[00:00] kiko, approved [00:00] although it makes me wonder why *you* can't approve it :) [00:01] beuno, sinzui: https://edge.launchpad.net/notify-osd persistently oopses with the same error I saw yesterday -- sorry I was out, but is this still not fixed? [00:02] kiko, PQM was closed, so the branch didn't go in [00:02] not sure if sinzui managed to get it in now [00:04] devel r9131 - it landed like 20 hours ago. [00:04] there you go [00:04] the benefits of being open source *and* having wgrant around [00:05] heh [00:05] okay, I gotta split [00:05] catch you all later === kiko is now known as kiko-afk [00:06] Oh bugger, looks like I need to port a bugs page... === Ursinha-nom is now known as Ursinha [00:34] beuno: you around for a chat? [00:37] * thumper is a very frustrated monkey [01:00] * thumper does lunch === matsubara is now known as matsubara-dinner [01:48] wgrant: that's not true, all xmlrpc commands on the master side are asynchronous [01:48] wgrant: the slave side isn't. === matsubara-dinner is now known as matsubara [01:55] cprov: But I've seen it cease to make XML-RPC requests to the other slave while the first one is downloading the chroot. [01:55] (I verified this morning) [01:59] wgrant: are you using daemon/buildd-manager.tac ? [01:59] cprov: That's the one. [02:01] It certainly looks like it should be async. [02:01] wgrant: right, check BuilddManager.scan(), the slave used for dispatching is non-blocking [02:05] cprov: I'll try again this evening when I have my other buildd back, but I'm pretty sure it ceased to scan while the download was occurring. [02:05] Which makes no sense, given the code. [02:06] wgrant: maybe you can add more debug to the manager in order to prove your hypothesis [02:06] cprov: How do I inform twistd that I want to increase the logging level? [02:07] There's lots of useful debug logging points, but they're not shown. [02:07] wgrant: you don't twistd doesn't accept args, needs code/config change [02:07] cprov: Ah, OK. [02:08] wgrant: BM._setupLogger() [02:08] cprov: Yup, just found that. [02:10] wgrant: you probably need more debugs, specially for ensurePresent() at the xmlrpc proxy. The current logging timestamp precision should be enough for spotting blockers. [02:18] cprov: It is definitely synchronous somehow. I added a time.sleep(30) in the slave's ensurepresent, started a build on a real builder, and marked a non-existent builder as OK. As expected, I can see it calling ensurepresent for each file for the build. 32 or so seconds later, it logs the response. [02:18] There is no scan in between. [02:18] And my builder that is uncontactable is still marked OK, when it should have been deactivated 90 seconds ago. [02:19] Once the build call went through after the 30 second wait for each file, the scanning started again and killed the bogus builder. [02:19] So it is synchronous, somehow. [02:20] wgrant: the scan cycle is synchronous with and 5 s delay, it won't finish the cycle if the dispatching started haven't finished. [02:22] cprov: Ohh, I see. That's a bit bad. It is still fairly synchronous! [02:23] wgrant: not really, there is an async burst of dispatching ... and that combined with 60 s timeout on the xmlrpc proxy guarantees that cycles won't take much longer than that. [02:24] wgrant: of course we could be more async, I don't see any harm on that, but it requires more refactoring. [02:25] cprov: This explains why builds often seem to take ages to finish. I'd been wondering about that. [02:26] wgrant: how so ? *ages* to be scanned again ? [02:26] cprov: Right. [02:27] Often I'll see a logtail that is right near the end of the build. [02:27] But the build won't complete for a couple of minutes afterwards. [02:27] wgrant: this perception comes mostly from the collecting task [02:28] wgrant: specially when we have to collect and process dozens of jobs, calling process-upload several times. [02:28] Yep. [02:28] And starting that isn't exactly quick... [02:29] wgrant: the dispatching cycle itself will always take about 80s (10s resuming, up to 60s dispatching, and some cruft) for 1 or 100 builders [02:29] wgrant: no, initZopeless() takes about 8s (IIRC) [02:31] wgrant: one change that would immediately remove that would be replacing the process-upload calls by directly instantiation of UploadProcessor() [02:31] cprov: Why wasn't that done in the first place? [02:31] wgrant: the only gotcha with it is that we need to switch db_user [02:32] cprov: Isn't it more appropriate to grant the builddmanager DB user the required privs? [02:32] wgrant: because the gain on async dispatching was higher [02:32] Ah. [02:32] wgrant: ehe, not from the isolation PoV [02:33] wgrant: uploading and publishing binaries doesn't necessarily fit in build-dispatching domain. [02:34] cprov: True. [02:34] wgrant: but if you ask my personal opinion, I'd say that we have too-many-db-users atm ;) === cprov is now known as cprov-zzz [02:45] good night, guys! === matsubara is now known as matsubara-afk [03:43] who changed make TAGS to make the tags file 53Meg? at 15 it was bad enough [03:44] or is it just the result of buildout chagnes? [04:19] thumper: i suspect it should exclude the paths that contain the dependencies [06:06] arse [06:06] for the second time in as many days my laptop's keyboard stops responding [06:18] Hmm, some bugs say slow-keys [06:18] * thumper must check that next time [06:54] jml, did you get word of the Mysterious Unfinished Branch Jobs? [08:19] Hmm. Did the session DB get nuked when I wasn't looking? [08:26] good morning [08:28] gmb: Yup [08:28] stub: Okay. As long as I've not gone mad. [08:28] Okay, madder. [08:28] Thats a seperate bug [08:32] Argh! I've just double-filed two bugs. Damn chromium. [08:32] Hitting enter for form auto-completion apparently submits the form, too. [08:50] hi henninge [08:51] hi stub, gmb & all the rest :) [08:51] hi jtv ! [08:52] jtv: Goedemorgen. [08:52] gmb: Geweldig, je spreekt Nederlands! [08:53] jtv: Niet erg veel. [08:53] gmb: I seem to be making progress with your language as well: I think in the past week at least 4 Brits had me pinpointed as English with absolute confidence. [08:53] gmb: my girlfriend's learning Dutch... want to set up a weekly practice session on skype? :) [08:54] jtv: That generally means that you're making all the right mistakes. As BjornT once told me: English doesn't have many rules, but it has a lot of exceptions. [08:55] jtv: It's tempting. I think I'll try to convince someone to let us sprint in Amsterdam first. It's nice to have an aim in mind :) [08:55] gmb: English is the lest prescriptive European language I can think of... matches the legislative culture as well. You just do whatever everyone else does. :) [08:55] :) [08:55] Oooh, sprint in Amsterdam... high time for another one of those [08:56] gmb: the drawback is that more than any other European language, English makes me feel I don't know the meanings of many words. I merely _associate_ them with things. [08:56] jtv: Oh, that isn't how language is supposed to work? Damn, I've been doing it wrong all this time... [08:57] But yes. that's a problem even for native English speakers. [08:57] * jtv finally looks up "acrid" [08:58] Apparently it means "The program 'dict' is currently not installed." Some examples follow. [08:59] jtv: The trouble is that once you learn the words there's a good chance you'll hold on to their meanings when no-one else does and become something of a sesquipedalian. [08:59] Sharp and harsh, or bitter and not harsh; pungent (but presumably in some particular way). Basically a copout for "we don't know either, we only ever use it to describe gun smoke" [08:59] dict sesquipedalian [08:59] a dwarf? [09:00] jtv: Also, the smell of powersupply smoke. [09:00] lol [09:00] jtv: Sadly, not a dwarf. Though the idea is interesting. Sesquipedalian == "Person who uses long words." Synonyms: Smartarse. [09:01] ah, from: one apt to the usage of words like sesquipedalian [09:02] Dict said "measuring or containing a foot and a half; as, a sequipedalian pygmy." [09:02] Awesome! [09:02] And, this being English, probably completely accurate. [09:03] Yes, according to wiktionary anyway, "a foot and a half" is it's literal meaning. [09:04] In Dutch, or German, or French etc, when you look up a word, you expect a definition. Not a series of loose associations, a list of references in last year's TV programs, and an artist's impression. (Yes Wikipedia, I'm looking at you). [09:05] Now Thai, Thai is the master of loosely-associated words. Try "translating" some on thai2english.com. It won't even find all the words' beginnings and endings, let alone conclusive meanings. [09:05] lol. [09:06] That used to be a game on some radio station or other: Translate a song title into Thai, translate it back again, see what comes out. [09:16] gmb: an observation. Like English, Thai doesn't normally distinguish between orange the fruit and orange the colour. The word for juice also means water, or pretty much anything that's fluid or gel-like (basically anything you could take on a plane before 2002 but not after). And "100%" basically means "to a significant extent." [09:16] Thus, the Thai for "100% orange juice," strictly read, says no more than "a fluid that is to a significant extent orange in colour." [09:16] Excellent. [09:21] Hello! [09:55] gmb: Does PQM still hate me? [09:59] wgrant: It did earlier this morning. You're number 3 in the queue at the moment so we should know within an hour whether you're cursed or not. [09:59] gmb: It has so far expressed its intense dislike for me twice on this branch? [10:00] wgrant: Well, once was because we were in testfix, the other *looked* like a testfix issue but there's no email for a relevant breakage so I don't know. [10:00] wgrant: You have to remember, though, that PQM hates *me*. You, OTOH, might well have done nothing wrong :) [10:01] (Ask sinzui; it hates him too) [10:01] Heh. [10:01] Third time lucky.. [10:02] Here's hopping [10:02] Er. [10:02] Hoping. [10:10] bigjools: Um, so it's not a bug that the pillar shown on the DSP code index is the distro? [10:11] Oh, but the app tabs link to the source package. [10:11] That's confusing,. [10:12] on phone, one sec [10:12] launchpad just passed 200000 branches \o/ [10:13] thumper: Ouch. [10:13] thumper: Can you say how much disk space? [10:13] wgrant: we'll hit 300k before all the package branches are up [10:13] wgrant: lots [10:19] Ah, PQM is alive after all. [10:19] wgrant: yes it's confusing [10:20] mrevell: shall we mark https://launchpad.net/launchpad/2.1 as either 'stable', 'supported' or 'obsolete'? (I hate how it's showing up among active series on the graph on launchpad.net/launchpad) [10:20] we talking about having a packaging tab, but that would screw over bugs [10:20] bigjools: A packaging tab only makes sense for products/projects/people. [10:20] exactly [10:21] so if it's a bug on a package ... [10:21] And has little to do with this strange dual-context thingy. [10:21] well it has a lot to do with it [10:21] danilos: I'm not sure what applies in our case. Obsolete I guess because you can't really run 2.1. [10:21] mrevell: yeah, I'd likely agree [10:23] danilos: done [10:23] mrevell: great, thanks (fwiw, that's what we have on old series like https://launchpad.net/rosetta/1.0) [10:24] danilos: Cool, I'll keep on top of that then, at least until our PM starts :) [10:24] mrevell: yes, prime minister :) [10:24] heh [10:25] PM? New project manager? [10:26] yes, there a position open [10:28] Ah. [10:29] can you say "singing from the same hymn sheet?" and "blue-sky thinking" ? [10:31] Huh? [10:33] PM BS bingo [10:34] Ah, right. [10:34] * wgrant isn't quite sure who is what. [10:35] The closest I can get is 'drinking from the same toilet'. [10:35] laying the same cable [10:36] :) [10:36] Heh. [10:37] wgrant: So, I suck at reading regexes. Resubmitting again. Should work this time. [10:41] gmb: Thanks. [10:51] good morning, all. [10:52] Morning deryck. [10:53] hi deryck [11:42] jml: did you get word yesterday about the mysterious lingering branch jobs? === beuno_ is now known as beuno === al-maisan_ is now known as al-maisan === salgado-afk is now known as salgado [13:04] gmb: I presume it didn't? [13:04] Or is PQM even slower than I thought? [13:06] wgrant: pqm is processing a lazr.restful branch, which means that it runs the full test suite [13:07] BjornT: Ah. === kiko-afk is now known as kiko === matsubara-afk is now known as matsubara [13:41] bac: ping [13:41] good morning barry [13:42] bac: good morning! i have a question for you regarding private teams [13:42] ok [13:42] if i'm an admin for a private team, but not an owner, it appears as if i cannot make other members admins [13:42] bac: i'm not sure of the exact combination of PMT/private, owner/admin, etc [13:43] bac: is this a bug or a feature? [13:43] barry: that is true of all teams. even on public teams only the owner can beknight others [13:43] barry: i think it is a (questionable) design decision [13:43] bac: i guess when i've done this before i've been the owner too [13:43] bac: agreed [13:44] yeah, i was surprised when i first discovered that behavior [13:44] bac: thanks. i'm going to file a bug on this, unless you know there's already one open [13:45] barry: should be a super easy fix. i don't know if there is a bug yet. it's worth a discussion on the list, though, to ensure there is no reason for why it is so. [13:45] bac: cool, thanks [13:56] intellectronica: I saw you tried to land the lazr.restful thing on pqm last night. I got a bunch of librarian-related failures on mine that seem verrry unlikely to be caused by my change (a two line patch that casts a token from unicode to str and passes on ec2test). Did yours succeed? [13:58] gary_poster: yeah, my patch landed [14:00] intellectronica: huh. ok, I'll just try again, I guess. I have a branch I'm working on now to switch lazr.* out of sourcecode/pqm land, but canonical-identity-provider and shipit will be pqm forever atm. :-/ thanks [14:01] gary_poster: oh cool, does that mean that we'll converge on a single restful trunk which will be included using buildout? [14:01] intellectronica: yes [14:01] excellent! [14:20] EdwinGrubbs, ping [14:20] deryck: pong === carlos_ is now known as carlos [14:52] Why wasn't edge update? [14:54] sinzui: hi there, I have a case where object doesn't have canonical_url defined (and it shouldn't), but there are subpages on it [14:54] ?? [14:54] sinzui: that totally breaks with new templates with something like "NoCanonicalUrl: No url for because broke the chain." [14:55] danilos: what breaks? The code that makes the branding? [14:55] yes, I bet that is it [14:55] sinzui: yeah, or perhaps breadcrumbs (I'd suspect breadcrumbs) [14:56] if there are breadcrumbs :) [14:56] danilos: the breadcrumb code is from 2.0 [14:56] reviewers (and lurkers) -> #launchpad-meeting in 4m [14:56] danilos: The branching searches (traverses) to IHasLogo [14:57] sinzui: it might come as a surprise then that this page was never migrated to 2.0 either (I am actually trying to use generic-edit.pt for the form) [14:57] danilos: Send an email to thumper. He is creating an alternate way to get branding...IRootContext. I think his solution will fix your problem [14:58] sinzui: right, it does fail in IHasLogo, will do that [15:00] reviewers (and lurkers) -> #launchpad-meeting [15:00] Chex: hi there! could you delete the following vcs-import when you have time: https://code.edge.launchpad.net/~vcs-imports/windmill/trunk ? [15:01] Chex: nevermind, it seems i have permission to do so [15:01] danilos: thanks for bring this up. your use case is better justification than the ones I was thinking of [15:01] sinzui: sure, thanks [15:02] flacoste: ok, no problem :) [15:08] beuno: do you want to join our reviewer's meeting? [15:16] flacoste, http://morgamic.com/2009/08/18/optimizing-your-js-and-css/ [15:17] beuno: tell me something i don't already know :-) [15:17] flacoste, this is my attempt to keep the CSS minifying on the table, rather than informing! :) [15:18] did edge rollout this morning? [15:18] something seems to have happened at ~08:44 [15:19] james_w: i think there was an edge server that didn't updated properly [15:19] james_w: btw, we should be upgrading to HAProxy soon for load balancing, that should really makes thing better for you [15:19] as it catches when backend server are not available [15:20] james_w: https://wiki.canonical.com/IncidentReports/2009-08-19-launchpad-edge-fail/ [15:20] that might be related [15:20] I'm still seeing issues though [15:20] not described there [15:21] james_w: what issue? [15:21] hmm, might have *just* gone away [15:21] bug 415943 [15:21] Bug #415943: branch.lp_save() often give PreconditionFailed [15:21] nope, still there [15:22] leonardr: could you help out james_w in diagnosing the source of bug 415943? [15:22] Bug #415943: branch.lp_save() often give PreconditionFailed [15:22] flacoste, sure [15:22] thanks === henninge_ is now known as henninge [15:23] ah, a little more insight I think [15:25] so something changed this morning that means lp_save() on a branch will now fail if you didn't change anything [15:25] but before it worked whether you changed it or not [15:26] james_w: the first step in diagnosis is to set httplib2.debuglevel = 1 before creating the Launchpad object [15:26] this will give you dumps of the http requests/responses [15:26] we're especially interested in the http headers ETag, Last-Modified, If-Modified-Since, and If-None-Match [15:27] since those are the ones that determine whether the server will give you Precondition Failed [15:28] http://paste.ubuntu.com/255765/ [15:29] you commented on bug 336866. that bug was caused by an interaction between an ObjectModifiedEvent and the unmarshalled field cache [15:29] Bug #336866: When adding tag or updating description, lp_save() gives "HTTP Error 412: Precondition Failed"

[15:29] i backported a fix to the lazr.restful branch used by launchpad, but apparently launchpad didn't pick it up, because people are still having that problem [15:29] are there any interesting ObjectModifiedEvents triggered when save() is called on a branch object? [15:30] I think my comment on that bug was mistaken [15:30] I think when I was getting this before the errors were legitimate, as scanning updates the last_scanned entry in the branch or something [15:30] and that would happen shortly after pushing, around the time I was setting the status [15:31] I implemented retry support to handle that [15:31] but this morning the retries don't help [15:31] further to the paste above, retrieving the object again gives the same ETag: 7bdbf8234351ddfa1c58adf96ab423eb41ab8f6e [15:31] thanks, i was just about to ask that [15:31] so something fishy is going on [15:33] james_w, can you reproduce this in your local launchpad installation? [15:33] I haven't tried [15:33] sinzui: It seems like account creation refuses to create accounts for email addresses used by SSO, etc, but password recovery only works for Launchpad accounts. Is that possible? example address: paul.szwajkiewicz@wp.pl [15:34] salgado: ^ [15:34] james_w: please give it a try, as the only ways to debug it at this point are by backtraces in the code or manual examination of the code [15:34] on the http level everything looks like it should be working [15:34] sinzui: where is PillarView tested? [15:34] abentley: I believe this /was/ true, but salgado fixed it [15:35] leonardr: yeah, that's why I'm asking if edge rolled out today, that would narrow it down [15:35] abentley, password recovery on login.launchpad.net should work for SSO accounts [15:35] I'll try and get launchpad.dev fired up [15:35] lp/registry/browser/tests/pillar-views.txt [15:35] sinzui: It seems to not be fixed as of r9129 [15:35] adeuring: gary_poster mentioned something about failures related to the librarian he had trying to land branches [15:36] intellectronica: thanks! [15:36] sinzui, salgado: Try the example address I gave on edge. [15:36] sinzui: thanks, grep was not my friend [15:36] sinzui, salgado: paul.szwajkiewicz@wp.pl [15:36] adeuring, intellectronica: yes, I don't see the background in history, but I got a *lot* of librarian failures on pqm [15:36] abentley, if that email is of an SSO account, then it can't use edge.launchpad.net [15:36] adeuring: I just retried, since intellectronica had success [15:37] gary_poster: OK, I'll try that [15:37] salgado: Just tested on prod, and it also doesn't work there. [15:38] abentley, prod shouldn't work either [15:38] only login.lp.net should work [15:38] salgado: !? [15:38] abentley, this is not a launchpad account -- it's an SSO one [15:39] salgado: A user tried to create an LP account and was told they couldn't, because the address was already in use. So they tried to recover their password, and they couldn't, because it wasn't a Launchpad account. [15:40] abentley, as I said earlier, they can reset the password on login.launchpad.net. once that's done they can use the new password to log into launchpad.net [15:41] salgado: I think that we should make that much easier for users, potentially by letting them recover an SSO password from launchpad. [15:43] salgado: There is nothing to indicate to the user that they need to do this. [15:45] abentley, agreed, but when we prevent the user from registering we say that they can use the SSO credentials to login [15:45] (agreed about making it easier, I mean) [15:45] salgado: But we don't say how to recover the SSO credentials, so that's not really enough. [15:49] salgado: Is the launchpad.net password recovery a subset of the login.launchpad.net password recovery? If so, I think it would be sensible to always use the login.launchpad.net password recovery. [15:49] it's not [15:49] abentley, ^ [15:50] salgado: What are you pointing me to? [15:50] the line above, where I said it's not a subset [15:53] salgado: Okay, in that case it would be great if we could add "If you have forgotten your password, use login.launchpad.net (not this page) to recover it." [15:53] barry, I think you forgot that we decided I'd attend the AsiaPac reviewer meetings now. [15:53] salgado: To the specific error. [15:54] rockstar: right! === Ursinha is now known as Ursinha-food [16:01] abentley, agreed, can you file a bug about that? [16:01] I'm sick of IRC clients now. Can we please go back to carrier pigeons? [16:02] * gmb goes to find a cup of tea === gmb is now known as gmb-afk [16:02] gmb: Actually, I'm using a Pidgin right now :-) === gmb-afk is now known as gmb [16:02] abentley, boo. :) [16:02] Argh. Didn't want to change my nick. Damn thing... [16:02] leonardr: I can't seem to reproduce with the latest devel === jtv is now known as jtv-night [16:03] aaargh [16:04] james_w: what about my earlier question about ObjectModifiedEvents? [16:04] that's almost certianly not the problem because you'd have it reliably [16:07] " are there any interesting ObjectModifiedEvents triggered when save() is called on a branch object?" ? [16:08] yes. more generally, does anything unusual at all happen when you save a branch, that doesn't happen for other objects? [16:08] I assumed you were asking your fellow LP devs, I don't know the answer, sorry [16:08] oh, i see [16:08] I've just updated the bug [16:08] rockstar, maybe you can help? [16:08] lp.me.lp_save() also gives the error [16:14] flacoste: New python-apt available in ~maxb/launchpad for syncing to ~launchpad/ppa (in response to new version in karmic primary) [16:14] maxb: great, thanks! i'll copy over [16:15] done [16:16] leonardr, james_w, we haven't gone out of way to do anything special on save through the API. What's the actual issue? [16:16] rockstar: james_w gets a 412 Precondition Failed every time he saves a branch object [16:16] apparently this just started happening recently [16:17] Hm, does it say what the precondition is, or give an oops? [16:17] rockstar: well, the precondition is almost certainly an etag failure [16:17] meaning the server-side calculated etag doesn't match what he sent [16:17] * rockstar hasn't gotten that far in the Rest book yet... [16:17] {'status': '412', 'content-length': '0', 'via': '1.1 wildcard.edge.launchpad.net', 'x-powered-by': 'Zope (www.zope.org), Python (www.python.org)', 'vary': 'Cookie,Authorization,Accept', 'server': 'zope.server.http (HTTP)', 'x-content-type-warning': 'guessed from content', 'date': 'Wed, 19 Aug 2009 15:05:22 GMT', 'content-type': 'text/plain'} [16:17] but when he GETs the object again immediately after the failure, it still has the same etag [16:18] So the etag is kinda like a hash then? [16:18] mrevell: just watched the screencast... Love it! The only thing missing is that the screen that says "The Next Day" should be while lettering on a scratchy and wobbly black background, bordered in Art Nouveau style, and accompanied by piano music. [16:18] white. White lettering. [16:18] jtv-night: :-D [16:19] thanks jtv-night [16:19] :) [16:19] rockstar: yes [16:19] it's calculated from the current values of all the object's non-read-only fields [16:19] james_w, do you have a script I can test? [16:19] rockster: he put a really simple on in the bug [16:19] lp.me.lp_save() [16:20] or yeah, see the bug [16:20] [16:20] Bug #415943: lp_save now gives PreconditionFailed if unchanged [16:20] james_w, where's the bug? [16:20] james_w: you're getting it on lp.me?!?! [16:20] yeah, I said that 10 minutes ago! :-) [16:21] i must have missed it in the confusion [16:21] ok, so it's a thoroughly general problem [16:21] and since it only shows up in production i think it might have something to do with an intermediary [16:21] such as a cache [16:21] rockstar: nm, it isn't a problem with branches per se [16:22] So I'm off the hook then. [16:22] Yay! [16:23] james_w: give me a minute to get a branch into review and then i'll pay more attention to this [16:23] thanks [16:33] sinzui, how much time should I spend trying to improve https://devpad.canonical.com/~salgado/poll-shots/poll-30.png? === beuno is now known as beuno-lunch === matsubara is now known as matsubara-lunch [16:40] salgado: Did we commit to redesigning that page? [16:40] * sinzui looks [16:41] sinzui, I hope not. where is the list of pages that we committed to redesign? [16:42] salgado: I would not spend more than 1h cleaning the presentation. [16:44] ok === salgado is now known as salgado-lunch [17:07] yo barry. We didn't talk about transferring the wiki stuff that I saw in the reviewers mtg. that should be a reviewers mtg action item for next week, and we ought to coordinate. Maybe coordination should just be as simple as pinging the other person whenever we start doing a bit of it, so we don't bump into each other [17:08] james_w: i can't duplicate your problem with launchpad.me.lp_save(), either on staging or edge [17:09] neither can I now :-/ [17:10] nor with your original load() code [17:11] james_w: are you willing to come back to this if/when it recurs? [17:12] i think the problem may have been some transient problem with the cache [17:12] I'll be back, don't worry [17:12] it's rather unsatisfying though [17:13] agreed [17:13] the last failure I have is slightly before 17:00 [17:15] danilos: http://paste.ubuntu.com/255805/ :-) [17:15] Note the huge block under wgrant's name. [17:16] kfogel: yeah, looks great :) === bigjools is now known as bigjools-afk [17:16] james_w: when it happens again we should probably get the losas involved, because a problem with this profile is almost certainly being caused by an intermediary [17:16] since the http looks fine [17:17] something dropping Etag in the response perhaps? [17:18] james_w: if something were modifying the ETag on the request before launchpad saw it, launchpad would say "this doesn't match" and give a 412 [17:18] i know that apache plugins for things like content compression munge the etag on the response (in a way that defeats the purpose of etags) [17:18] yeah, that's what I meant [17:18] nothing like what I actually said, granted :-) [17:19] but i don't know of anything that would modify it on the request [17:19] did anyone do anything 20 minutes ago? === Ursinha-food is now known as Ursinha [17:19] restart servers, move a cache or anything? [17:20] sinzui, I just noticed that question-add.pt is not in the zcml but in the view code. Is there a reason for this? [17:23] hi! does launchpad have a trademark policy? i'm making a project that stores bug data from launchpad into couchdb, and I was considering calling it "offpad", but I don't want to have any hassle over the name [17:23] short for "offline launchpad" [17:23] kfogel, ^^ [17:24] leonardr: I updated edge - the auto update failed overnight so we had servers on different versions, was just fixing that [17:24] james_w -^ [17:24]

A public poll that has not opened yet