[00:00] bdmurray: does bzr info for your download-cache indicate (in the Location) that it is "checkout of branch: bzr+ssh://bazaar.launchpad.net/~launchpad/lp-source-dependencies/trunk/" ? [00:01] gary_poster: yes and thanks for looking into this [00:04] ok, bdmurray, does the distribution appear to be in the download-cache when you do ``ls download-cache/dist/zc.buildout-1.5.0dev-gary-r10*``? You should see ``download-cache/dist/zc.buildout-1.5.0dev-gary-r108342.tar.gz`` as one of the entries [00:05] gary_poster: yes, that is there [00:05] ok, trying next step [00:08] bdmurray: argh, I started with a completely fresh eggs directory, and a ``make build``, and it worked just fine for me. ok, here's an interesting idea. what if you do a ``ls eggs/zc.buildout-1.5.0dev_gary_r10*``? Does that include eggs/zc.buildout-1.5.0dev_gary_r108342-py2.5.egg ? [00:10] gary_poster: no, I only have 105072 there [00:11] darn, there goes that theory [00:15] bdmurray: ok, flailing even more obviously now. Does ``ls -hl download-cache/dist/zc.buildout-1.5.0dev-gary-r108342.tar.gz`` show 274K? [00:16] gary_poster: yes it does [00:17] argh [00:18] bdmurray, how did you build this tree? I can try to dupe. [00:19] gary_poster: I believe I just ran rocketfuel-setup [00:20] bdmurray, what happens if you run rocketfuel-get, out of random curiosity? [00:20] Same thing? [00:23] gary_poster: a make build after rocketfuel-get fails the same way [00:23] bdmurray: rocketfuel-get does a make also though... [00:24] bdmurray: try a ``make clean`` and then a ``make build`` [00:25] gary_poster: still the same [00:25] * gary_poster curses [00:27] bdmurray, the only other thing that came to mind--this is the last ditch effort before I try to do rocketfuel-setup tomorrow myself--is to make sure that the user that you are running as can read those files. I noticed that the home directory in your pastebin was ubuntu. Is everything owned by ubuntu? [00:28] (that is, the ubuntu user) [00:28] (and you are running as the "ubuntu" user) [00:28] gary_poster: yes, it is a vm and that's the only user === joey is now known as Guest52456 === Guest52456 is now known as joey [00:32] bdmurray: ok. what kind of vm is it, and how gargantuan is it? If I could rustle up some place for you to upload it to (which may not be possible), would that be reasonable to upload for your bandwidth? (It wouldn't for me, for instance; my download is good but not my upload). [00:32] Failing this, I'll try to dupe the problem myself in a vm. I'm the release manager for this release and so I'm pretty busy, and really busy tomorrow, so I'm might not get to it till Wednesday or so. [00:32] bdmurray: everything was fine, and then it stopped working recently, right? [00:32] gary_poster: no, this is actually a brand new setup [00:33] bdmurray: oh! ok. Did you set up the vm yourself, or is this something we provided somehow? (There's been talk of doing this) [00:34] bdmurray: and karmic, right? [00:34] gary_poster: I set it up myself. Karmic - yes. [00:35] bdmurray: ok cool. I'll try to get one set up and started tonight if I can, and get back to you if I figure anything out. Feel free to ping me tomorrow, but forgive me if I don't have it ready. [00:35] gary_poster: okay, thanks! its not a huge hurry [00:36] bdmurray ok cool. I know you have done some nice contribs lately, so don't want something silly like this to be in your (or anyone elses) way [00:37] gary_poster: thanks [00:45] gary_poster: to be clear eggs/zc.buildout...108342 does not exist only 1050702 does [01:21] thumper, ping [01:23] rockstar: pong [01:23] thumper, skype? [01:23] ok [01:32] thumper, lp:~rockstar/launchpad/upgrade-branch-rc [03:03] rockstar: you realise that you deleted 4 qa items and only added 2? [03:03] thumper, yes, because the 4 were actually only two. [03:03] rockstar: actually 5 and 2 [03:03] ok [03:03] They were the same issue. [03:03] rockstar: is the bad rcfixed yet? [03:03] thumper, yes. [03:04] It's in db-devel now, should be in stable next year some time. [03:04] ok [03:34] hooray (?) for 3g [03:35] thumper: did we have much more to talk about [03:35] ? [03:35] mwhudson: hmm.. I think we covered most of it [03:35] mwhudson: I'll file some bugs, update the wiki [03:35] mwhudson: and we can take it from there [03:35] thumper: cool [03:36] thumper: i'll start working on the "no mirrorRequest for noop pull" case [03:41] ok [03:53] rtt min/avg/max/mdev = 1171.359/9666.795/18292.795/6516.203 ms, pipe 17 [03:53] wow, this is slow :( [03:54] although i wonder if the thunder is affecting my connection.... [03:56] mwhudson: Will your new place have an Internet connection immediately? [03:56] wgrant: yes, i guess that's the upside to this screw up [04:29] * thumper goes to make dinner [07:19] Ah good. My rc candidate has passed all its tests. [07:19] Er, I guess it's been elected, so it's not a candidate anymore. [07:34] Good morning! [07:35] Any chance of having the Assignee column removed on the answers main list? No one uses it (in Ubuntu at least) and it's a waste of space. [08:24] hi henninge [08:24] hey jtv ! [08:24] henninge: did you see the oopses in pofile-js-footer? [08:24] noe [08:24] autofocus_html_id breaks in some vague way :( [08:24] p [08:25] Got some edge oopses, e.g. OOPS-1486EC432 [08:37] henninge: ahhh, I think this happens when the current batch is empty. [08:38] browser/pofile.py, _initializeTranslationMessageViews [08:38] jtv: sound reasonable [08:54] henninge: just lost a bunch of work on the bug report because I pressed but not while trying to switch workspaces... shorter version is now at bug 512698; I'm reconstructing the rest of the details in a comment. [08:54] Bug #512698: Oops while rendering +translate page for empty POFile [08:54] hi danilos [08:57] hi jtv [09:08] wgrant: hi [09:13] jtv: this is related to bug 359180 [09:13] Bug #359180: Missing keyboard shortcut for navigation [09:14] adiroiban: hi! what is related to that? [09:14] bug 512698 [09:14] Bug #512698: Oops while rendering +translate page for empty POFile [09:15] Morning [09:18] hi mrevell! [09:18] hey jtv [09:18] adiroiban: oh, it's related to the QA failure? [09:18] adiroiban: want me to put up a fix? [09:19] (do have a standup now) [09:20] jtv: I have created a new branch for bug 359180, it was approved, but not landed [09:20] Bug #359180: Missing keyboard shortcut for navigation [09:20] adiroiban: then we'd still want a quick fix for this oops... hang on, call first. [09:46] jtv: this can be a quick fix http://paste.ubuntu.com/363127/ [09:47] bigjools: Hi. [09:49] adiroiban: the "autofocus_html_id view/autofocus_html_id | string:;" part? [09:49] wgrant: I spoke to Joneau and we're happy to remove SPRBU on the proviso that any API is designed to not preclude its addition later [09:49] Jøñeau? [09:50] bigjools: Ah, excellent. [09:50] * bigjools can't find the compse key [09:50] compose, either [09:50] Any API will only be used in a couple of places, so it shouldn't be too terible. [09:51] jtv: yes [09:51] adiroiban: that catches an unhandled exception!? [09:52] jtv: nope. it should deal with LocationNotFound [09:54] adiroiban: why does it produce that exception type anyway? With these deep-zope tracebacks I usually end up just getting one or two keywords and ignoring the rest. [09:56] What comes out of the code on our end should be an AttributeError, right? [09:57] jtv, adiroiban: hi, can you guys please comment on https://bugs.edge.launchpad.net/rosetta/+bug/506714 [09:57] danilos: will do [09:57] jtv, thanks [09:58] jtv: I don't know why the view is not visible from the template [09:59] adiroiban: I thought it was that the template can access the view, but an attribute of the view then fails with an AttributeError. [10:00] Ah! Maybe things get confused because it's a @property that fails with that error. Maybe somewhere along the line it's assumed that if you evaluate just foo.bar, a property, and get an AttributeError then it must be because foo.bar does not exist. [10:00] But in this case, foo.bar is really a method invocation that happens to fail with an AttributeError. [10:02] ok back [10:14] jtv: I have commented on the bug and added a possible fix [10:15] adiroiban: thanks... meanwhile I'm having a look at the one danilo pointed out [10:17] jtv: should I push the branch for review as a RC ? [10:18] don't know what is the process [10:18] adiroiban: the first thing is to get a review... I can review it, if you like. There doesn't seem to be anyone on call. [10:20] jtv: should I get a review for this branch https://code.edge.launchpad.net/~adiroiban/launchpad/bug-359180-take-2/+merge/17785 ? [10:20] or I should create a new one? [10:22] adiroiban: separate branch please... this is a separate, if related, problem. [10:22] and if we do want to get a fix for the actual oopses rolled out outside of the normal release schedule, we want a minimal change. [10:22] but the fill fill depend on this branch [10:22] and it is not commited [10:23] but the fix will depend on this branch [10:23] adiroiban: what other problems are there with the branch that was landed, besides these oopses? [10:24] a javascript keybinding overlap with Compiz [10:30] jtv: I will create a new branch to only fix this issue [10:30] adiroiban: how much trouble can that one cause? [10:30] creating a new branch ? [10:30] I don't understand the question. [10:30] adiroiban: what problems does the keybinding overlap cause? [10:31] jtv: minor... if you have compiz, the keybindings will be used by compiz and not send to the Webbrowser [10:35] adiroiban: so not a big problem, but we're basically telling people to press keys they shouldn't be pressing... How about we request that your original fix be backed out? [10:35] You've got a better fix waiting for next release, right? [10:35] yep [10:36] So then I suggest we fix all these problems by backing out the broken fix. [10:36] fine by me [10:36] OK, I'll request it. Thanks! [11:02] Morning, everyone. [11:04] deryck, good morning. [11:19] morning deryck [11:22] * jml is off to run some errands. [11:40] hey guys, sf.net recently blocked access for some countries (http://sourceforge.net/blog/). is this something that will happen on launchpad, too? [11:45] traceroute suggests that the Launchpad datacentre is in London, so US law seems unlikely to intrude [12:07] maxb: it is [12:08] mdd: I'm not aware of any proposal to do that, however we do have a US arm so its not impossible that we might need to do address it [12:09] lifeless: okey, thanks! === henninge is now known as heninnge-lunch === mrevell is now known as mrevell-lunch === heninnge-lunch is now known as heninnge === mrevell-lunch is now known as mrevell === jamalta_ is now known as jamalta [14:29] james_w, are you around to talk about your three outstanding lazr.restfulclient branches? [14:32] leonardr: I believe james_w is on vacation this week. [14:33] al-maisan: well, that takes some of the time pressure off of me [14:34] Is there anyone else who would know about UDD branches? [14:34] leonardr: aha :) IIRC there's a platform sprint on next week, so probably even less pressure on you :) [14:34] I have been wondering why the branches for debian/+source/subversion have been deleted from LP [14:35] maxb: it depends, jml or thumper or mwhudson, lifeless as well probably. [14:38] leonardr: i just pasted the stack trace on #512552 [14:38] Bug #512552: Large POST fails with createComment on merge proposal [14:39] mup, ok [14:39] er, dobey, ok [14:40] heh [14:41] dobey, responded [14:42] ok, i'll try that === salgado is now known as salgado-afk [14:53] the build is broken [14:53] maxb, hi [14:53] hello [14:54] maxb, I don't know why the branches for debian/+source/subversion have been deleted from LP. [14:54] Do you think I should wait for james_w to be back to find out more? [14:54] maxb, sadly, yes. [14:54] hmm. [14:54] actually. [14:55] maxb, I think he sent an email to the UDD list about how he's got a web page that logs the stuff that the importer does [14:55] Indeed. But I don't think the importer can automatically delete things [14:56] maxb, hmm. [14:56] maxb, are you sure? [14:57] The API to do that was only added very very recently. [14:57] So it seems unlikely. [14:58] (10 days ago, in fact) === matsubara is now known as matsubara-lunch [15:30] mars: ping [15:30] hi EdwinGrubbs [15:31] mars: did you get a chance to look at my email on the rhinos list concerning sprites? [15:32] EdwinGrubbs, yes. I'm not sure how to answer in that case though. I defer to Martin and Matthew. [15:32] mars: did you learn how sidnei is consolidating the sprite images with a script? [15:33] EdwinGrubbs, yes. I'm not sure if the tool they use is configurable to take whitespace into account. [15:33] EdwinGrubbs, I'll check [15:34] mars: I assume we have the source code for it. Is that not correct? [15:37] EdwinGrubbs, nope. They use spriteme.org right now. If they switch to SmartSprites (http://csssprites.org), then it is configurable, sort of. === beuno is now known as beuno-lunch [15:38] EdwinGrubbs, looking at the SmartSprites docs, doing so would still be a hack [15:39] EdwinGrubbs, is there no way to restructure the HTML to do this? Because all of the sprite tools seem to assume there is. [15:45] mars: there are several ways to restructure the HTML, but they all assume that the element using a sprite as the background is only one line. In the most extreme solution, we put a sprite in an element by itself, and it becomes equivalent to an tag. [15:47] mars: As described in the email, the most difficult layout to handle is where the first line is indented after the sprite, but the second line is not indented, so it appears under the sprite. If we didn't care about that situation at all, we would still have a fairly large amount of work converting all the long tags with sprites to be so that wrapping wouldn't expose the next sprite. === salgado-afk is now known as salgado [16:06] EdwinGrubbs, found something that may be of interest here: http://archivist.incutio.com/viewlist/css-discuss/104801 [16:07] EdwinGrubbs, difficult to get enough context though: http://archivist.incutio.com/viewlist/css-discuss/104799 [16:09] EdwinGrubbs, it seems like you have two problems: 1) You need inline elements to obey a hanging indent; And 2) You need the sprites to not have the bottoms blown out of them [16:18] mars: I don't see how that mailing list thread applies. It uses vertical-align:botton to make sure that each row of
  • elements line up at the bottom of the element where the image is as opposed to the top of the
  • . [16:18] mars: your summary of the problems is accurate. [16:33] mars: look at the answer to question #2 at http://csssprites.org/#faq [16:35] EdwinGrubbs, yes, so every sprite definition using the tool would need the margin directive [16:36] mars: my point is that we are not the only people encountering this problem. [16:36] EdwinGrubbs, yep. Looking at the Yahoo and AOL sprites, they space them with a line height of 2-3x [16:36] EdwinGrubbs, anticipating that if menu items exceed that, then you have a problem === matsubara-lunch is now known as matsubara [16:54] EdwinGrubbs, so one way to fix this would be to regenerate the sprites with each sprite starting at 3x line height [16:54] is staging deliberately down? [16:54] gary_poster, ^ ? [16:54] mars: exactly [16:55] EdwinGrubbs, 2x line height between each [16:56] EdwinGrubbs, I'm tempted to say that going beyond that height for items means that you can stop using sprites. For example, switch to list item bullet images. [16:57] mars: why would you stop using sprites? [16:58] EdwinGrubbs, because they aren't meant to be used for
  • elements that span 5+ lines. They are meant to clean up a whole bunch of short little items, not big things. [16:59] mars: but is there any real drawback to using them for 5 line elements? [16:59] Sprites are not the One True Way for every image on the page. They are a way to save you from making 40+ server requests. [16:59] EdwinGrubbs, these padding issues would be one drawback. [17:00] mars: how is it a drawback for 5 lines if the solution is the same as for 2 lines? [17:01] EdwinGrubbs, well, imagine a page with a menu of options, each option has 8 lines. You could fit maybe 5-10 bullets on the page? [17:02] EdwinGrubbs, so sprites would save you 10 unique requests, say. Compare that to a typical Launchpad page, where you have say 35 unique images. [17:03] 35:1 is a big performance gain [17:03] 10:1 or 5:1, not so much [17:03] now, it is a pain to engineer the sprite for the 5:1 case [17:03] mars: but those bullets will often use the same sprites that already appear on the page. For example, the picker displays person and team sprites, and it can easily be 3-5 lines in an
  • [17:04] EdwinGrubbs, true, but is it worth the time to re-engineer the sprite image so you can make it fit this one case? So you can save two requests to the server? (one for person, one for team) [17:05] that's assuming that those two stand-alone images aren't cached, either [17:05] EdwinGrubbs, if we didn't have sprites, we would just use the person and team image bullets in the
  • , no problem [17:06] Using sprites is a performance optimization for initial page loads [17:07] So the cost of not using that optimization in the picker list (2 requests) is trivial compared to the cost of not using it for 90% of the other page icons (35+ requests) [17:08] EdwinGrubbs, just thought of another approach: did you consider switching the picker list over to a
    ? [17:09] You could use the sprite on the
    , indent the
    using the CSS text-indent property [17:11] You still get to use the sprites, but don't have to worry about a 5-line list item any more [17:11] EdwinGrubbs, I have to go to lunch, but let me know if the
    idea would work. [17:11] mars: that is not really any different than an
  • with two spans in it. You would still have the issue of a long
    wrapping. [17:12] EdwinGrubbs, a long
    would not make much sense. 2 lines at most. [17:12] The
    can be as long as needed. [17:16] mars: my point is that the differences between a
    and a
      is not where the problems lie. We could do everything we want with spans inside an
    • . Using a
      might be simpler, but it really is irrelevant to the sprite problem, which is difficult to re-use. [17:27] adeuring: ayt? [17:28] kfogel: yes? [17:28] EdwinGrubbs: I agree. A
      and a
    • are block-like and should not be used [17:30] adeuring: I'm writing tests for other entities, for bug #506018. I thought I'd start with SourcePackage, which means my next step is to create a source package in my testdata. Oh, or find one, duh. Let me look around. Anyway, maybe I don't actually have a question yet. I just wanted to let you know what I'm doing, in case it sounds insane. [17:30] Bug #506018: Need a "+patches" view: report lists patches attached to bugs. [17:31] kfogel: OK ;) I'm AFK for the 10-15 minutes but will stay around for 2 or 3 hours [17:31] adeuring: but, do we have any packages in default test data? I'm not even sure how to search for them. [17:31] just got an OOPS with this: [17:31] https://launchpad.dev/+search?field.text=package [17:31] kfogel: let me look... [17:34] adeuring: oh, wait, I think we found them before [17:34] https://bugs.launchpad.dev/ubuntu/+source/libstdc++ [17:34] https://bugs.launchpad.dev/ubuntu/+source/cdrkit [17:34] https://bugs.launchpad.dev/ubuntu/+source/alsa-utils [17:34] kfogel: yeah, right [17:35] etc [17:35] and we opened bugtasks for some of those packages (on "Bug C", which has a patch attachment) [17:35] ok [17:36] so, https://bugs.launchpad.dev/ubuntu/+source/cdrkit/+patches should work! [17:36] adeuring: and it does work, heh [17:36] kfogel: great! [17:37] adeuring: if you have to go afk for 10-15, no worries. I'm going to write a test for this in lib/lp/bugs/stories/patches-view/patches-view.txt. [18:00] jml (and mars), staging was down for an update about when you asked, yes [18:04] nytol [18:05] EdwinGrubbs, fair enough. My point was that image reuse is not the reason that you use sprites. [18:06] mars: it just seems strange to avoid an opportunity to use DRY if there isn't a drawback that we aren't already incuring with the 2 line solution. === beuno-lunch is now known as beuno [18:18] adeuring: when looking at patches view on a distro series (e.g., https://launchpad.dev/ubuntu/hoary/+patches), we *should* have the package column, right? (Currently we don't.) [18:18] (and I think I know how to fix it, I just want to make sure it is something that needs fixing) [18:19] kfogel: yes, I think that would be useful [18:20] adeuring: *nod* [18:20] kfogel: BTW, we should similary show the project name on the +patches view for project groups [18:21] leonardr: re: #512552 the OOPS I just got in HTTPError.content id is: OOPS-1487EA832 [18:21] Bug #512552: Large POST fails with createComment on merge proposal [18:23] dobey: so it looks like the size of the post caused the database to timeout? [18:24] leonardr: it would appear so, yeah [18:24] adeuring: oh, good idea! [18:31] adeuring: hmm, I think we already should show individual targets in project groups; I just added the "IDistroSeries" in this snippet, but the "IProject" was already there: http://paste.ubuntu.com/363372/ ...or are you saying that we should show product instead of target package, or that we should show both? [18:32] kfogel: sorry, i should have looked what we already have for project gruops. I menat that we should display the projects in this case. [18:33] adeuring: instead of packages? [18:34] adeuring: (I'm a little confused by the "target" terminology. Target seems very generic to me -- could refer to whatever the bug(task) is attached to. But we seem to always use it to refer to a package, in practice. Am I missing something?) [18:34] kfogel: I think we don't have packages for projects. [18:34] adeuring: where "projects" means "project groups"? [18:36] kfogel: yes, bugtargets can be a bit confusing ;) And I meant the we should should on the patches view for project groups (IProjects) the individual projects (IProducts) in the traget column [18:36] s/should should/should show/ [18:38] adeuring: gotcha. note that we don't label any column "target"; we label it by what entity it is. So right now it's "package" for most views (or it's absent), but for the IProject view, it will say... well, not "Product", because users don't know that word, but "Project" (meaning IProduct). [18:38] kfogel: exactly [18:54] adeuring: so currently, in lib/lp/bugs/stories/patches-view/patches-view.txt, I want to associate a bugtask (for one of the bugs we made that has a patch attachment) with a series like https://launchpad.dev/ubuntu/hoary. I'm using lib/lp/bugs/stories/bug-release-management/40-nominate-bug-for-productseries.txt as a guide. I'm being very, very verbose about what I'm doing, because I badly wish I were sitting next to you :-). [18:56] kfogel: looks good. And I think we can set up virtual desks if necessary ;) [18:56] (hunh... I guess one nominates a bug for a series directly; it doesn't require a separate bugtask. The question of where bugtasks do and don't apply can be a bit confusing). [18:56] ...virtual common desks... [18:56] adeuring: that virtual desks thing might be a good idea. so far I'm not blocked, but just in case: how does one do it? [18:57] kfogel: no idea in practice. But skype calls for example are sometimes auite useful [18:57] Or using gobby [18:58] kfogel: actually, I think you don't needs to go through the nomination cycle. Don't we have a factory methods for bugtasks that allows us to dircetly create a task for a product series? === EdwinGrubbs is now known as Edwin-afk [18:59] adeuring: let me see [18:59] adeuring: we wrote our own make_bugtask() (meta-factory method for make_thing(), if you recall). make_bugtask() calls factory.makeBugTask()... let me take a look at that. [19:01] kfogel: I think there is an odd restriction that you can't directly create a bug for a product series. But if you have an existing bug, you can add a task for a product series. [19:02] kfogel: and we used the helper function make_bug in order to set the importance ad the status [19:03] adeuring: indeed. (although I have it working the other way already, so unless there's a preference, I could just stick with this) [19:03] kfogel: well, if it works, things are fine. [19:39] good morning [19:39] leonardr: did that OOPS provide any useful info for you? [19:39] dobey: only to the extent that it confirms my belief that it's not my fault [19:40] i'll add a comment [19:40] ok [19:48] is the bugs subdomain broken on edge right now? [20:07] morning [20:08] jamalta: broken how? [20:08] bac: is https://edge.launchpad.net/sachco really proprietary? [20:09] dobey: i get an oops every time i try to load a bugs page [20:10] not the case anymore, so i guess it fixed itself [20:10] jamalta: individual bugs work fine. i think it's timing out for any lists of bugs pages, afaict [20:10] dobey: oh right, it is [20:11] jamalta: You can disable edge when search bugs, or write very targeted bug searches [20:11] sinzui: i disabled edge for now [20:11] i just wanted to see if it was an issue that affected everyone or just me [20:11] sinzui: nah, i just needed a test project === matsubara is now known as matsubara-afk === salgado is now known as salgado-afk === Ursinha is now known as Ursinha-bbk === Ursinha-bbk is now known as Ursinha-bbl [20:52] biab [21:10] leonardr: so the web page is timing out talking to the db as well, with similar content size in the comment. [21:30] dobey: well, that lets me off the hook [21:31] either launchpad needs to be made faster, or the timeout time needs to be made proportional to the incoming payload [21:31] * dobey votes both [21:36] that oops from the api has 46s of non db time [21:36] * thumper wonders what it is donig [22:03] biab, again === Ursinha-bbl is now known as Ursinha [23:33] gary_poster, yeah, thanks. it interrupted my UDW preparataions, so it was a bit surprising [23:33] * jml gone