/srv/irclogs.ubuntu.com/2012/01/27/#launchpad-dev.txt

wallyworld_rick_h: i've discovered a fatal flaw in our changes to how we instantiate the YUI instances01:47
rick_hwallyworld_: ruh roh /me hides01:49
wallyworld_rick_h: the change from using a single YUI instance LPS to creating a new YUI instanc eeach time we do a YUI().use(...)01:50
wallyworld_there's code which puts stuff in a well known global namespace01:50
wallyworld_and now each YUI().use(...) is sandboxed01:51
rick_hwallyworld_: right, which code is this? I've been working on moving that code out01:51
wallyworld_so different bits of YUI code cannot see this shared info anymore01:51
wallyworld_there's 2 places - one i'm developing now (and is broken after merging trunk), and also the butask deletion code which is on prod01:52
rick_hwallyworld_: right, that shared stuff needs to become a YUI module and then each use() that needs it has to require it01:52
wallyworld_so the current bug task deletion setup is in form-picker-macros.pt01:52
wallyworld_it attaches a function to a 'lp.app.picker.connect' namespace01:53
rick_hgah, yea that's my task after all this, getting JS out of the .pt files01:53
rick_hwallyworld_: ok, so can that code be moved to lp.app.picker YUI module?01:53
wallyworld_the issue here is that we have all this js in pt files and we need a global namespace to be able to glue stuff together01:53
wallyworld_no, because it's bespoke code01:54
wallyworld_the picker is generic infrastructure01:54
rick_hwallyworld_: so that's why I've got the huge glob of JS code in the layout macro file01:54
rick_hsince that makes it global01:54
rick_hhmmm, but I see, it gets attached to the YUI module ugh01:54
rick_hit's just so wrong...01:55
wallyworld_hmmm. but here, the code is paramaterised according to the config of the field the picker pertains to01:55
wallyworld_and that config is available in form-picker-macros01:55
wallyworld_and nowhere else easily accessible01:55
rick_hsec, let me pull up the code so I can see what it is01:55
wallyworld_rick_h: wanna mumble briefly?01:55
rick_hwallyworld_: yea, give me a sec to dock and reset the laptop01:56
wallyworld_ok. sorry to intrude after EOD01:56
rick_hall good, I was just merging/psuhing out convoy changes to trunk anyway01:56
wallyworld_rick_h: meet you in Yellow 1-1?01:57
rick_hk01:58
rick_hhttp://paste.mitechie.com/show/521/02:16
lifelesswallyworld_: hi02:26
wallyworld_lifeless: g'day02:27
lifelesswallyworld_: I've just replied to your MP about branches; I'm letting you know in case my reply doesn't make sense.02:27
wallyworld_ok. let me have a look02:27
lifelessgeneral pattern of EAFP vs LBYL02:28
wallyworld_your suggestion makes sense.02:28
lifelesscool02:29
lifelessthanks for looking02:29
wallyworld_i was trying to do it in the picker though so that they could simply pick someone else if they misclicked02:29
wgrantAnd this is on +register-merge, right?02:29
wallyworld_yes02:29
wallyworld_the same pattern as we use for bug tasking assigning02:30
lifelesswallyworld_: this would still look like being in the picker, wouldn't it? They click on 'OK', it comes back and says 'nooo'02:30
wallyworld_ie if a use is not a contributer02:30
lifelesswallyworld_: then they can click to really do it, or click on someone else.02:30
wallyworld_there is no ok in the picker, just the link of the person02:30
lifeless bah yes02:31
wallyworld_and they use the picker as part of filling out the form02:31
wallyworld_ie it's just one field02:31
lifelesswell, there are two places you need to do this02:31
lifeless+register-merge and also when you request an additional review.02:31
wallyworld_yes02:31
lifelessoh interesting corner case for you02:32
lifelesswhat if someone in the default review team does not have access to one of the involved branches.02:32
lifelesse.g. they never click on the picker.02:32
wallyworld_so the thinking was not not have them submit the mp with a reviewer who could not see stuff if we could catch it beforehand02:32
wallyworld_hmmm.02:32
wallyworld_that default reviewer would be auto subscribed atm02:33
lifelessorly?02:33
wallyworld_not sure if that's the best thing02:33
wgrantI doubt it.02:33
wgrantWe cannot have autosubscription.02:33
lifelesswallyworld_: or do you mean with your code changes?02:33
wallyworld_no, i mean in my branch02:33
lifelessah02:33
wallyworld_so i think we need a check on form submission as well02:33
wallyworld_as when using the picker02:34
wallyworld_so we catch the issue early if we can02:34
wallyworld_but there's a safety net also02:34
lifelessso, I would expect the default review team to get the subscription (which is near enough to equivalent), but I do think this needs to be vetted, not automatic.02:34
wallyworld_ok, so if there are any visibility issues when the form is submitted, we return and let them confirm02:34
wallyworld_as per your suggestion in the mp02:35
wallyworld_so we attempt to alert them when using the picker, but also catch things at the back end if required02:36
* wallyworld_ has to go to airport to pick up someone, back soonish02:37
wallyworld_lifeless: thanks for the input, much appreciated as always02:37
lifelessde nada02:40
rick_hStevenK: so heads up, convoy stuff is landed in their trunk.02:49
rick_hStevenK: let me know if you need anything else then to look at setting up our packages with the directory traversal/etc02:49
StevenKrick_h: Right, let me look at updating our packages.03:22
wallyworld_wgrant: if i put an exported api  method on IBranchSet, how to a call it from js, using an instance of Y.lp.client.Launchpad()? Using get(...)?04:10
wallyworld_in python it seems to be done as lp.branches.getByUniqueName(...) etc04:12
wgrantwallyworld_: The URL is /branches04:14
wgrantHowever you were calling something on the person URL, use /branches instead.04:14
* StevenK grumbles at his test convoy recipe04:17
StevenK"nest-part packaging lp:~launchpad/convoy/packaging debian" doesn't seem to work :-(04:18
wgrantStevenK: Oh?04:19
wgrantAlso, packaging-only branches are vile.04:20
StevenKbzr: ERROR: Merge-into failed because Source tree does not contain debian.04:20
wgrantDon't do it.04:20
StevenKNo?04:22
wgrantmerge04:22
StevenK:-(04:22
wallyworld_wgrant: yes, i was using a named_get() and thought i needed a context object04:24
wgrantwallyworld_: You should just be able to use a URL. Otherwise construct an object from the URL first.04:26
StevenKwgrant: I guess you want a one branch recipe04:31
wgrantStevenK: No.04:37
wgrantStevenK: Have a branch of trunk, with packaging added.04:38
wgrantCreate a recipe based on trunk with that merged.04:38
lifelesswow05:11
lifeless'The membership status of Allison Randal (allison) in the team Launchpad05:11
lifelessStakeholders (private-canonical-launchpad-stakeholders) was changed by05:11
lifelessthe user himself from Approved to Deactivated.'05:11
wgrantYep05:11
wgrantLongstanding bug about that.05:11
lifelessanyone else see the glaring issue there?05:11
lifelesswgrant: whats the bug # ?05:12
lifelessah 11475305:12
lifelesshmm, https://bugs.launchpad.net/launchpad?field.searchtext=himself does not find it05:12
lifelessI suspect our FTI logic didn't include description at that point, or something05:13
StevenKlifeless: Ugh05:18
StevenK77 references to himself in our code -- most of them in doctests.05:20
StevenKAnd referencing sampledata to boot05:20
lifelesswin06:03
lifelessare you going to JFDI this wart?06:04
StevenKlifeless: https://code.launchpad.net/~stevenk/launchpad/user-himself/+merge/9037606:46
benonsoftwareHello06:47
benonsoftwareI am wondering how do I update my local install of Launchpad?06:47
nigelbrocketfuel-get will work.06:53
benonsoftwareSo just run that in the repo folder and it should work?07:00
nigelbWell, how did oyu setup the first time?07:13
nigelbIf you did rf-setup the first up, rf-get should "Just Work"07:13
benonsoftwarenigelb: Ok, thanks very much07:14
=== wallyworld_ changed the topic of #launchpad-dev to: | https://dev.launchpad.net/ | On call reviewer: - | Firefighting: - | Critical bugtasks: 4*10^2
wgrantstub: We've been running with targetnamecache search disabled for about a year.08:04
wgrantAnd I have a branch to remove the only other use of it.08:05
stubcool08:09
adeuringgood morning08:25
mrevellMorning09:00
=== almaisan-away is now known as al-maisan
=== adeuring changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: adeuring | Firefighting: - | Critical bugtasks: 4*10^2
benonsoftwareWhat does Firefighting mean?09:30
lifelessbigjools: donw09:33
bigjoolslifeless: !  thank you09:33
=== al-maisan is now known as almaisan-away
=== jelmer_ is now known as jelmer
StevenKrick_h: Feel free to merge my approved convoy MP into trunk12:33
rick_hStevenK: was there another one? I merged in your tests_require fix last night12:34
rick_hah, the copyright stuff12:35
rick_hStevenK: sure thing, will do12:35
rick_hwallyworld_: you still around?12:35
rick_hStevenK: ok, merged and pushed. Thanks for that!12:39
=== bac changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: adeuring,bac | Firefighting: - | Critical bugtasks: 4*10^2
wallyworld_rick_h: hello13:21
rick_hwallyworld_: hey, just a heads up that I commented on your MP13:22
rick_hsmall notes on just conventions, nothing big13:23
wallyworld_just saw, thanks. i used what i cargo culted from elsewhere for the wait13:23
rick_hyea, definitely13:24
wallyworld_i didn't want to use a simulate - i was black box testing the method which was supposed to raise the event13:24
wallyworld_so all i needed to do in the test was subscribe to it and see that it was called with the right arguments13:24
rick_hright, cool13:26
wallyworld_i'll look at using wait/resume, although there should still be a timeout for dafety so the test doesn't hang13:26
rick_hit won't13:26
rick_hso what happens is that the test hits and errors that "wait called without resume" and it fails13:27
wallyworld_ok, will look into it13:27
wallyworld_the validation work is a little more problematic - there's no notion ahead of time how many validators there are and what the chaining order should be13:28
wallyworld_with attaching the callbacks to the namespace, it was easy13:29
wallyworld_so i'll give it some more thought13:29
rick_hk13:29
rick_hwell glad at least some of it was easy :)13:30
wallyworld_async events are good but they make some things a lot harder, especially collaboration between loosely coupled components13:31
wallyworld_where you need ordering or other behaviours13:31
wallyworld_how goes the convoy deployment stuff?13:31
rick_hyea, you have to control the flow and document the expected chain so you can figure out what's up when it fails13:31
rick_hwallyworld_: honestly not sure atm. Convoy is done and merged and in trunk and StevenK was putting together an updated package13:32
wallyworld_and controlling the flow is hard when an arbitary component needs to participate13:32
wallyworld_i really want a way to register validation callbacks - that's a common pattern for this type of thing13:33
rick_hwallyworld_: well, it depends. In some ways it's easy to bring in an arbitrary bit just by listening for the same event, but yea, it's harder for that guy to then effect what tohers do13:33
wallyworld_and the global namespace was a convenient point to register them on13:33
rick_hwallyworld_: right13:33
wallyworld_and i want them to fire one after the other, and if one fails, the others don't fire13:34
rick_hwhere's this code at? I'll see if I can look at it and see an easy way to attack it13:34
wallyworld_there is none yet - it's all done by registering the callbacks and looping over them13:35
wallyworld_no event based implementation i mean13:35
rick_hgotcha13:35
wallyworld_i'll have a think about it over the w/e13:36
rick_hcool13:36
rick_hsorry, bit cloud headed this morning due to cold meds so having fun following13:36
wallyworld_i'm off to bed, talk later13:36
rick_henjoy13:36
wallyworld_will do, very tired tonight13:36
=== almaisan-away is now known as al-maisan
deryckMorning, all.14:03
deryckIf anyone needs tips on staying motivated while working from home, I'm happy to oblige.15:00
rick_hderyck: I feel like i need to grow my hair out so I can get styling tips15:01
deryckI'm not sure I could do that anymore, even if grew it out again. ;)15:02
sinzuiYou get use gel to put your hair into tips, but stay away from open flames15:06
benjihmm, fetching any code.launchpad.net URL never finishes for me15:07
abentleyderyck: I'm assuming bug comments and code review comments are unicode.  Let me know if that's wrong.16:09
=== al-maisan is now known as almaisan-away
deryckabentley, that's right.16:18
flacostebigjools: about question https://answers.launchpad.net/launchpad/+question/182545, i'm going to contact non-virtualised ppas owner, that's Archive where purpose = 2 and required_virtualized = False, right?16:34
bigjoolsflacoste: yup16:35
flacostethx16:35
sinzuiadeuring, bac: do either of you have time to review https://code.launchpad.net/~sinzui/launchpad/mailing-list-name-field/+merge/9047916:49
bacsinzui: i do16:49
sinzuithank you16:50
bacsinzui: when you had the pre-implementation call with yourself did you act it out with puppets?  any disagreements?16:52
sinzuiI quite. I wrote one thing 9 hours ago in a somnambulitic state. After 4 hours of sleep, I realised that +review on users was also broken so I told my previous self to fix the base view instead of creating a whole new view to support teams16:54
=== deryck is now known as deryck[lunch]
bacsinzui: i'm confused by TeamAdminisiterViewTestCase -- the team doesn't (appear) to have a PPA or mailing list, so why is the field not there?17:08
sinzuibac: the ppa/mailing lists tests are from the mixin17:09
sinzuiThe tests I wrote verify the fields show to ~admin or ~registry17:10
bacsinzui: gotcha17:11
bacsinzui: based on that it looks good.  i'll finish up shortly.17:11
sinzuibac: these are the tests I converted to the mixin: http://pastebin.ubuntu.com/819036/17:11
sinzuidiffs often no show what was really done17:12
=== salgado is now known as salgado-afk
rick_hallenap: benji can either of you check and make sure this is the changes you decided as the *right* way at the end of yesterday? https://code.launchpad.net/~rharding/wadllib/fix_tests_922599/+merge/9049217:38
rick_hallenap: benji and then while I'm getting you guys should we remove buildout from this since it doesn't work in py3?17:39
rick_hjames_w: ^^ as well17:39
benjiwhy is code.launchpad.net not working for me?  when I look at the source of the page it looks like it's all there; I wonder if there's some JS request that's hanging17:40
rick_hbenji: if there is should be able to see it in the dev tools17:40
SpamapSOOPS-24aa7bfe6e5dbf33996f05a9614d7d0217:40
rick_hbenji: screenshot of the waterfall from the dev tools?17:40
benjirick_h: yeah, that's what I was thinking of17:41
SpamapSI believe this is a known bug...17:41
SpamapSis anybody working on it? (regression where searching for New/Undecided bugs with ubuntu-server as the bug administrator times out17:41
rick_hSpamapS: I don't think I've seen this one recently. adeuring this isn't any of the timeout stuff you were poking at right?17:43
james_wrick_h, I don't think I would change the test. There's a test just below for a stream input, and changing it means that string inputs aren't tested17:43
* adeuring is looking...17:43
SpamapSIts been over a month where ubuntu-server can't triage bugs.. :-/17:44
rick_hjames_w: ok, let me change that back then and see if that didn't help the tests pass for me17:44
SpamapSI had a chat w/ lifeless in here about it.. he did some query plans and showed where it would take 7 minutes or something like that17:44
adeuringno, that's nothing I touched. It's probably not easy to fix, I'm afraid17:45
SpamapSI think we'll have to write a report to be able to triage then.. and just query each package that we supervise individually. :-/17:47
adeuringSpamapS: yes, that could help. ANd actually, the slowest query is just a COUNT(*) for bugtasks related to ubuntu -- but the count is limited to bugs the user can see. Removing this filter should not do any harm17:50
SpamapSadeuring: I'm being told this only happens because I'm in too many teams and so the number of bugs I can see is too high.17:51
adeuringSpamapS: maybe -- but the Ubuntu bug searches tend to time out too often anyway.17:52
jmlgary_poster: I'm (finally!) watching Rich Hickey's talk on simple vs easy. The graph on development speed looks painfully, painfully familiar.17:53
SpamapSCould this bug bug 901122 ?17:53
_mup_Bug #901122: New bug listings need to preload more attributes <bug-columns> <regression> <timeout> <Launchpad itself:In Progress by deryck> < https://launchpad.net/bugs/901122 >17:53
gary_posterjml :-)17:54
gary_posterglad you are watching; hope you enjoy.17:54
adeuringSpamapS: no, that's another issue that might crrep in once the cause for the timeout you are seeing is fixed ;)17:55
jmlgary_poster: yeah, I am enjoying it so far.17:55
gary_postergreat17:55
jmlgary_poster: it's good having such deep thinking clearly expressed17:55
jmlreminds me of the opening SICP lectures in some ways.17:56
sinzuijcsackett, Maybe you need to fix a trivial bug that really irks you17:56
gary_posterjml, clear expression of deep thinking: yes.  Hickey has impressed me with that repeatedly.  he gave another talk advocating hammock-driven development, which I think is related. :-)  I haven't taken time for SICP lectures; probably should.17:59
jmlgary_poster: the sicp book is a little blurred with the lecture in my mind. Am pretty sure it has most of the same stuff and can be consumed about 5x faster18:03
gary_posterheh, cool18:04
gary_posterwill give it a whirl18:04
=== deryck[lunch] is now known as deryck
=== dpm is now known as dpm-afk
rick_hjames_w: ok, updated though line 55 seems like a giant line of hack https://code.launchpad.net/~rharding/wadllib/fix_tests_922599/+merge/9049218:26
james_wrick_h, it does seem to solve the problem though :-)18:27
rick_hjames_w: yea, exactly, ok. I'm going to put this up for review and try to get the powers to review it. I *think* this is the only blocker for the release18:28
rick_hthe other items, fixing the buildout (or removing it) can come later18:28
james_wyeah18:28
rick_hcjwatson: allenap benji any of you guys check out https://code.launchpad.net/~rharding/wadllib/fix_tests_922599/+merge/90492 please?18:34
rick_hand let me know what should be done from here on the buildout side?18:34
abentleyderyck: should BugComments and CodeReviewComments have separate config values for the maximum length before a "Read More..." link?18:34
deryckabentley, I would tend toward saying no, that we should converge.  But if it's too hard to converge -- i.e. on form really needs longer snippets -- I'm fine to have two different configs.18:35
benjirick_h: it looks good to me; want an approval?18:36
abentleyderyck: happy to converge.18:36
rick_hbenji: please18:36
deryckabentley, awesome.18:37
benjirick_h: done with one small suggestion18:38
rick_hbenji: awesome, ty18:39
benjimy pleasure18:39
=== salgado-afk is now known as salgado
jmlgary_poster: I don't think I understand what he's saying about using data for information, rather than objects.18:49
gary_posterjml, it's been awhile since I've seen it now, but he, and Clojure, have an opinion that standard object oriented design has gone off the deep end.  We should have simple data structures that represent our "objects" and not code that has both data and behavior.  Clojure then has a simple data-y concept of type that can be applied to a data structure,  and generic functions and another simpler and faster dispatch syst18:54
gary_posterem.18:54
gary_posterI suspect that's the background for (and/or current best practical example of) what you mentioned but could be wrong.18:55
jmlgary_poster: hmm. interesting. I can begin to see what he's saying but don't grok it the same way as his other points. I wish I knew more about Clojure.18:57
jml(not enough to actually *learn* more right now, you understand)18:57
gary_posterlol, I understand18:57
rick_hanyone know how to tell gpg which key to use to sign a file?18:57
gary_posterrick_h, that sounds suspiciously like the first line to a joke.18:58
jmlrick_h: look into the --local-user and/or --default-key options18:59
rick_hgary_poster: heh, looks like I have to change the --default-key19:00
rick_hjml: yea, that seems to work thanks19:00
rick_hthis whole split private/work life is a mess sometimes heh19:00
deryckrick_h, you mean you try to split work and private life? ;)  you're bordering on lifeless levels of online time. :)19:02
rick_hheh19:03
rick_hderyck: have to make up for being a bit dense :)19:03
rick_hlike where is this magic launchpadlib/bin/py19:03
rick_hbah, that's a buildout reference I bet19:04
deryckyeah, buildout create bin/py.  for most projects.  I don't know how launchpadlib is setup.19:05
rick_hright, the bin/py threw me off hunting. I'm on the right track again I think19:05
rick_hok, maybe not. So the docs say I'm looking for: launchpadlib/contrib/upload_release_tarball.py19:07
rick_hbut my launchpadlib library has no contrib, seperate package?19:08
rick_hderyck: do you know if anyone can upload the pypi package or do I need to get a special user credientials? pypi says the author is: LAZR Developers19:19
deryckrick_h, I don't know.  My guess would be we have a login for LAZR dev.19:20
deryckgary_poster, do you know ^^ ?19:20
rick_hjames_w: so I *think* I've got the 1.3.0 release all set on the LP side, just not in pypi yet19:20
rick_hjames_w: let me know if there's any issues and I can check it out19:21
james_wrick_h, excellent, thanks19:21
rick_hjames_w: sorry for the delay on all that :)19:21
james_wno problem, thanks for catching the bug and promptly fixing it19:21
gary_posterderyck, rick_h, you'll need creds for pypi.  looking at pypi.19:27
rick_hgary_poster: ty19:27
gary_posterrick_h, is your pypi id rick_h?19:29
rick_hgary_poster: no, actually not sure beena while19:31
gary_posterrick_h, tell me what it is and I can give you permissions19:32
rick_hgary_poster: sure thing, getting reminder sent one sec19:32
gary_postercool19:32
rick_hgary_poster: mitechie19:32
gary_posterrick_h, cool.  mitechie has owner permissions19:32
rick_hty much19:32
gary_posterwelcome19:33
rick_hgah, py3 damn you!19:34
rick_hgary_poster: I'll wait a bit to see if something needs time to update, but getting Server response (403): You are not allowed to store 'wadllib' package information19:38
gary_posterrick_h, sounds like you need wadllib too! :-) I only did launchpadlib.19:38
gary_posterrick_h, there is no big switch :-/19:38
gary_posterrick_h, doing wadllib now19:39
rick_hah! sorry yea.19:39
gary_posterrick_h, pypi is not a speed demon.  You are now added for wadllib.  any others?19:42
rick_hgary_poster: ty much, that worked19:42
rick_hgary_poster: not atm, wadllib was all I needed for now19:42
rick_hjames_w: so not sure if you need pypi, but should be up there soon as well19:42
gary_postercool welcome. rick_h I'm a one-stop-shop for a lot of our pypi privs; feel free to ask for more. ;-)19:43
james_wthanks rick_h, we're fine with LP19:43
abentleybac: Could you please review https://code.launchpad.net/~abentley/launchpad/attachment-timeout/+merge/90514 ?19:43
bacabentley: sure.19:44
abentleybac: thanks.19:44
bacabentley: you know it seems we have this conversation about this time every friday!  :)  you're very consistent.19:45
abentleybac: :-)19:45
abentleyrick_h: When you file bugs on Launchpad projects such as wadllib, please also triage them.20:24
bacabentley: why did you decide to put the "Download" link at the top of the message instead of below?  i'm not asserting one is preferable but i was surprised to see it there.20:26
abentleybac: I thought that if you want to download the comment, you probably don't want to read it in the Launchpad UI.20:27
bacabentley: but you'll read the snippet before deciding, perhaps.20:29
abentleybac: The snippet is 3200 characters long.20:32
bacabentley: is the condition at line 557 correct?  shouldn't it be too_long_to_render?20:36
abentleybac: No, the conditional is what was intended.  It seems like downloading is a reasonable option to present for long comments, even if they're not so long that they cannot be rendered.20:39
* bac looks again20:39
abentleybac: Thanks!20:48
=== bac-lunch is now known as bac__
sinzuibac: do you have time to review a follow up to my previous branch: https://code.launchpad.net/~sinzui/launchpad/administer-team/+merge/9053021:14
bac__sinzui: in a bit.  i need to reboot.  and i may disappear due to flaky internet21:14
sinzuiunderstood21:15
=== gary_poster changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: adeuring | Firefighting: - | Critical bugtasks: 4*10^2

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!