[01:44] thumper: Did you notice that ~vcs-imports have odd extra privileges when registering projects, too? [01:45] ~vcs-imports members can set the owner and license_reviewed flag of projects when registering a new project, specifically [01:56] maxb: I don't think so... [01:56] maxb: that is registry experts [02:38] I can and do set every project I register or change as reviewed. And anyone in Registry Admins can too [03:38] sinzui: ping [03:44] sinzui: emailed you [04:03] hi thumper [04:16] lifeless: hi [04:16] lifeless: aren't you at UDS? [04:17] yes [04:17] lifeless: early start then? [04:17] jetlag [04:18] had stomach ache style crookness yesterday afternoon, crashed out, woke at 4am [05:53] lifeless: hey! we had similar days then ... I've been up since 2am after having gone to sleep at 6:$0 [05:53] 6:30 [05:54] thumper: you know what would be great - if merge proposals that were dependent on each other were sorted so that the one that includes all of the other ones was really easy to find... [07:57] Where is the mailinglist unsubscribe link hidden? [08:00] stub: It's pretty obvious on the team page for me. [08:00] In the 'Mailing list' section, an actual button labelled 'Unsubscribe'. [08:01] My eyes must have skipped the button looking for icons. duh. [08:02] Yeah, it is just about the only button. === almaisan-away is now known as al-maisan [09:53] mwhudson: were you publishing runs of pydoctor on launchpad anywhere? [09:53] jelmer: http://people.canonical.com/~mwh/canonicalapi/ [09:54] mwhudson: thanks! [09:54] also http://people.canonical.com/~mwh/bzrlibapi/ fwiw [09:55] mwhudson: I've added a local rule for pydoctor as well, but for some reason it doesn't seem to include the css stuff [09:55] (https://code.edge.launchpad.net/~jelmer/launchpad/pydoctor, couldn't find anything existing in the launchpad sourcecode) [09:56] jelmer: css stuff? [09:56] mwhudson: I'm not getting any of the colors that the URL you just mentioned has [09:59] jelmer: i can't remember how that works :-) [10:00] mwhudson: hmm, I'll have a look then [10:01] jelmer: but i'm pretty sure it should just be copying the css file into the directory [10:03] jelmer: yes, it should copy apidocs.css to the target directory [10:05] mwhudson: what command are you running to generate the html docs at that location? [10:05] bigjools: Bug #527551 [10:05] Bug #527551: Intra-archive copying of a source with a failed build may leave that source uncopyable [10:06] ~/pydoctor-2a/bin/pydoctor ~/db-devel/lib/lp ~/db-devel/lib/canonical/ --make-html --html-output ~/public_html/canonicalapi/ -q -q --verbose-about epytext-summary --project-name=Launchpad --docformat restructuredtext [10:06] jelmer: ^ [10:06] mwhudson: thanks [10:20] mwhudson: whoa, that's spitting out a lot of warnings in the new version [10:20] yeah probably [10:21] about a billion launchpad docstrings aren't actually valid rest [10:21] for one thing [10:23] jelmer: how goes the bzr-git, dulwich updates? [10:29] thumper: oh, right. I've deferred any bzr stuff so far, but I'd like to spend some time on bzr things this UDS. [10:29] * thumper nods [10:45] mwhudson: If you have a chance, can you review the trivial branch at https://code.edge.launchpad.net/~jelmer/launchpad/pydoctor/+merge/25133 ? [10:45] * jelmer goes back to the more complicated stuff [11:01] mwhudson: thanks [11:57] bigjools, wgrant: where are we at wrt getting visible stats for ppa usage? [11:58] jml: The script tried to murder germanium. [11:58] It nearly did, too [11:58] wgrant, after which rollout? [11:59] The first run of the librarian log parser would have run into this too, but I don't know how it was solved then. [11:59] jml: It was run a couple of weeks after 10.03, IIRC. [11:59] I'm thinking I'll just tell it to stop after processing some x lines of log, where x is yet to be defined. [11:59] So it will take lots of runs to process the existing logs, but it won't eat all of the RAM. [12:00] The API stuff is all landed, but there's no UI at all yet. [12:00] you mean there's a manual process to get the historical data in? [12:02] * jml has to go to a GTG thing [12:31] Is there an ETA on a CP for the copy-packages bug? [12:42] It will be submitted to PQM in a few minutes. So, soon. === matsubara-afk is now known as matsubara [15:12] Anyone else seeing rocketfuel-get fail on the CSS build? The problem seems to be some extra dots between classes in l/c/l/javascript/test.css [15:25] mars: do you see rocketfuel-get failing on the CSS build? See above. [15:26] jtv, haven't seen it. CHR today. sinzui or noodles785, ^ any idea what's going on? [15:27] mars: it went away when I fixed l/c/l/javascript/test.css [15:27] gary_poster: ping? [15:27] * sinzui knows nothing [15:27] mars: it had lines of the form ".class1..class2.class3 {" [15:27] BjornT: pong [15:28] jtv, "bzr praise"? [15:28] ooh, didn't know about that one... do we have a secret backchannels for stats on blame/praise? [15:29] gary_poster, have a few minutes to spare ? [15:29] gary_poster: when running 'make', i get a lot of: 'import site' failed; use -v for traceback [15:29] gary_poster: what's that about, and to what should i pass -v? [15:29] mars: sinzui. [15:29] jtv, ? blame/praise/annotate are the same. If you see a funky line like that, I would say check the source of the build chain. Something changed, and the machine didn't do it :) [15:29] ha! [15:30] mars: the machine didn't do it? there's a novel concept... we meatbags usually help, but... [15:30] BjornT: you can typically pass -v to the interpreter, so bin/py -v should tell you what is going on [15:30] BjornT: and generally, I don't know what that's about, but it's not good [15:31] When you gimme the -v output that will tell me [15:33] manish, I saw your email. I sympathisize but I don't know the answers to your questions. leonardr will not be here until next week, as I said. flacoste could possibly know. If this is just a matter of interpreting the wadl spec then I can try to help, but you've read the spec much more than I, I'm sure. I'll reread your email now to make sure I don't see anything more concrete I can offer... [15:34] gary_poster, Thanks. I have read the spec, but the spec is just too flexible and extensible ( a bit too much). I can try to contact the people who created WADL spec, but I hardly expect any reply from them. [15:35] gary_poster, anyway leonardr created the LP WADL? or flacoste ? [15:36] manish, I reread your email. if I had to guess, I would say that you will want to be able to send parameters and representation_type together === Ursinha is now known as Ursinha-afk [15:36] gary_poster, Nope. I just wanted to know if both can be together? [15:37] but as the spec says multiple representation_type means multiple ways to pas data which means overloaded method ? right? [15:37] manish, I think so. I'd give it 90% probability yes. [15:37] manish: leonardr is responsible for this infrastructure, and is the WADL expert [15:37] gary_poster, If both can go together then it makes the situation more complex :( [15:38] manish, yes. however... [15:38] till now i was working thinking both can mutually exclusive [15:38] s/can/are [15:38] manish: I think launchpadlib itself only ever requests JSON [15:39] gary_poster, sorry I didn't get it. Can you rephrase it please? [15:39] gary_poster: ok, found the problem: ImportError: Bad magic number in /devel/launchpad/rocketfuel/lib/site.pyc [15:39] gary_poster: this makes the problems go away: http://pastebin.ubuntu.com/432258/ [15:40] BjornT: typically that means that your pyc was from a different version of Python [15:40] BjornT: +1 on that patch [15:40] gary_poster: cool, i'll merge it then, thanks [15:41] thank you BjornT [15:41] manish, rephrasing... [15:48] StevenK, do you know what's pretty hard? [15:48] StevenK, getting it right the first time :( [15:49] jml: I bet [15:49] What broke? [15:49] mars: it's getting late here and I'm putting out a fire... any chance you could pick up that css issue? [15:49] jml: Does that mean there is further stuff to be abstracted? [15:49] StevenK, ish. I made a couple of mistakes when I did it. [15:50] manish, I believe it is theoretically and practically possible to ask for different representation types. We do this in Launchpad, in fact: in our browser API calls, sometimes we ask for HTML back instead of JSON. [15:50] *However*, in launchpadib, I believe we only ever ask for one response type: JSON. Therefore, for your case, I believe the following is reasonable advice: [15:50] 1) your C# code should generate code that can accept parameters. [15:50] 2) it should never have to accept a representation type, because you always want JSON back, (which your code will process transparently.) [15:50] 3) if you ever want to make code that used a different representation type, you could have it specify a representation type in a different place. I do not think you should implement this now. My personal idea would be that you could generate a root object that sends a different representation type be default. You also could change it on the fly: a Python version might be launchpadlib.representation('application/xml').foo( [15:50] jtv, please file a bug and assign it to me. I'm a bit buried myself, and doing CHR. [15:50] mars: oh yes forgot about that... maybe we should just bully sinzui into fixing it? [15:50] gary_poster, a minute. I need time to read and understand [15:51] *That* is what I forgot. Breakfast, lunch & dinner. [15:51] stub: just heard some pretty rousing words from Dr. Weng... this ain't over yet. [15:51] jml: Should we do another branch that sorts out more abstraction issues? [15:51] manish, of course. when you have digested what I wrote: did that make sense, and does that seem helpful? If so, I'll put it on the list so that you and others can comment on it, and leonardr and flaocste can see it later. [15:52] jtv, what exactly is broken? [15:52] sinzui: lib/canonical/launchpad/javascript/test.css [15:52] and as a consequence, rocketfuel-get [15:52] StevenK, yeah, probably the best thing to do is use pipelines or whatever to split out that level of work from the poppy level of work [15:52] gary_poster, it did make some sense. Well I know that I should expect only JSON [15:52] sinzui: it's got some "class1..class2" (note double dot) [15:52] StevenK, because the only sane way to drive the abstraction stuff now is by actually doing poppy [15:53] jtv r=me to remove it [15:53] (now that all the mentions of "codehosting" and "bazaar" have been factored out) [15:53] gary_poster, till now what I was doing was that - either param[] or represenation_type[] can be present [15:53] jml: Pipelines are one of those that I saw and went "That's cool", but I have no idea how to drive [15:53] sinzui: late night & putting out fire [15:53] one of those things, even [15:53] sinzui: I was hoping I could guilt or bully you into it. [15:53] StevenK, me either, I just do it the hard way by having two branches and merging one into the otther [15:53] jtv: I will do it during my lunch [15:53] gary_poster, when param[] is present, i create a method with returnType foo (param1,param2... paramn) [15:54] sinzui: wayyy to bounce the guilt back. Well played. :-) And thanks! [15:54] gary_poster, and when represenation_type[] I create n methods and in each method I expanded a represenation_type and that becomes it's parameter [15:55] jml: That's sounds like work. Let me investigate pipelines [15:55] StevenK, :D [15:58] bigjools, does poppy have a config section? [15:58] manish: ok. I'm personally thinking that you should just ignore representation_type for now. It will make using your library in (what I believe will be) the common way easier. Later...if you want to support it I'd suspect you'd want to do it a different way. it strikes me that we could be looking at lazr.restfulclient and seeing what leonardr does with representation types. Let me see if I can find that... [15:58] gary_poster, thanks a lot [15:58] sure [15:59] gary_poster, I have one more issue in this WADL [15:59] manish, ok [15:59] have a look at https://edge.launchpad.net/+apidoc/1.0.html#branches [16:00] getByUrls method [16:00] there is no return type specified [16:00] even the WADL doesnt have anything [16:00] jml: no [16:01] jml: pipeline looks full of goodness [16:01] gary_poster, http://pastebin.com/kqZtj9zW is the section in WADL for branches.getByUrls [16:02] gary_poster, so the C# code generated is "public void branches_getByUrls(string urls)" [16:04] bigjools, oh, I see. it just hardcodes strings [16:05] manish: I think it returns an application/json representation of multiple branches (a sequence) [16:05] gary_poster, difference between think and present. From the name I can make out, but it is not present in the WADL [16:05] which mean I need to hand modify the WADL or the generated code which is BAD :( [16:06] from name I was sure it returned a branch list [16:06] jml: Where are you heading? [16:06] but being absent in WADL and the documentation makes the case weak as every update to the app will break since every time the WADL has to be hand edited [16:11] manish, I understand your concerns. I don't know enough to say the cause or the best remedy. My suggestion is that you open a but against launchpad-foundations (https://bugs.edge.launchpad.net/launchpad-foundations/+filebug) about the WADL. I'm hopeful that we can generate better WADL for you. Either way, we can record the conversation there. [16:13] gary_poster, sure. I will file. I just wanted to bring it to notice to some LP dev [16:18] gary_poster, done. should I subscribe Leonard too? [16:19] manish, that's ok. thank you! [16:19] https://bugs.edge.launchpad.net/launchpad-foundations/+bug/579516 [16:19] Bug #579516: Launchpad WADL's "branches" section with method getByUrls doesn't have a return type [16:19] got it [16:31] mars, sinzui: IRT https://bugs.edge.launchpad.net/launchpad-foundations/+bug/537298 and https://bugs.edge.launchpad.net/launchpad-foundations/+bug/575254 , do we know conclusively if the trigger is repeating background images? Do either of you know of someone able to dupe? If that is the case, we now have five reports. Can we trivially fix the CSS to possibly make this go away? [16:31] Bug #537298: Lucid firefox renders Launchpad very slowly [16:31] Bug #575254: Switching to a tab containing Launchpad is extremely slow in Firefox [16:33] gary_poster, I do not know anything about the issue. I could speculate the size of compressed CSS and JS frustrates FF. [16:33] * sinzui uses Epiphany and is very happy [16:33] heh [16:34] gary_poster, I wonder most engineers are not Chrome users...they do not see the problem [16:35] gary_poster, no one could conclusively narrow down the issue. Major sites are also affected. It really is Firefox' problem, not the Internet's. [16:35] mars, Thanks for the sober summary. WONTFIX [16:35] ok thanks, guys, [16:37] sinzui, social puzzle for you: what the heck do I do with this? https://answers.edge.launchpad.net/launchpad/+question/110662 [16:39] mars: wtf [16:40] mwhudson, excellent summary :) [16:40] lol [16:40] mars: i hadn't read the comment [16:41] that makes me happy the way Dali sometimes makes me happy [16:46] mars: That is possibly the best thing I've read in *days* [16:57] jml: #python-testing may interest you [16:58] jml: I plan to make testtools only emit one outcome per startTest/stopTest pair [16:58] jml: any objection? [16:58] lifeless, I'd like to see an email explaining why, if that's ok [17:01] jml: I blogged a while back :) [17:01] and I'm sure there is a bug; digging [17:01] bug 335816 [17:01] BjornT, danilos, your reviews kindly requested: https://code.launchpad.net/~jtv/launchpad/bug-578331/+merge/25168 [17:01] Bug #335816: tests report multiple results [17:01] jml: ^ [17:02] lifeless, thanks. [17:02] adding details there, now. [17:02] ok thanks. [17:02] lifeless, I'm doing some twisted hacking w/ StevenK right now so my brain is a bit full [17:03] sure [17:03] I want 'u' in launchpad bugs === matsubara is now known as matsubara-lunch === Ursinha-afk is now known as Ursinha === beuno is now known as beuno-lunch === Pilky_ is now known as Pilky [18:06] kfogel: so, rob [gnash] is pinging me, face2face === matsubara-lunch is now known as matsubara === cperrin881 is now known as cperrin88 [18:14] Hey Peaople [18:14] I'm not sure if I'm in the right channel but I will ask [18:14] Can anyone help me with the Apache HTTP Commons lib and the OAuth signing process of launcpad [18:34] gary_poster, are you able to help cperrin88 with the OAuth process? Or is that something for Leonard when he gets back? [18:37] The problem is driving me nuts because I don't see the mistake [18:37] and I can't sniff a damn https connection >.< [18:37] mars, cperrin88, that's probably leonardr I'm afraid, next week. If someone else can help, they are probably doing UDS things [18:37] *sighs* [18:38] cperrin88: I'm not familiar with that lib. Does this mean that you want to use the REST webservice without launchpadlib? [18:39] Yep [18:39] I mean [18:39] I have no other choice [18:39] Android is Java [18:40] cperrin88: right, got it. flacoste (who is not here) would be able to help; salgado (who is not here) might be able to help. If you give focussed questions there's a chance I can dig up answers for you, but that's all I can offer now, I'm sorry. [18:41] That's a start [18:41] well I'm trying to get a request token as described here https://help.launchpad.net/API/SigningRequests [18:42] ack [18:42] mm, sorry, I meant, got it [18:42] ack ;) [18:42] :-) [18:43] I'm using the Apache HTTP commons for this request as they are preinstalled [18:43] ok [18:44] I have tried to send this requst to non SSL server and it looks like this http://pastebin.com/E8SrwLjv [18:44] of course i send it to edge.launchpad.net over https and to the path /+request-token [18:45] If I send this as a raw HTTP request over a tool named Fiddler, I get what I want [18:45] If i send this with my code I get a 401 Unauthorized [18:46] cperrin88: (probably where you are going, but to join along...) so that seems to suggest the HTTP Commons lib is doing something a bit different than the vanilla request? [18:47] probably [18:47] cperrin88, you said you can't sniff the connection: does Apache HTTP commons have debug logging output that you could inspect (assuming your application has logging)? [18:47] Have you gotten a look at the actual request over the wire? (assuming the answer is "I can't" for some reason :-) ) [18:47] log4j or Apache commons logging, I don't remember which, been too long [18:48] I can log some things but not what I need to get a clue [18:48] What I can log seems correct [18:49] It would be cool the see what the server really gets [18:49] but I don't have a SSL cert [18:49] so I can't write a logging server [18:50] cperrin88: wireshark an option? It seems to have SSL support, based on googling around [18:50] nope [18:51] for my own knowledge, why not? [18:51] I'd need the private key of the launchpad cert [18:51] ah right [18:51] ... [18:51] If I had that first S in SSL would be in vain for launchpad [18:51] :-) [18:53] I think i got an idea [18:53] cperrin88: an off-chance, but does the body of the 401 response seem to have anything I could do something with on my side? [18:53] nope [18:53] it's empty [18:53] * gary_poster is happy cperrin88 has an idea [18:53] ok [18:53] cperrin88, maybe try stunnel as a man-in-the-middle, as described here: http://codepainters.wordpress.com/2010/01/23/androids-https-implementation-slow-under-debugger/ [18:53] I will try to write a logging app on app engine [18:53] that has htttps === beuno-lunch is now known as beuno [18:55] fwiw, in Python something else I'd try is actually hacking the underlying library to tell me what the heck is going on, but that can be problematic. That said, *some* way of seeing what you are actually sending us does seem like a pretty reasonable diagnostic tool to have. [18:55] so sure, the https app engine logging engine sounds easyish [18:56] I like mars' idea too [18:57] I was reading that [18:59] cperrin88: no idea what I'm talking about, but http://hc.apache.org/httpclient-3.x/logging.html ? [19:00] about to go on call; back in a bit [19:04] gary_poster: I don't have that lib in Android SDK [19:04] but I will try it another way [19:04] ok [19:08] cperrin88, Can you tell your problem again clearly. This is nearly the same problem I had when I started with LP APi [19:08] I build my requst with APche HTTP commons for a requst token and everythime i send it I get a 401 Unauthorized back [19:09] a vanilla request, using a tool named fiddler, works [19:09] It drives me nuts [19:10] cperrin88, I have used Fiddler. That was what I too used to figure out my problem [19:10] you get a 401 even in step 1? [19:10] oauth_consumer_key=LaunchDroid&oauth_signature_method=PLAINTEXT&oauth_signature=%26 [19:13] cperrin88, ^ [19:14] yep [19:14] but fiddler works [19:14] cperrin88, you sending it to? https://edge.launchpad.net/+request-token [19:14] https or http? [19:15] https [19:15] but http woudl refirect my request [19:16] *would redirect [19:16] oh. crap. This is even weirder. What is the content of the exception you get? [19:16] means the exception description [19:18] cperrin88, you using this? http://developer.netflix.com/docs/Security#0_18325 [19:19] nope [19:19] your HTTP request shows this? [19:19] I just send my request somewhere that doesn't use https to sniff it once [19:19] Host: api.netflix.com at http://pastebin.com/E8SrwLjv [19:19] I know [19:20] you can use HTTPS withj fiddler too [19:20] *with [19:21] cperrin88, using fiddler with https is a bit tricky, but easy http://www.fiddler2.com/Fiddler/help/httpsdecryption.asp [19:22] I know but https commons is a bit tricky with selfsigned certs [19:22] I'm jsut writing a logging server [19:22] I'm nearly don [19:22] e [19:22] that's good [19:23] cperrin88, but I *strongly* using Fiddler for HTTPS since it saves a lot of headaches. Even I had this problem. Fiddler solved it [19:23] and yeah with the help of wgrant [19:23] manish, btw, I'm replying to your email with a summary of what we discussed on IRC. I doublechecked and lazr.restfulclient does in fact control the representation type itself (that is, you can't request another representation type, at least in the common cases that I saw). It specifies the JSON, as you'd expect. [19:23] The approach wadllib and lazr.restfulclient take are that wadllib simply is responsible for parsing the wadl--you can ask wadllib resources (like nodes in the WADL) questions like WADL_RESOURCE.parameter_names(media_type) to get a list of the parameter names for that call and parameter. Then lazr.restfulclient uses that library to ask for the JSON parts of things. [19:23] This led me to the thought that you might find it interesting and useful to try porting wadllib and lazr.restfulclient, rather than figuring out your own architecture. [19:24] gary_poster, Thanks. I would look into wadllib and lazr. restfulclient [19:24] getting them now === EdwinGrubbs is now known as Edwin-lunch [20:04] gary_poster: *sigh* I'm don't know more now [20:04] *I [20:05] cperrin88: so, you got the logging server, you have the full request, and it is pretty much identical to what you had before (http://pastebin.com/E8SrwLjv with the replacements)? [20:05] and that same request works with Fiddler? [20:05] Yep [20:06] ... [20:06] that's pretty depressing -_- [20:06] Searching an error that doesn't seem to be existent ..... [20:08] an error taht should not be there [20:13] cperrin88: [20:13] (1) I'm happy to look at a pastebin of the what you got from the logging server, if you want another pair of eyes. [20:13] (2) maybe try to dupe outside of android but with the same library in Java? Then conceivably you could turn on the logging we discussed earlier, if you can dupe? (If you can't dupe that way....!) [20:13] (3) in case there's some horrible fundamental bug in our code, you've duped the *exact* same output you saw in the logging server (for instance, all the same headers all in the same order, and whatever else you can think of) with Fiddler and still gotten success? [20:18] gary_poster: I'm double chacking everything now [20:18] *checking [20:18] ok [20:54] thumper, already there? good morning :) [20:54] there's an oops that I've seen for a while, a ProcessTerminated one, like https://lp-oops.canonical.com/oops.py/?oopsid=OOPS-1557CIW15 [20:55] I was pretty sure you replied the list asking about it, but I can't find the message, I guess I'm (or was) on crack [21:54] Ursinha: morning [21:54] Ursinha: I'm about to get a coffee [21:54] thumper, go for it :) [21:54] Ursinha: that is a job being killed [21:54] an import worker [21:55] thumper, is that a real problem? [21:55] not sure === matsubara is now known as matsubara-afk [22:05] thumper, well, any investigation would be appreciated :) [22:11] ack [23:57] rockstar: ping