] <thomi> wgrant: of course!
[03:47] <thomi> I'm just adding the links between the two page snow
[03:47] <thomi> *pages now
[03:47] <thomi> then will add tests for those links
[03:48] <wgrant> :)
[03:48] <wgrant> And the new page doesn't require reauthentication?"
[03:48] <thomi> nope
[03:48] <wgrant> Marvellous.
[03:48] <thomi> yeah, it's pretty sweet :D
[03:48] <thomi> I'm sure I've done something stupid though
[04:02] <thomi> hmmm, I'm really not sure where to put these additional links.
[04:02] <thomi> I'll leave it for today, and come back to it tomorrow.
[13:49] <cjwatson> wgrant: https://code.launchpad.net/~cjwatson/launchpad/git-defaults/+merge/250474 should be reviewable again, in the unlikely event you're still vertical.
[17:18] <cjwatson> wgrant: And https://code.launchpad.net/~cjwatson/launchpad/git-lookup/+merge/250628 as well now, phew.  Fairly substantial redesign, ended up being able to throw away most of IGitNamespaceSet.
[19:24] <thomi> cjwatson: still around perchance?
[19:28] <cjwatson> thomi: little bit
[19:30] <thomi> cjwatson: I read your comment on bug 1425646 - I added a screenshot with both links as text links... what do you think?
[19:30] <thomi> I'd love to get this finished today (as I have to go work on CI next week) - I also added to that bug screenshots of the mailing list & email settings pages
[19:30] <thomi> so, feedback desired - blr and wgrant too ^^
[19:30] <mup> Bug #1425646: It's too hard to find the +editemails page. <trivial> <ui> <Launchpad itself:New> <https://launchpad.net/bugs/1425646>
[19:34] <cjwatson> sure - commented
[19:34] <cjwatson> (with the caveat that I'm relatively new to doing serious LP UI work)
[19:35] <thomi> cjwatson: thanks. My eye for resign is... lacking I'll try that out
[19:35] <cjwatson> yeah me too :)
[19:35] <cjwatson> what's the equivalent of a tin ear?
[19:35] <thomi> heh
[19:36] <thomi> cjwatson: something like https://launchpadlibrarian.net/198833330/colin_suggestion2.png ?
[19:38] <cjwatson> I think I prefer that personally.  But if others feel that uses too much vertical space then maybe the links can just be spaced out a bit more horizontally.  My main problem was that they were bunched up right against each other, I think.
[19:38] <cjwatson> Oh, I would put "Change e-mail settings" first since it's directly associated with the list of addresses just above.
[19:41] <thomi> good catch
[19:41] <thomi> I won't upload a new SS - I'm already spamming the bug
[19:42] <thomi> blr: any thoughts on the above?
[20:19] <blr> thomi: that looks better - I think we're kind of stretching the semantics of <dl> and <dt> there as well, they're not really definitions (arguably should be <Hn>s with the colons removed)
[20:20] <blr> you'd have to fix the css up if you changed that though
[20:20] <thomi> blr: and if we fix it in one place, we have to fix it everywhere
[20:21] <thomi> what is <hn>?
[20:21] <blr> <h1> <h2> etc
[20:22] <thomi> oh, right
[20:23] <blr> I wouldn't worry about it, from the user perspective they're jus tall block elements, not important.
[20:23] <blr> if we get some design resource it might be worth going back and addressing that kind of thing
[20:24] <blr> cjwatson: the create-api branch is ready for re-review if you have a moment to look over it.
[20:28] <cjwatson> Right, K+J just got back from karate so I'm about to be distracted so let's see
[20:30] <blr> cjwatson: no worries, will have the ref list bit for you soon too.
[20:30] <cjwatson> oh cool
[20:31] <cjwatson> I spent all of today rearranging my defaults/lookup branches
[20:31] <cjwatson> but they feel more sensibly-structured now
[20:49] <cjwatson> blr: r=me with comments
[20:50] <blr> cjwatson: thanks colin :)
[20:51] <blr> cjwatson: the print statements are placeholders until I sort out some proper logging really
[20:51] <cjwatson> Yeah, but they might as well be good placeholders. :-)
[20:51] <blr> that is true!
[20:59] <wgrant> cjwatson: Excellent, a good redesign was what I thought might work well.
[21:00] <wgrant> Will rereview today.
[21:06] <blr> wgrant: did you get any feedback from the juju devs yesterday?
[21:06] <blr> I missed the tail end of it
[21:08] <wgrant> blr: Juju should apparently be detecting that the instance failed to start and trying the other AZ, but it doesn't wait quite long enough.
[21:08] <wgrant> https://bugs.launchpad.net/juju-core/+bug/1425808
[21:08] <mup> Bug #1425808: OpenStack provider doesn't try another AZ if the scheduler fails to find a valid host <juju-core:Triaged> <https://launchpad.net/bugs/1425808>
[21:10] <blr> wgrant: great
[21:33] <thomi> wgrant: can you please remind me what to do to get a branch scan completed? I thought it was a script in lp-dev-utils?
[21:33] <cjwatson> thomi: http://paste.ubuntu.com/10436489/
[21:33] <cjwatson> it's not in lp-dev-utils
[21:33] <thomi> ahh
[21:34] <thomi> can we add that to lp:launchpad ?
[21:34] <thomi> or ... I guess if the problem is going to go away soon
[21:34] <thomi> maybe it's not worth it
[21:34] <wgrant> I just Ctrl+R lp-shell, Ctrl+R resc, then fix the branch name :)
[21:34] <wgrant> And no, that sort of code doesn't belong in lp:launchpad. lp:lp-dev-utils maybe.
[21:36] <wgrant> cjwatson: Hm, is git-defaults meant to be half the length it was?
[21:40] <cjwatson> wgrant: Sounds about right.  Removed a fair bit of cruft, and I applied some mixin refactoring to the tests to let them be a lot less wordy.
[21:40] <wgrant> Yeah, looks good, just surprising.
[21:41] <cjwatson> I reckoned it was a good sign.
[21:41] <cjwatson> git-lookup becomes a bit longer, IIRC, but not too much.
[21:43] <cjwatson> Oh.  It becomes slightly more longer than git-defaults became shorter.  Oh well.
[21:44] <wgrant> Heh
[21:44] <cjwatson> Not directly related, it's just that I tore a bunch of stuff out of git-namespace.
[21:47] <wgrant> Yeah
[21:48] <wgrant> git-defaults is muuuuch cleaner now, thanks for reworking it.
[21:49] <cjwatson> I almost wonder whether I needed to do PersonDSP at all.  But it'll probably still be handy for browser stuff.
[21:49] <wgrant> I think it may become sensible to make it exist only for browser purposes, yes.
[21:49] <wgrant> Handing around an owner and a target isn't terribly onerous most of the time.
[21:50] <wgrant> It is probably easier to excise once this series is landed, though.
[21:50] <wgrant> Which is why I didn't request its death earlier.
[21:50] <cjwatson> Yeah.  Ditto the anomalous argument order of get_git_namespace.
[21:53] <cjwatson> Anyway, family time, night.
[21:53] <wgrant> Night.
[21:53] <cjwatson> Thanks again for the review.
[21:53] <wgrant> Thanks for not telling me to get stuffed when I ask you to basically rewrite it :P
[21:53] <cjwatson> Heh
[21:54] <cjwatson> Rough direction helps.  I think my cognitive error had been thinking that all traverse-style methods had to have roughly the same signature, and in particular a single return value.
[21:55] <wgrant> Which would usually be a good policy, but it makes things so much uglier here that breaking with convention probably works well.
[22:01] <wgrant> It's a similar problem to series, really. We're trying to handle two dimensions in a single object for no clear reason.
[22:04] <thomi> I wonder if I could get a review on https://code.launchpad.net/~thomir/launchpad/devel-fix-editemails-link/+merge/251176 please?
[22:04] <thomi> I'm sure I've done something stupid.
[22:05] <wgrant> thomi: Sure. Just rereviewing a couple of Colin's big ones, then I'll look at yours.
[22:05] <wgrant> I doubt it :)
[22:05] <thomi> lol - we'll see