').append(result.getHTML(attribute)).getContent(); [02:24] I couldn't see a way to get a string representation directly. [02:24] wgrant: i'm trying something atm. give me a minute [02:25] So, erm, bugs and blueprints use different AJAX textarea widgets? [02:26] I wasn't aware we had more than one :/ [02:26] poolie: perhaps [02:27] poolie: I think there is great value in actually reaching a goal [02:27] poolie: it will take considerable time to turn around the perception that lp is slow [02:27] poolie: firstly we need to make it not slow, and then we need to keep it that way for a while ;) [02:27] (Windmill didn't catch this breakage, FWIW) [02:29] booyah [02:30] https://bugs.staging.launchpad.net/ubuntu [02:30] its not quite subsecond [02:30] but not terribly far off [02:30] I am disappoint. [02:30] yeah, i do also agree with hammering it thoroughly flat [02:30] just sayin [02:32] wgrant: this is what we want [02:32] - return Y.Node.create(value); [02:32] + var result = Y.Node.create(""); [02:32] + result.setContent(value); [02:32] + return result; [02:33] wallyworld: How is that different? [02:33] Apart from adding another layer, it looks like it should be the same. [02:33] Oh. [02:33] Does it get an innerHTML if it's a single element, whereas this is a sequence of

s? [02:34] wgrant: it's done in the getHTML method, not the _defaultFormatter method [02:34] wgrant: i think the getHTML method was constructing the node incorrectly [02:34] with the above change, there's no need to change _defaultFormatter [02:35] I consider this timing a success: 03976-04169@SQL-launchpad-main-master SELECT BugTask.assignee, BugTask [02:36] wallyworld: A span is not the right thing, but we do need something like that. [02:36] Ugh, this is hard to do correctly. [02:37] good things often are [02:37] wgrant: the span is just a container - it doesn't come out in the final html [02:44] *blink* 40 / 54 BugTask:+index [02:45] that's good right? [02:46] yeah [02:46] was sitting at 180/* before [02:46] I'm wondering wtf changed [02:53] thumper: I really wouldn't tackle what you're working on by blocking branch renames. [02:54] thumper: Its up to you of course, but I fear it will just decrease usability and increase user frustration [02:54] lifeless: I'm not [02:54] thumper: oh cool; what are you doing? [02:55] lifeless: I'm trying to catch the object modified event for the branch and checking the unique name [02:55] lifeless: creating a rename job if the unique name has changed [02:55] and there are branches stacked on it [02:55] however, since the unique name is a string col that is updated with a db trigger [02:55] it isn't being reloaded on Store.flush [02:55] yeah [02:56] I could easily use old skool logic [02:56] which would work [02:57] thumper: _mark_autoreload [02:58] is related [02:58] obj_info.variables[column].set(AutoReload) [02:58] looks private(ish) though [02:58] IIRC you do this: [02:58] obj.column = AutoReload [03:00] thumper: did you consider making the stacking location be id based rather than unique-name ? [03:00] thumper: and rewriting all the current branches Just Once to stack by id ? [03:01] lifeless: I don't know how do work that well for HTTP access, or enough in the smart server to do it right [03:02] thumper: so for http its pretty easy - we already export branches by id in the dc; we just need to make a matching export to the world with the existing branch-privacy logic attached to it. [03:03] for the smart server, we have a hook already that mangles the stacked on location on branch open - a 95% solution would just decorate the set_stacked location in the backend [03:05] actually that may well be a better solution than what I'm doing [03:05] bugger! [03:05] I suspect its a little more fiddly, but much more robust - no race conditions, immune to any changes we make to branch names in the future etc [03:05] yeah [03:06] wgrant: https://code.launchpad.net/~wallyworld/launchpad/text-area-widget-html-fix/+merge/55860 [03:08] * lifeless forces himself away from coding [03:08] lifeless: I think what we need is a special traversal hook, like +branch-id [03:08] wallyworld: I don't quite understand what the root problem is. Is it that we can't get the innerHTML from a Node created from a fragment, instead of a single element? [03:08] lifeless: so we could stack on /+branch-id/12345 [03:08] internally [03:08] not sure what it would look like to the outside world [03:09] wgrant: no exactly. innerhtml assumes that there is an element eg

, etc wrapping the content [03:09] thumper: I think exporting that to the outside world would be fine [03:09] wgrant: and so innerhtml returns what's inside that containing node [03:10] thumper: that is the heart of the solution (for working with http and sftp) [03:10] yeah... [03:11] unfortunately I think I'm probably the most qualified to actually implement this [03:11] given jml and mwhudson not actively coding on LP now [03:11] wallyworld: This is probably a hack, but it seems to work. [03:11] at least I understand much more of it now since my sloecode workl [03:12] wgrant: not sure if it's a hack or not. seems to match what i've seen elsewhere. but i'm not expert [03:23] Project windmill build #128: STILL FAILING in 1 hr 11 min: https://lpci.wedontsleep.org/job/windmill/128/ [03:26] wallyworld: You've tested a good variety of (potentially duplicated) widgets? [03:27] wgrant: i tested bugs, blueprints and am about to double check recipes [03:28] Great. [03:28] We have more than two hours to confirm everything's really working. [03:31] huwshimi: have you formed an opinion of the my revision to the mailing list archive header? [03:31] sinzui: ah sorry, yes, I think it's good [03:32] sinzui: Again it would be nice to have this fully integrated, but let's go with this for now [03:32] sinzui: At least it's consistent with Loggerhead [03:33] huwshimi: Then are the wiki's next? The Lp primary css is now usable by other Lp wsite [03:34] sinzui: The spacing under the "$TEAM team mailing list archive" seems to be a little small, but it looks good to me too. [03:35] sinzui: I'm not sure what to do with the wikis. At the least the help wiki should probably look more like Launchpad, but it's difficult to know exactly what to do when the auth system etc. isn't integrated. [03:36] sinzui: We have deeper issues about navigation etc. that need to be resolved to help us with that [03:38] wgrant: I thought so to. I suspect that lose of the image changed the positioning. I can ad some pixels [03:38] of white-space [03:38] sinzui: Is the title an h1? [03:39] thumper: textareawidget - if hide_empty=True, then there's no way to enter text apart from using the html form. why do we have a hide_empty? [03:39] eg sourcepackagerecipe index page [03:39] Was that for the recipe description? [03:39] wgrant: yes [03:40] Most recipes don't have descriptions, so it took up too much room for too little gain, IIRC. [03:40] but if you wanted to enter a description later... [03:40] and it doesn't take up *that* much room [03:40] Then you're screwed, yeah. [03:41] It could be like the MP's commit message widget, I guess. [03:41] i mean, in the context of what else is on the page, it's stuff all wasted space [03:41] That's just a single line when empty (an add button, not a normal textarea widget) [03:41] yes, it is for the commit message and description [03:41] it has been suggested that we integrate the menu items into the widgets themselves [03:41] and have the code in one place for showing and hiding [03:41] That would be very sensible. [03:41] but I didn't get that far [03:42] thumper: and now you're desserting us [03:42] or deserting, i can't spell [03:42] wallyworld: I'll leave it in your capable hands :) [03:42] * thumper is eating dessert [03:42] * wallyworld lols [03:49] * wallyworld needs food. dessert even [03:54] I could sit here and play with the blueprint implementation status widget all day. The way the other fields magically change... [03:56] bacj shortly, vet run [03:56] wallyworld: lifecycle_status was relying on getHTMLOrDefault? [04:11] wgrant: yes. there is no html representation for that field [04:48] thumper: bug 745230. it that really a bug or do they just need to edit their recipe when they rename a branch? or perhaps there was already a build pending when the branch was renamed, hence the error? [04:49] <_mup_> Bug #745230: recipe fails when branch gets new owner < https://launchpad.net/bugs/745230 > [04:49] wallyworld: Ah, I didn't realise a bug was filed for that. [04:49] wallyworld: It is the stacked-on-a-renamed-branch bug. [04:51] Is there any way to make my local launchpad dev server accessible over my network? [04:51] wgrant: ah yes. i knew about that one. i didn't pick up that was at play here though. it looked like just a simple branch rename and a recipe referring to the old branch [04:51] huwshimi: https://dev.launchpad.net/Running/RemoteAccess should still mostly work. [04:51] huwshimi: You're not going to try and unbreak some of our JS in IE, are you? [04:52] huwshimi: The other thing you can do is 'utilites/ec2 demo' [04:52] i reckon we shouldn't even attemp to support IE < 9 :-) [04:52] wgrant: Would I dare? :P [04:52] StevenK: That was broken at one point. Not sure if it still is. [04:52] why go to all the bother? [04:52] wallyworld: Sadly we have to :( [04:53] why? sure our lp users are smarter than joe sixpack who swallows whatever dross ms dishes up with windows? [04:53] s/sure/surely/ [04:53] Ha ha ha. [04:53] It's not quite that simple. [04:54] how so? [04:54] especially with ie 9 out now [04:56] wallyworld: Some people work in corporate environments that won't upgrade their browsers for various reasons. We have a significant number of stakeholders that fall into that category. [04:56] huwshimi: how should i triage bug 744808? i'm thinking medium? or high? [04:56] <_mup_> Bug #744808: Redrawing of Bug View page very slow in Firefox < https://launchpad.net/bugs/744808 > [04:57] huwshimi: yes, i realise corporates still use ie 6 - i used to work at one. i didn;t realise there were that many of them [04:57] or that they couldn;t install ff for lp use [04:58] even at my old employer where ie 6 was the standard, we could still install ff etc. i went one step further and used linux :-) [04:59] wallyworld: I don't know how you should classify that bug. We're going to need some more info about browser version etc. [04:59] huwshimi: It's an nvidia driver bug :/ [04:59] huwshimi: Well, plus Firefox, I guess. [04:59] But all the other drivers do fine. [04:59] So I tend to blame nvidia. [05:00] wgrant: If it's just one image that is cause it there are things we can do. [05:05] huwshimi: i'll mark it as medium i think then [05:06] wallyworld: lifeless will hate you forever. [05:06] wallyworld: I don't think we use medium [05:06] Medium doesn't exist :) [05:06] wallyworld: It should probably be high [05:06] high it i [05:06] is [05:06] oh, i forgot about not using medium [05:08] wallyworld: the rationale is this: high is nominally 6 months worth of work queued; with reviews to winnow it; anything below that is simply 'unscheduled' [05:09] thanks for the reminder [05:11] jtv: do you think 743913 is a duplicate as suggested in the bug report? [05:11] wallyworld: looking [05:11] Bug #743913 [05:11] * wgrant pokes _mup_ [05:11] mup has been a bit slack of late [05:14] wallyworld: that does look like it's probably the same issue—even though it's not a simple one and the individual reporters may just as well have run into different sides of the problem. [05:15] jtv: thanks. i'll mark it as a dup [05:18] jtv: dpm poses an interesting question in this bug #743029 - when you have a moment it would be cool if you could take a look to provide an answer. regardless of the answer, it's a problem that needs fixing it appears [05:18] <_mup_> Bug #743029: Launchpad Translations only allows submitting 3 translation suggestions and discards the rest < https://launchpad.net/bugs/743029 > [05:19] wallyworld: if you're doing triage; there are a couple of links in the BugTriage wiki page that jointly should be driven to 0 [05:19] ok [05:21] I'm guessing there is no way to log into launchpad.dev without openid? [05:21] huwshimi: Basic auth might work. [05:21] admin@canonical.com:test [05:22] wgrant: how do I get a prompt for that? [05:22] huwshimi: you don't [05:22] huwshimi: Put it in the URL. [05:22] Although I guess you'll need to escape the @... [05:22] Let's see if it works. [05:22] huwshimi: note that we don't support basic auth in prod [05:23] lifeless: Yeah this is just to test over my network [05:23] lifeless: From a vm running ie that doesn't like the openid stuff [05:23] Hm, doesn't seem to work. [05:23] wallyworld: wow, that one is just insane. But frankly it sounds as if it's more of an "only the latest suggestions are displayed" issue. I'm not aware of anything like that having been done deliberately while we still had the Rosetta team, though I could well imagine someone deciding that if the same user enters successive suggestions for the same message in the same language, it's probably to improve on the older ones rather than to offer lots of alternati [05:23] huwshimi: How does it error? [05:24] StevenK: It won't load https://testopenid.dev/+openid [05:24] StevenK: I don't know how to get the header code [05:24] Can it even resolve it? [05:24] I think I ended up with SSL errors of some kind last time i tried... I don't remember details. [05:24] jtv: yeah weird. thanks for looking [05:25] StevenK: Possibly not. I assumed it was something to do with my network setup [05:25] wallyworld: first I'd make sure that we understand what steps dpm went through. [05:25] Can haz review? https://code.launchpad.net/~stevenk/launchpad/dsd-dont-request-child-diff/+merge/55863 [05:25] https://bugs.launchpad.net/launchpad/+bug/741092/comments/11 <- and they say python is slow [05:25] <_mup_> Bug #741092: Archive:+subscriptions times out with many subscribers < https://launchpad.net/bugs/741092 > [05:26] wallyworld: It's almost buildbot time. [05:26] wallyworld: that may sound strange given his description, but our UI doesn't actually support the steps he describes directly. He'd have to submit the page every time, then page back to get to the message he was entering suggestions for. [05:26] jtv: i'm going to try it on qastaging using the bug report as a guide [05:26] wallyworld: Your branch seems fine, so it should probably get into the next buildbot run. [05:26] wallyworld: same here [05:26] I'll try a different page then [05:27] wgrant: i've submitted it to ec2 just after it was approved [05:27] wgrant: Damn, doesn't like that either [05:27] wallyworld: Hmm, but we don't run windmill in ec2 yet, so will that actually test anything? [05:27] huwshimi: No :/ [05:27] huwshimi: Can you see if the testopenid.dev request is making it to the appserver? [05:27] wallyworld: by the way, be careful to do this in Translator mode. Do it in Reviewer mode once, and the translation you enter will become official and your old ones will be counted as reviewed. [05:27] wgrant: erm... [05:28] wgrant: the tests are plain javascript ones. i thought windmill ones were separate again? [05:28] huwshimi: 'make run' prints all the requests it receives. [05:28] wgrant: Ah [05:28] wallyworld: Ah, they're not still run in WindmillLayer? [05:28] test_yuitests lives in lib/lp/*/windmill.... [05:29] jtv: ok. on qastaging though i can mess up as much data as i want though? [05:29] class CodeYUIUnitTestCase(YUIUnitTestCase): [05:29] layer = CodeWindmillLayer [05:29] wallyworld: yes, but: [05:29] wgrant: i didn't appreciate that fact. i would have thought they were separate :-( [05:30] wallyworld: dpm gave a specific link, and on staging you can tell that link doesn't do quite what he expected any more. Be careful not to change the effect of the same link on qastaging—it can be a useful reference. [05:30] wallyworld: I guess they need similar infrastructure, to actually get them to run in the browser and such. [05:30] jtv: yes, true [05:30] wgrant: This is what I get after clicking on the login link: http://paste.ubuntu.com/588109/ [05:32] jtv: so to double check, the fact that it says "Your suggestions will be held for review ...." means I am in reviewer mode? [05:32] No [05:33] That means you're in translator mode. [05:33] ah sorry, my fingers got mixed up [05:34] i meant to say translator mode [05:35] wallyworld: I just tried it and yes, it only shows my latest 3 suggestions for that one particular message. [05:35] jtv: so the bug is legit [05:35] ? [05:36] or if it's be design we need to say that [05:36] by [05:36] and why [05:36] In the sense that we can reproduce it, yes it's legit. Another question is whether it's a bug. To be honest I don't see why, if a single person suggests >3 different translations for the exact same message into the exact same language, we should expect all of those to be useful. I think it makes sense to assume that the person is improving, rather than throwing a bunch of garbage the reviewer's way after making a good suggestion. [05:37] Think about the spam or shelf-stuffing potential alone. [05:37] huwshimi: Hm, do other browsers work? [05:37] huwshimi: I'm able to log in from another machine OK. [05:37] huwshimi: Although it's not Windows. [05:38] huwshimi: Do you happen to have Firefox on that Windows machine? [05:38] jtv: i agree, and it seems any change was likely done for that reason. so i can mark it as won't fix and explain why? [05:38] StevenK: Just installing it [05:38] wallyworld: to be blunt, "won't fix" may be too mild. [05:39] More of a feature-not-bug. [05:39] opinion [05:39] :-) [05:39] wgrant: Firefox doesn't work either [05:39] lifeless: I agree completely with that being an ideal resolution for this, except in this case the discussion hasn't played out to that point yet. [05:40] wallyworld: BTW the messages are not actually gone. [05:40] huwshimi: What error does the browser show? [05:40] wallyworld: The reviewer can still get to them by "zooming in" on the message. [05:40] wgrant: Same thing. Can't find the server [05:40] * wallyworld tries that [05:40] wallyworld: They just don't clutter up the regular translation page which must show multiple translations. [05:41] huwshimi: You added testopenid.dev and launchpad.dev and bugs.launchpad.dev and anything else of interest to C:\Windows\System32\drivers\etc\hosts, or whatever it is these days? [05:41] http://pastebin.ubuntu.com/588111/ is the config diff I use. [05:41] wallyworld: …which on the one hand makes perfect sense, though on the other the reviewer obviously has no sane reason to assume that the translator may still want them to see the suggestion they made 4 page submissions ago. [05:41] wgrant: Yeah it's all there. The other urls work so it must be working [05:42] I thought testopenid.dev was a different IP [05:42] jtv: all valid points. [05:42] wallyworld: bear in mind that, thanks to the back-paging, the translator must go to considerable extra trouble to enter the redundant messages. So it ends up being a bit of a "doctor it hurts when I punch myself in the eye." [05:42] StevenK: It's not really configured properly. [05:42] jtv: i didn'h have to backpage at all [05:42] perhaps there's an ajax version working for me [05:44] wallyworld: no, that's probably because you happen to be on a view that only has one page to display then. [05:44] Even so, it's not like the submit is instantaneous. So this is a pretty rare case. [05:45] Or should be. [05:45] yeah [05:47] If anyone's running into this, I think it's probably because of some tangentially-related misconception or misleading UI subtlety. [05:49] jtv: i'm commenting on the bug now and we'll see what reply we get. did you want me to subscribe you to it? [05:49] wallyworld: sure, thanks [05:49] jtv: thanks for your help with this [05:51] No worries. Bear in mind by the way that this is David acting as a conduit, which is part of the Ubuntu Translations Coordinator role he's been fulfilling wonderfully. It's great for all of us to establish contact with him about these things; he lost a Rosetta team as a support point so it'd be good to know he's gaining a Launchpad team in its place. [05:52] wgrant: Any other ideas :) [05:52] huwshimi: How does your Apache config look compared to the diff I provided? [05:52] jtv: agreed. the past few times i've come across input from him i've tried to act promptly on it. [05:52] I don't *think* I changed anything else to get it to work. [05:53] But I remember it didn't initially. [05:53] That's great. He represents a lot of people's needs! [05:54] wgrant: http://paste.ubuntu.com/588114/ [05:54] wgrant: Not a diff [05:56] huwshimi: Try http://pastebin.ubuntu.com/588115/ (you might need to fix the branch-rewrite path, but I don't think Apache will complain too much if it's wrong) === almaisan-away is now known as al-maisan [06:03] * wallyworld does the school run [06:04] wgrant: No change [06:04] huwshimi: Do you even get to the login prompt? [06:05] nope. I think it's just 404ing [06:05] but I don't know the actual status code cause the browsers are trying to be nice and hiding that info from me [06:05] Firefox at least should be able to tell you. [06:06] Otherwise maybe tail /var/log/apache2/other_vhosts_access.log? [06:07] wgrant: Yeah pretty sure it's a 404 [06:08] wgrant: "Firefox can't find the server at testopenid.dev" [06:08] tcpdump ? :-) [06:08] That doesn't sound like a 404. [06:09] Can you ping testopenid.dev OK? [06:10] Ah hm. [06:10] wgrant: It gives exactly the same error when I go to a non-existent url [06:10] Is the default-ssl site enabled in /etc/apache2/sites-enabled? [06:10] Mine isn't. [06:10] That may be what I did to fix it. [06:10] sudo a2dissite default-ssl [06:11] wgrant: Pinging fails [06:11] That could also be a problem. [06:11] Sounds like your C:\Windows\System32\Drivers\etc\hosts is broken... [06:11] wgrant: it's possible [06:12] but it's weird that it works for some urls [06:13] wgrant: This is what I have in my hosts: http://paste.ubuntu.com/588122/ [06:13] Yet 'ping testopenid.dev' fails to resolve? [06:13] wgrant: yep [06:14] wgrant: "Ping requests could not find host testopenid.dev. Please check the name and try again." [06:16] And it's XP, so it's not UAC virtualisation... [06:16] I'm out of ideas :( [06:16] wgrant: Ok thanks anyway :) [06:16] I guess I'll leave IE broken for now, I can't think of anything else to try either. [06:28] 7M renders yesterday [06:30] lifeless: Nice. [06:31] wgrant: If this helps all the base urls: bugs. translations. etc. work but anything else breaks: shipit. lists. [06:31] wgrant: after opstats etc [06:31] huwshimi: Interesting. [06:31] wgrant: I think I must have some kind of port set up incorrectly or something [06:31] huwshimi: That doesn't explain why ping wouldn't work. [06:32] wgrant: by breaks I mean I can't ping any of them [06:33] huwshimi: I wonder if Windows' hosts file implementation has a limited line length. [06:33] huwshimi: I would not put it past them. Try removing all except the hostnames you need. [06:33] wgrant: I thought about that but they're spread out through that list [06:34] Bah [06:34] wgrant: I actually wonder if it's the Allow from in the apache file [06:34] huwshimi: That doesn't affect name resolution. [06:34] wgrant: Ah right [06:35] testopenid, lists and shipit are all in the last half of the line. [06:36] wgrant: oh, maybe I was wrong before [06:37] wgrant: Oh I am going to stab things. Hard. [06:37] huwshimi: Oh? [06:37] wgrant: Moving it to the start of the line makes it work [06:37] Yeah. [06:38] When in doubt, suspect shoddy Microsoft code? :) [06:40] wgrant: That really is aweful. I just don't understand. [06:42] Yes :/ [06:43] huwshimi: So, it's all working now? [06:44] wgrant: It exploded in an oops at first, but now it's working fine. Thanks heaps for your help :) [06:44] wgrant: Now I probably owe you cake too [06:44] Haha. [06:44] I'm glad that it's working. [06:44] Maybe you can even make it work slightly in IE9 :) [06:45] wgrant: Maybe I should just stab myself in the face with my desk? [06:45] IE will encourage you to sharpen your desk. [06:46] So maybe. [06:46] Haha [06:47] (if the hosts brokenness didn't already) [06:47] huwshimi: hosts files in windows have a maximum line length [06:47] huwshimi, i would love it so much if you proceeded with your idea to remove the variable sizes from the bug tag cloud [06:47] wgrant: I think I just blunted my desk from that [06:48] lifeless: hahahahaha, thanks :D [06:48] ever since i saw the mockup, the current ui hurts me even more [06:48] poolie: Me too! [06:48] lifeless: Yes, we worked that out after an hour or so :P [06:48] Which mockup? [06:48] so i think you're morally obligated [06:48] wgrant: I was writing a mammoth mail; sorry [06:49] wgrant: https://bugs.launchpad.net/launchpad/+bug/709009 [06:49] <_mup_> Bug #709009: Tag clouds are very hard to read < https://launchpad.net/bugs/709009 > [06:49] oh so *now* mup decides to work, just after I've given up and started pasting bug links into the channel [06:50] Bug #1234 [06:50] <_mup_> Bug #1234: Gina is an unmaintainable mess of command line options, environment variables and shell scripts < https://launchpad.net/bugs/1234 > [06:50] Huh. [06:50] Indeed. [06:50] Bad _mup_ [06:50] wgrant: Specifically: https://launchpadlibrarian.net/62993362/tag_cloud.png [06:50] (nice mockup) [06:50] wgrant: But I think I would like to revise that slightly again when I get to it [06:51] Yeah. [06:51] A lot of other things need revision, too :/ [06:51] i wonder now, given robert's point about performance [06:51] if it would be better to just simply show all the official tags [06:55] poolie: My brain no longer functions. I can not reply [06:59] ! sorry [07:04] Project windmill build #129: STILL FAILING in 48 min: https://lpci.wedontsleep.org/job/windmill/129/ === al-maisan is now known as almaisan-away [08:03] wow, trying to get a sorted list of tagged bugs is awful [08:06] As awful as