[01:52] * karni waves goodbye === smb` is now known as smb [08:34] morning, peeps [09:31] Happy Friday, everyone! :D [09:41] JamesTait: you're way too happy [09:41] czajkowski: Only because I see your face on my Facebook timeline in the mornings. ;) [09:42] JamesTait: ROFLOL [09:48] Chipaca, hello. I'm going to ask for an upload today, with what we have in trunk. Any other change we could have missed? [09:51] davidcalle: nope. Only remaining change in my mind at this moment would be when we move the server to something.ubuntu.com [09:53] Chipaca, ok [11:09] good morning! [11:31] gatox, morning! [11:32] mandel, hi [12:36] brb (in 5min) [12:42] back [12:49] hola gatox [12:49] nessita, hi [12:50] gatox: show-errors is approved, I'm looking at the last one now [12:50] nessita, great! [12:50] nessita, everything looks prettier now? :P [12:50] yes! [12:50] at least a little bit more..... [12:50] jejeje [12:51] gatox: while doing the last review, I found: https://bugs.launchpad.net/ubuntu-sso-client/+bug/944767 [12:51] mmmm maybe i miss a show_overlay there [12:51] gatox: you can queue that up for after the INTERNAL SERVER ERROR [12:51] gatox: yeap [12:51] nessita, yes [12:51] can i assign that to me? [12:52] ah..... i assume that i should queu that :P [12:52] gatox: I assigned it already [12:52] refresh pliz! :-) [12:55] gatox: also reported https://bugs.launchpad.net/ubuntu-sso-client/+bug/944769 (as a Medium) [12:56] nessita, roger that [12:57] gatox: approved [12:57] nessita, thanks [12:57] hello, all! [12:58] alecu, hi! o/ [12:58] holaaaa alecu [12:59] good morning! [12:59] Despite appearances, I am not working until 2PM. So I may or may not do things I should be doing. You are warned ;-) [13:00] ralsina, jeeje ok [13:01] * nessita does not see ralsina [13:01] ralsina: is great you're coming after 2pm, I will have tons of reviews for you :-D [13:03] nessita: awesome! [13:03] nessita, don't scare ralsina [13:03] jeje [13:03] let him enjoy his morning [13:03] nessita: if you have any, I may do one as a communitymember in the morning, as well ;-) [13:08] ralsina, you have to sign the agreement then :P [13:09] lol [13:09] mandel: I have! === yofel_ is now known as yofel [13:45] mandel, gatox: if you have some minutes, would you please review https://code.launchpad.net/~nataliabidart/ubuntuone-control-panel/fix-933576-1/+merge/95557 ? [13:46] nessita, sure! [13:46] nessita, ok [13:55] nessita, I'm off to have lunch and will finish the review when I'm back [13:55] mandel: thanks! === m_conley_away is now known as m_conley [14:14] the really annoying thing about debugging the music player, is that i can't play music while i'm doing it :( [14:17] ugh. 464M for libwebkitgtk-3.0-dbg [14:20] gatox: there is a conflict in https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/934502/+merge/94841 [14:20] nessita, ok.... fixing [14:27] gatox, ping [14:27] alecu, pong [14:27] gatox, did you ever encounter something like this in pyqt? [14:28] """QObject.connect: Cannot queue arguments of type 'MyType' [14:28] (Make sure 'MyType' is registered using qRegisterMetaType().)""" [14:28] :S..... nop [14:28] alecu, what are you doing? [14:28] nessita, approve [14:29] gatox, I'm trying to connect to this signal: http://doc.qt.nokia.com/4.7-snapshot/qabstractsocket.html#stateChanged [14:29] gatox: thanks! [14:29] * gatox looking [14:29] gatox, in that doc it says very clearly: "QAbstractSocket::SocketState is not a registered metatype, so for queued connections, you will have to register it with Q_REGISTER_METATYPE() and qRegisterMetaType()." [14:29] gatox, but I'm not finding the right docs for those two methods in the pyqt docs [14:29] Q_REGISTER_METATYPE() and qRegisterMetaType() [14:29] alecu, give me a sec [14:29] * gatox looking [14:30] alecu, do you have qt assistant installed? [14:30] i wonder if anyone even ever uses this feature in the music store plug-in, where we have clicking on the song in the "my downloads" page, start playing that song in your library [14:30] aquarius_: ping [14:31] dobey, pong === aquarius_ is now known as aquarius [14:31] gatox, seems I don't. what is the package for that? [14:31] aquarius_: you were doing some stuff with rhythmdb queries outside of rhythmbox, right? [14:31] dobey, I was [14:32] aquarius: do you know if you can query it for a list of tracks from a certain library, in the list of libraries? [14:32] I don't know off the top of my head [14:32] gatox, I've got the qt designer, but not the assistant. And I can't find a package for that [14:32] ie "all tracks in ~/.ubuntuone/Purcahsed from Ubuntu One/" ? [14:32] could maybe work it out [14:32] alecu, i think you can install libqtassistantclient4..... if that doesn't work.... you can always install qt4-dev-tools [14:32] ok [14:33] alecu, the web documentation is not loading here....... that's why i mention qt assistant [14:36] alecu, take a look at the answer here, to see if it is useful for you: http://stackoverflow.com/questions/2585442/sending-custom-pyqt-signals [14:37] aquarius: is the code for the queries stuff you were doing somewhere i could look at it? [14:37] alecu, i think that probably using: Q_REGISTER_METATYPE() is more c++ way..... and the one in that page is more python-ish [14:37] dobey, it's only a one-liner, but it's not helpful to you, because it was about querying the ExtDB (where the album art is stored), not the RhythmDB. [14:37] ah right [14:38] dobey, I'm currently looking into it [14:38] i'm looking at the .gir and can't even figure out how to do a query and tell it what to query for [14:38] thanks [14:38] I am a little worried that rhythmdb_query_parse and cousins are not bound into gir :( [14:38] http://developer.gnome.org/rhythmbox/unstable/RhythmDB.html#rhythmdb-query-parse [14:39] hrmm [14:39] it's introspectable=0 [14:42] ah, because a RhythmDBQuery isn't an object [14:42] dobey, but... if you've got multiple libraries, I think they're all separate sources [14:43] so what you want to do is not construct a query; I think you want to get the list of sources, find the one in it that's your library, and then that source's query_model contains all the tracks [14:43] aquarius: nah, they're all under the same source it seems [14:43] really? how do you get the source? [14:43] if you can get the source I ma be able to build a query for it [14:44] libsrc = self.shell.props.library_source [14:48] that only works inside a plugin :( [14:48] external pytohn scripts don't get a shell object :( === IAmNotThatGuy is now known as M0hi [14:52] dobey, ah, you've gotta use it in a plugin; all the other stuff doesn't work safely externally. bah. So, I don't have time to fiddle with that right now :( [14:54] aquarius: right. new rb has some behavioral differences, and i think making it so that if you select the "Ubuntu One" store, you can hit play and it will play your purchased music, would be nice, and get rid of some of the weirdness i'm seeing in fixing other stuff :) [14:57] gatox: ping [14:57] ralsina, pong [14:57] gatox: please talk to Christine on canonical IRC *now* and give her your picture :-) [14:57] ralsina, ok [14:58] gatox: not the darth maul one, please [14:58] ralsina, jeje okok [15:00] briancurtin: http://pypi.python.org/pypi/MacFSEvents [15:01] urbanape: inttttteresting [15:01] me [15:02] me [15:02] gatox, mandel, alecu, dobey? [15:03] urbanape: ? [15:03] me [15:03] me [15:03] me [15:04] alecu: standup [15:04] me [15:04] DONE: lots of reviews, helped gatox with a branch to make sso qt more beautiful, team call, more on u1cp sigin screen branch [15:04] TODO: proposed the branch for the aforementioned UIFe, finish reviews [15:04] BLOCKED: nopes [15:04] NEXT: briancurtin [15:04] * alecu writting notes [15:04] DONE: got to where SSO tests are passing, moving on with client [15:04] TODO: push forward with client tests [15:04] BLOCKED: nope [15:04] NEXT: mandel [15:04] DONE: Debugging why libsoup throws an IO erro when a cert is wrong and ssl-strict is set to false. Some reviews for alecu, [15:04] TODO: finish nessitas review, more debugging. [15:04] BLOCKED: no, I mean, devugging.. [15:04] urbanape, please [15:04] DONE: Made significant progress getting the ubuntu-sso-client tests passing in OS X (had to hard-code the network state). 1 random window size test failure. [15:04] TODO: starting on the syncdaemon stuff. Need to write our own platform package. [15:04] BLOCKED: Nada [15:04] gatox: go [15:04] meh [15:04] DONE: [15:04] Fixed UI stuff and branches finally landed. Couple of reviews. [15:04] TODO: [15:04] Working on INTERNAL SERVER ERROR when executing an installed sso. [15:04] BLOCKED: [15:04] No [15:04] alecu, gord [15:04] go [15:04] :P [15:05] DONE: tunnel server use from Syncdaemon [15:05] TODO: fix some bugs, more of the above [15:05] BLOCKED: stuck with weird Qt signal and metatypes [15:06] dobey: go [15:06] dobey, go [15:07] λ DONE: team meeting, rb debugging, fixed one CRITICAL in rb upstream, fixed click-to-play on 'my downloads' page [15:07] λ TODO: more rb debugging, twisted glib2/gireactor debugging, tarmac locking? [15:07] λ BLCK: none. [15:08] dobey: can you please delay the twisted glib2/gireactor debugging and start working on the installer UI refactopring? the UI fix has way more priority than the other thing [15:08] ralsina: I know you're not here, but do we have the bug for the installer UI refactoring? [15:09] alecu, in https://code.launchpad.net/~alecu/ubuntuone-client/proxy-tunnel-server/+merge/95075 in FakeServerTunnelProtocol why is remote_disconnected empty? [15:10] mandel, probably it's not needed by the fake. lemme check [15:10] dobey: can you please give me an ack? :-) [15:10] I know you're not happy about that, but we have no other choice for that [15:11] there is no bug filed for installer ui [15:11] and all i have for it is a screenshot [15:11] nessita: I don't think there is a bug [15:12] alecu: can I help you with the weird signal/metatype thing? [15:12] ralsina: can you please file the bug with the details (or ask rtgrant to do it)? [15:12] ralsina, if you have some minutes, please do :-) [15:12] alecu: ok! [15:12] nessita: I will ping rtgrant about it [15:13] nessita: and no. the rhythmbox stuff has much higher priority. and it's what i'm concentrating on at the moment. [15:13] dobey: I said to delay the glib/gi debugging, not the RB stuff [15:13] ralsina, did you ever encounter something like this in pyqt? [15:13] """QObject.connect: Cannot queue arguments of type 'MyType' [15:13] (Make sure 'MyType' is registered using qRegisterMetaType().)""" [15:13] ralsina, I'm trying to connect to this signal: http://doc.qt.nokia.com/4.7-snapshot/qabstractsocket.html#stateChanged [15:14] ralsina, in that doc it says very clearly: "QAbstractSocket::SocketState is not a registered metatype, so for queued connections, you will have to register it with Q_REGISTER_METATYPE() and qRegisterMetaType()." [15:14] alecu: are you connecting accross threads for some reason? [15:14] and i think the twisted stuff is also a higher priority [15:14] alecu: because queued connections are accross threads and they have to be marshalled [15:15] ralsina, if I don't use "2" as the second parameter to connect I'm getting segfaults [15:15] alecu: then yes, it's across threads [15:15] alecu: and you are officially in some trouble :-/ [15:15] ralsina, the weird thing is that if I print the "thread.get_ident()" then it seems to be the same. [15:15] * gatox lunch === gatox is now known as gatox_lunch [15:16] alecu: that is weird, yes [15:16] urbanape: mumble? [15:16] ralsina, this is probably qt4reactor playing with our heads. [15:16] alecu: but keep in mind that Qt uses threads for non-blocking I/O so maybe you are later emitting the signal from another thread [15:16] briancurtin: omw [15:16] ralsina, oh, that makes more sense. [15:17] ralsina, the thing is that I looked for .connect() in qt4reactor, and they *are not* using queued connections. [15:17] alecu: hmmm [15:17] alecu: let me think 1' [15:17] which sounds awfully dangerous [15:17] (not you thinking) [15:18] alecu: usually people don't use queued connections, andthey get used automatically if needed [15:18] ralsina, what's "if needed" ? [15:19] alecu: if you are doing it from a non-UI thread [15:19] ralsina, if I'm connecting to the signal, or using the signal? [15:19] alecu: not sure, never tried it ;-) [15:20] ralsina, :-) [15:20] alecu: a 0 (the default) means "auto" and it will guess if it's director queued [15:20] alecu: 0 (default) If the signal is emitted from a different thread than the receiving object, the signal is queued, behaving as Qt::QueuedConnection. Otherwise, the slot is invoked directly, behaving as Qt::DirectConnection. The type of connection is determined when the signal is emitted. [15:21] ralsina, so, state changing signals like "disconnected" or "stateChanged" look to be coming from a different thread at some point... because sometimes the process ends with a segfault. [15:21] (those are QTcpSocket signals) [15:22] as usual with thread stuff, it's impossible to reproduce. [15:23] alecu: right [15:23] so, for instance, if I add a "print" it stops crashing. [15:23] alecu: is it at all possible to do this without connecting that signal? [15:25] ralsina, in order for tests to pass correctly I need the tcp connections to be completely closed. [15:25] The "right" way is to use the disconnect signal, but it's not fired if the socket is looking up the hostname. [15:25] so I'm trying to use the statechanged signal. [15:25] I *could* use a "callLater" with a check of the state()... [15:26] alecu: yes [15:26] alecu: do some polling [15:26] but it makes the test ugly. [15:26] alecu: and locking while looking up is not unusual behaviour [15:26] mandel: hola! [15:26] nessita, dime :) [15:26] mandel: were you able you finish my review? [15:26] ralsina, well, not exactly polling, because I'll still be releasing the cpu with the reactor.callLater... [15:26] ralsina, but it's very ugly. [15:26] alecu: yes, a slow polling ;-) [15:26] nessita, running tests atm and will +1 [15:27] mandel: awesome, thanks! [15:27] ralsina, I'll give it a try, thanks for the input :-) [15:27] nessita, es un placer! [15:27] alecu: sorry I couldn't solve it :-( [15:27] alecu: but you are in the dark, seamy underground of Qt since a while ago ;-) [15:28] ralsina, I guess it would have been easier if I did this in C++; pyqt seems to be focused on the gui parts of qt :-) [15:28] alecu: actually, maybe that's not as crazy as it sounds. Qt'sidea of C++ is very python-like [15:29] ok, need to run a quick erran, brb! [15:30] ralsina, let's see how the tunnel turns out; in any case we can side-port it to C++ later :-) [15:30] nessita, +1 done [15:30] mandel: thanks! [15:31] alecu: indeed [15:32] alecu: may be a fun weekend project, actually (a generic SSL over HTTP proxy tunnel?) [15:36] ralsina, alecu I'll avoid C++ if possible :P [15:36] chicken! [15:37] mandel: C++ is the natural language for this. Or go, if there is a decent network library :-) [15:37] alecu, velociraptor! [15:37] ralsina, alecu go sounds more reasonable for this [15:37] sorry, I'm not here. I'm compiling! [15:37] lol [15:37] mandel, ralsina: Qt bindings for go? [15:38] Some people have a problem. Then they use C++ and now the C++ points to element problem+1 [15:38] alecu: not exists, AFAIK [15:38] alecu, you can call c from go, so we just look at the vtables :P [15:38] and my C++ joke has a syntax error [15:38] mandel: and do the mangling manually [15:38] did that once with c... painful! [15:39] ralsina: at least you aren't using templates in that joke [15:39] mandel, and it only worked for one compiler only... since they mangle differently. [15:39] briancurtin3: I could try to work that in, but it would specialize on compile time. [15:39] alecu, yep.. it was a very hacky thing [15:39] and produce a 50,000 word error message [15:40] briancurtin3: which amounts to "you left a space before the >" or "you missed a space before the >" [15:40] :) [15:40] briancurtin3, ralsina well, after dealing with erlang I think thisfred is ready to read template errors [15:40] I explicitly declined to deal with erlang, mostly ;) [15:41] mandel: you know C++ templates are turing complete? [15:41] mandel: you can write your program in *templates* with no actual code. [15:41] Just like XSLT then ;) [15:41] http://stackoverflow.com/questions/189172/c-templates-turing-complete for example [15:42] jikes!!! ugly ugly.. [15:44] * ralsina considers generating templates from XML descriptions via XSLT [15:45] ralsina, if you that I wont talk to you again, I'll use alecu as a proxy [15:46] http://gitorious.org/metatrace a C++ compile-time raytracer. You write the scene, you compile it, the image is in the binary. [15:46] Which is the greatest thing I have seen in the last 5 weeks at least [15:46] mandel, I'll be your tunnel [15:46] alecu: yikes [15:47] alecu, that is what she said.. [15:47] lol [15:47] nessita, if i'm right..... and you can try it there...... you should go to: /usr/lib/pymodules/python2.7/ubuntu-sso-client/ubuntu_sso$ ........ and then do: sudo find . -iname '*.pyc' -delete....... and try the problem between sso and control panel again, and should be fixed..... something went wrong between the update and the pycs [15:48] it seems === gatox_lunch is now known as gatox [15:48] ohhhh i was gatox_lunch [15:50] gatox, dude there is a guy with a nick very similar with lunch at the end, was it you? [15:51] mandel, no!! who who?? [15:51] jje [16:00] briancurtin3: http://pypi.python.org/pypi/watchdog [16:00] mandel, et al: what if we used watchdog for all platforms? [16:00] it abstracts away inotify, FSEvents, and Windows stuff. [16:04] urbanape, hm..it will add an abstraction layer for all which is good, but I'm sure they have not fixed the fact that moves are diff etc.. although, you can use it for mac, and then we can consider it for windows [16:04] urbanape, I think that adding a lib in there is a smart move.. the problem is how well tested it is, does it solve all the issues and how do we hook that to twisted [16:05] urbanape, so in summary, if you have the time is ok for me but you need facundobatista to approve such a major change [16:05] just found it looking to see if anyone had already mapped inotify events to FSEvents [16:07] urbanape, if you a library that abstracts them all into one set of semantics, *maybe* SD can be tuned to use those semantics, and we'd be happy [16:07] facundobatista, would mean loosing the open event for sure [16:08] facundobatista, urbanape it will also mean pushing all the dirty code to a lib maintained by others \o/ [16:09] mandel, which is double edged sword [16:09] facundobatista, true. [16:11] it's got a bunch of dependencies on it, too, which we might not want to take on [16:11] PyYAML, argh, argparse, pathtools, and Brownie [16:12] might not actually need those (just for the extra scripting support) [16:18] ok, need to get lunch. bbiab === briancurtin2 is now known as briancurtin [16:23] urbanape: sry, comcast seems extra flaky today [16:24] no problem [16:28] ralsina, nessita I have an appointment in the ophthalmologist, so I'm leaving in 5 mins to be there on time [16:32] ok, time to go and see if i'm more blind :P [16:36] mandel: ack! [16:59] nessita, ralsina, can you take a look at bug #940884 ? [16:59] https://bugs.launchpad.net/ubuntuone-client/+bug/940884 [17:00] we will not sync files with names longer than what the os allows. Is that our problem, or the os problem? I think you'll know better. [17:01] elopio: it's the OS fault but it's our problem [17:04] ralsina: ok, you have it on your plate. [17:05] good evening u1 team! I am now working, so hit me with reviews ;-) [17:05] ralsina: about to! [17:05] guh, when I fix this lint issues [17:09] ralsina: would you know if rtgrant filed the bug? [17:10] ralsina, nessita really trivial review: https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/back-button/+merge/95622 [17:10] gatox: looooking [17:11] nessita: seems he didn't. I'll do it myself [17:11] ralsina: :-/ [17:13] nessita: I pinged (with the request) but he never ponged, maybe he's not on IRC today, andit's late in the UK already, so... [17:13] ralsina: if you consider this can wait till Monday, we can email him asking him to please file it [17:14] I guess it can, since we are not going to file the UIFe today.Let's mail him and he can do it early monday [17:15] nessita: I'll send the mail right away [17:18] nessita: mailed, with you as CC [17:19] gatox: The "enter verification" page is the one for the token you get by email for new accounts? [17:19] ralsina, yes [17:20] gatox: why shouldn't there be a back button there? [17:20] ralsina, let me give you the bug description [17:20] And the right way to not have a back button is to make the previous page a "commitPage" IIRC [17:20] ralsina, https://bugs.launchpad.net/ubuntu-sso-client/+bug/944769 [17:20] gatox: looking [17:21] what happens if you registered with a wrong email? Then you want to go back... [17:21] nessita, ^ [17:25] ralsina: well, the user can close the window and re-open, no? Or, we can leave the button there but cleanup the whole fomr [17:26] ralsina: otherwise we'll let the user think he can "retry" the registration [17:26] which he can not [17:26] ralsina: I prefer not having a back button, the user already had to enter the email *twice* [17:30] ralsina: the review you were asking for is https://code.launchpad.net/~nataliabidart/ubuntuone-control-panel/fix-933576/+merge/95627 [17:30] gatox: if you could too... ^ [17:30] nessita, on it [17:31] so.... what should i do with the back button isssue..... i'm working in another issue now...... to know if i need to modify anything there [17:31] ralsina: any thoughts re: Back button? my food is ready so you can share those with me when I came back [17:31] gatox: mark the MP as WIP until ralsina and I decide what is best [17:32] nessita, ack [17:38] nessita: https://bugs.launchpad.net/ubuntuone-installer/+bug/944982 from Robert [17:38] nessita: buen provecho! [17:39] nessita: "removing the dbus service" means bin/ubuntuone-control-panel-backend also? are we certain no apps in ubuntu are using it? [17:44] nessita: deja-dup uses the dbus service in control panel [17:48] nessita: on your branch, if I login using an existing account, the SSO window is empty and u1cp is not usable until I close it manually [17:52] dobey: why would they? is not an official API :-/ [17:52] ralsina: do you have nightlies updated? [17:52] nessita: updated to last night. let me update and retry [17:52] ralsina: (remember that SSO code is independent from the control panel code) [17:53] nessita: yes, I was thinking if I had to file a bug on SSO and/or if you knew about it [17:53] ralsina: ah, ok. LEt's try to reproduce with latest nightlies [17:53] (please) [17:54] ralsina: with sso window empty, what do you mean? [17:54] nessita: because it's probably the only way to do it. and any published api is an official api. [17:55] nessita: if you know of a better way to do that, feel free to help mterry switch deja-dup over to it :) [17:55] dobey: I disagree with your last sentence, but will not discuss about that. Ok, if you are sure deja-dup uses it, I guess we need to leave that in place :-/ [17:56] nessita: all widgets disappear and the window stays open [17:56] well i just asked mterry in #ubuntu-desktop, because it is depending on ubuntuone-control-panel [17:56] ralsina: weird :-/. Would you let me know if this happens with latest sso? [17:56] nessita: yep [17:58] dobey: well, sadly, there is no better way to do the quota querying without doing all the REST thingy by hand. SO I guess we cna leave that (the controlpanel dbus service) in place. [17:58] right [18:08] nessita, ralsina another really small branch: https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/ensure-overlay/+merge/95636 [18:08] ralsina: what shall we do regarding the back button? saw my comments from before? [18:08] gatox: did you check all the screens? [18:09] nessita, yes [18:09] awesome :-) [18:09] nessita: let's remove it and prepare a request for UX [18:09] nessita: so that we can make it better in the not-so-close-future [18:09] ralsina: ack, gatox, did you get that ^? (also not sure if you should apply what ralsina said about the CommitPage) [18:10] nessita, yes..... i'll check about CommitPage [18:10] gatox: thanks! [18:26] gatox: I haven't tested IRL yet, but... in https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/ensure-overlay/+merge/95636, shouldn't the overlay be hide when the operation is completed? ie, in the method on_email_verified y en on_email_verification_error [18:27] ....... yep....... [18:27] gatox: I meant on_email_validated and on_email_validation_error [18:28] gatox: would you please add that? [18:28] nessita, of course.... [18:28] thanks! [18:34] ralsina: did it work? [18:34] nessita: still upgrading [18:35] ack [18:35] nessita: day of lots of updates, friday [18:42] meh [18:42] looks like rain [18:42] dobey: in a wedding day? [18:43] weddings are lame anyway [18:44] Oh. My attempt at alanis-rolling failed :-( [18:44] ralsina: I got it :-) [18:44] is like a free ride, when you already paid ;-) [18:44] it's a good advice! [18:44] and i really do not like debugging untraceable race conditions that corrupt memory :-/ [18:45] dobey: is the RB thingy? :-/ [18:45] dobey: ugh [18:45] yes [18:46] i am pretty certain now that it's not the entry that i'm creating in the plug-in, which is being dereferenced. [18:46] dobey: if you had no luck with valgrind, have you tried any of the debuggig mallocs? [18:47] well i'm just doing gdb [18:47] and using G_SLICE=always-malloc makes the crash go away [18:47]