[04:03] <blr> wgrant: how are the vocabulary strings resolved?
[04:05] <blr> e.g. ProductSeriesView.branch exports a RefrenceChoice with attrib vocabulary='BranchRestrictedOnProduct' ..
[04:28] <wgrant> blr: An IVocabularyFactory with the same name.
[04:28] <wgrant> You'll see securedutility ZCML directives for that registration.
[04:30] <blr> ah, yes I should probably remember to consult zcml config whenever something appears magica
[04:30] <blr> magical
[04:34] <wgrant> Yup
[04:35] <wgrant> Magic usually means adapters, and adapters are registered in ZCML.
[11:43] <cjwatson> blr: I picked up https://code.launchpad.net/~blr/launchpad/bug-1334577-verbose-diff/+merge/243751 and posted an updated review.  Should be possible to finish this up without too much more work, hopefully by end of month.
[11:45] <cjwatson> wgrant: Any chance you could find a bit of time to look over https://code.launchpad.net/~cjwatson/launchpad-buildd/avoid-pbuilder/+merge/261989 ?
[13:49] <xnox> cjwatson: wgrant: who is ~foli and why does he mess with ~ubuntu-dev members =)
[13:50] <xnox> cjwatson: wgrant: would you be able to PM me his email address (none are published on the launchpad page) to raise an issue with his actions w.r.t. ACLs of ~not-canonical per package uploader =)
[13:50] <cjwatson> xnox: He's a Canonical sysadmin
[13:51] <cjwatson> xnox: More detail, please?
[13:51] <xnox> cjwatson: so james hunt was deactivated from ~ubuntu-dev by foli, yet james hunt is a per-package uploader for e.g. upstart approved by DMB. And as far as I know DMB controls ~ubuntu-dev.
[13:52] <xnox> seems like too much clean-up was done during exit procedure?
[13:52] <cjwatson> I suspect that's an overenthusiastic script somewhere, indeed
[13:52] <xnox> hence i wanted to email DMB about it, and CC him, to possibly give more insight or correct things.
[13:52] <cjwatson> PMed
[13:52] <xnox> thanks
[14:00] <cjwatson> I'm reasonably sure you can just add back and it won't be done again, mind you
[14:00] <cjwatson> But I just found the whitelist, and it contains ubuntu-core-dev but not ubuntu-dev, so that's almost certainly an error
[14:01] <xnox> cjwatson: sent an email to DMB, with cc of james and the two people you have PMed me. Board also has enough canonical people to check that everything is in order.
[14:01] <cjwatson> xnox: Might want to reference lp:canonical-is-scripts blessed/leavers/leaver.csv (which you won't be able to see directly, but I'm pretty sure that's the whitelist)
[14:01] <xnox> cjwatson: meh, i'd rather be explicit such that things can be checked and corrected. I thing on grand scheme of things, this is trivially minimal thing.
[14:01] <xnox> haha, ok.
[14:02] <cjwatson> xnox: I'm pretty sure this is just because PPU is relatively rare so IS didn't previously notice that ubuntu-dev needed to be whitelisted directly
[14:02] <xnox> true.
[16:00] <cjwatson> hmm
[16:00] <cjwatson> I think I might have accidentally put together a halfway reasonable side-by-side diff implementation in a couple of hours of effort
[17:54] <cjwatson> hloeung: Do you need help finishing up https://code.launchpad.net/~hloeung/launchpad/openpgp-show-fingerprint/+merge/244614 ?
[22:46] <blr> wgrant: morning
[23:05] <blr> wgrant: added new browser tests (using BrowserTestCase rather than doctests) for product views - we didn't appear to have any.
[23:15] <hloeung> cjwatson: yeah I do actually. I'm not even sure where to begin fixing those xx*-pdf.txt as mentioned
[23:15] <hloeung> cjwatson: or ensure that my change doesn't break them
[23:16] <wgrant> blr: Morning.
[23:16] <wgrant> blr: BrowserTestCase is good for testing general cases, but lots of edge cases are nowadays better checked using create_initialized_view.
[23:16] <cjwatson> hloeung: Do you have a working test environment?
[23:16] <wgrant> blr: How close is it all to finished?
[23:17] <hloeung> cjwatson: nope, I haven't done any LP development before
[23:19] <cjwatson> hloeung: Would you like to get this sorted out so that you're better set up for the future, or do you just want to throw it over the wall to us and let one of us finish it off?
[23:19] <blr> wgrant: have an issue with my RestrictedGitRepositoryVocabularyOnProduct test, trying to resolve now (everything else is finished)
[23:20] <blr> cjwatson: thanks very much for looking at the verbose diff branch, will be good to get that squared away
[23:21] <hloeung> cjwatson: I think it might be best if I throw it over the wall. That is if one of you guys are willing to take over it
[23:22] <cjwatson> hloeung: sure, I can finish it up.  It's basically a matter of grepping for pks/lookup and applying the obvious URL changes
[23:23] <cjwatson> hloeung: test_web doesn't fail, but the other mentioned ones do
[23:25] <cjwatson> (But test_web should be adjusted anyway to make sure that testkeyserver can cope with the new form)
[23:27] <cjwatson> Anyway, I'll submit a fixed/updated version tomorrow
[23:28] <hloeung> cjwatson: thanks
[23:29] <wgrant> cjwatson: Oh, fmt:ssdiff, nice.
[23:33] <cjwatson> wgrant: http://people.canonical.com/~cjwatson/tmp/ssdiff-1.png
[23:36] <wgrant> cjwatson: Did you consider splitting the code into two halves to make it more comprehensible? Specifically, have one function which yields categorised lines with diff and file line numbers, and then another which can just process them without all the weird continue stuff.
[23:36] <blr> wgrant: cjwatson: point of confusion regarding valid repositories for a product, in the bzr case the vocabulary validates product.series.branch, however the git case sets the default repository - what property should have the vocab?
[23:36] <wgrant> blr: The form field must have the vocab directly.
[23:36] <cjwatson> wgrant: That might not be a bad idea.
[23:36] <cjwatson> generators++
[23:37] <wgrant> cjwatson: fmt:diff can probably be simplified based on that too.
[23:37] <wgrant> Though it wouldn't actually use the in-file line numbers.
[23:37] <cjwatson> They're structurally pretty similar, yeah.
[23:37] <blr> wgrant: ah, so I _don't_ want a copy_field, but should just set the vocab on the TextLine
[23:37] <wgrant> blr: Indeed, there's no corresponding model field for git, because it's controlled by IGitRepositorySet.setDefaultRepository
[23:37] <blr> right
[23:38] <blr> that makes sense, thanks.
[23:38] <wgrant> blr: It wouldn't be a TextLine, but rather a Choice or something, I think.
[23:38] <wgrant> blr: It should be just like ProductSeries.branch, except directly on the view.
[23:38] <wgrant> cjwatson: I think the tests become a good bit less ugly when they can test edge cases without parsing HTML...
[23:38] <wgrant> Though some of them do need to.
[23:39] <cjwatson> Indeed.
[23:39] <cjwatson> Could be fleshed out a bit without dying of boredom.
[23:40] <wgrant> I'd probably even split out a third layer which yields (type, diff_lineno, added_lineno, removed_lineno, added_text, removed_text) without doing HTML, but that's possibly little benefit once the complex parser is split out.
[23:40] <cjwatson> I'm going to claim that this was all part of the inline comments improvements requested by stakeholders, and not an epic exercise in procrastination, honest.
[23:40] <wgrant> Heh
[23:40] <wgrant> There is either a bug or a stakeholder request about side-by-side diffs.
[23:40] <cjwatson> Both.
[23:41] <cjwatson> I assigned both of them to me once I realised it was tractable.
[23:42] <cjwatson> (loggerhead does the transformation in JavaScript, I think, which is sort of arguably a better UX since you can switch back and forward without reloading; OTOH at least loggerhead's implementation is dreadfully slow and the query string isn't in fact too bad.)
[23:43] <cjwatson> (Also we could still flip it with AJAX later if we wanted.)
[23:48] <cjwatson> blr: I also reviewed https://code.launchpad.net/~blr/launchpad/bug-114705/+merge/243320, just in case that's stuck way back in your inbox.
[23:52] <blr> cjwatson: almost felt like leaving that open given who reported it :(
[23:59] <cjwatson> yes ...