[01:10] nixternal: before that were 50 or so commits in a branch :P [01:10] * PyObjectPtr also did another 30 since then in another branch [01:10] * PyObjectPtr is implementing the u1 api [01:10] * PyObjectPtr didnt even know u1 had an api until a few hours ago ^^ [01:26] PyObjectPtr: quit changing your name === PyObjectPtr is now known as static_name [01:27] nixternal: happy now? === AndChat- is now known as jjesse-droid [01:29] hm [01:29] seems my getQuota() is already working [01:30] * static_name thinks a caching system would be nice to have with such an api === static_name is now known as apachelogger [01:37] apachelogger->goTo(Bed); [01:46] g'nite apachelogger [01:46] what's up gang? [02:04] seeing how many canucks I can piss off tonight [02:04] http://www.cbc.ca/canada/montreal/story/2010/05/14/quebec-cycling-accident.html [02:04] that story combined with the comments prove they are just as stupid as they claim we are [04:34] * JontheEchidna back home [04:34] * JontheEchidna goes off to bed === fale_ is now known as fale [08:45] JontheEchidna: yes and no, while one could upstream that patch in a more appropriate version (i.e. if(NOT MYSQLD_EXECUTABLE) find_program etc.) which of course is of no concern to debian, the patch tries to archive something different, it essentially eliminates the need to search for mysqld and thus makes the package not build-dep on mysqld [08:45] however [08:46] we do not use mysqld but **COPY** it from the mysql-server package to akonadi as mysqld-akonadi [08:46] and, since debian in all its glory hardcoded the name of the daemon we just override that using debian/rules [08:47] otherwise we would have to patch the patch or change the patch which both are recepies for troubles and lost-in-merge-issues [09:45] Riddell: around? [10:07] Internal Server Error \o/ [10:07] * apachelogger broke u1 [10:11] Riddell: rekonq is translated [10:12] it’s in the kde extragear trunk for translation [10:21] http://flavio.castelli.name/qjson-qobject-serialization-deserialization now that puts implementing a json api into a whole new perspective [10:46] hm [10:46] ok, this is pretty cool [10:47] Quota q; [10:47] QJson::QObjectHelper::qvariant2qobject(data.toMap(), &q); [10:47] kubotu: identica dent <3 !qjson - it is uberawesome [10:47] status updated [11:36] \o [12:35] http://imagebin.ca/view/HM9fpP.html [13:26] apachelogger: fck yeah [13:26] but I use dropbox :P [13:26] holy shifting paradigm [13:26] ! [13:26] * apachelogger goes mad over implementing the api [13:31] /usr/include/qt4/QtCore/qobject.h: In member function ‘UbuntuOne::Api::SubscriptionObject& UbuntuOne::Api::SubscriptionObject::operator=(const UbuntuOne::Api::SubscriptionObject&)’: [13:31] /usr/include/qt4/QtCore/qobject.h:309: error: ‘QObject& QObject::operator=(const QObject&)’ is private [13:32] * apachelogger is wondering why moc cant just use the copy constructor -.- [13:39] apachelogger: QObject prohibits copying [13:40] yeah, but I am not feeling like implementing an assignment constructor and for some reason pointers do not work [13:40] apachelogger: well, I guess Qt has a reason for prohibiting copying... I think the whole object hierarchy thingy would end up in a burning mess if we started copying QObjects around [13:41] so fix the reason why pointers don't work for you instead :P [13:42] qt is privativing = and copy ctor so that one does not end up with shallow copies by accident ;) [13:46] ther is some qjson magic I do not see, but probably it has to do with QVariants vs. custom type [13:47] it appears as if the metacall for the variant in question never gets issued :( [14:26] ah! [14:26] so [14:26] the way qjson applies a json object to a qobject [14:26] looks like apachelogger rediscovered the wheel [14:26] :P [14:26] a) it converts the json object to a qvariant [14:27] b) it runs through all of target_qobject's properties and tries to use setPropert() using the appropriate part of the qvariant [14:27] problem: setProperty's prototype is bool QObject::setProperty ( const char * name, const QVariant & value ) [14:28] that means that above approach only works for data types supported by QVariant [14:29] so the only approach to marshal an object within an object is to first marshal the parent, then marshal the child and then link them together [14:29] suffice to say I already feel the headache on the run [14:40] apachelogger: QVariant can support any datatype [14:40] that's what qRegisterMetatype is for [14:41] you need to provide serializers for the type [14:51] hehe. the Federal Convention of the German Pirate Party is world trending topic in twitter. Take that, Mark! :) [14:52] *worldwide trending topic even [15:10] Sput: that sounds too evil for the mental condition I am in right now ;) [15:13] anyhow [15:14] ohh [15:14] * apachelogger will have to get a new brain if his sister continues to sing like that [15:15] oh my, she cant even hold a note -.- [15:17] apachelogger: defining an operator>>() for QDataStream isn't hard :) [15:17] * apachelogger is doing marshaling between qt and dbus datatypes, so he knows :P [15:17] "To use the type T in QVariant, using Q_DECLARE_METATYPE() is sufficient. " [15:18] * apachelogger actually had done that [15:18] well, this defenitely needs some looking into [15:19] but for now, marshaling the json to the child-objects manually should be sufficient [15:19] m_plan->setText(account.subscription()->description()); is so much better than m_plan->setText(account.subscription()["description"].toString()); [15:24] apachelogger: hmm, quite sure that you also need the stream operators [15:24] Q_DECLARE_METATYPE_STREAM_OPERATORS() [15:24] or something [15:25] Note that if you intend to use the type in queued signal and slot connections or in QObject's property system, you also have to call qRegisterMetaType() since the names are resolved at runtime. [15:25] ah, yeah, you need that obviously for the marshalling [15:25] *nod* [15:25] right, for queued signal/slots QVariant needs to do actual serialization [15:26] for direct calls, it doesn't need to bother, so no need to register it in the dynamic type system [15:26] and it's qRegisterMetaTypeStreamOperators() :) [15:37] well then [15:37] Sput: there comes the need for copy constructors [15:37] apachelogger: no? [15:38] I don't see where you'd need to copy objects [15:38] gcc wants to eat me once I added the metatype stuff [15:38] YDIW then [15:38] you could show a buildlog :P [15:40] Sput: http://paste.ubuntu.com/433873/ [15:41] ah [15:41] Sput: This macro makes the type Type known to QMetaType as long as it provides a public default constructor, a public copy constructor and a public destructor. [15:41] from the documentatoin of Q_DECLARE_METATYPE [15:41] so obviously I did not do it wrong :P [15:46] apachelogger: very weird [15:48] Sput: maybe it is there so that people who dont know how to implement a cctor wont mess with the metatypes ;) [15:48] i.e. people like Quintasan [15:51] well [15:51] something I do wrong [15:51] I have a class Account, that has a member of type Subscription and that has a member of type SubscriptionUpgrade [15:51] Subscription and SubscriptionUpgrade both have SubscriptionObject as base, and that is a QObject [15:53] now if I try to let qjson do magic on an instance of Account automoc will cause failure because ... if the Subscription instance is not a pointer, it will want to have an assignment operator overload [15:53] only god knows why [15:53] if I make it a pointer the qjson magic fails for whatever reason there might be (or maybe the Qt property stuff does...) [15:53] this is all very meh. [16:03] Neat @ kde bug 227127 [16:03] KDE bug 227127 in general "Implement an "undo close-tab" feature" [Wishlist,Resolved: fixed] http://bugs.kde.org/show_bug.cgi?id=227127 [16:03] I'm sold for rekonq ;) === jtechidna changed the topic of #kubuntu-devel to: Welcome to the Kubuntu Development channel | Let's all be mavericks | Congratulations to neversfelde for making MOTU \o \o o/ o/ \o/ \o/ === jtechidna changed the topic of #kubuntu-devel to: Welcome to #kubuntu-devel | Let's all be mavericks | Congratulations to neversfelde for making MOTU \o \o o/ o/ \o/ \o/ [16:14] hi all [16:15] There is a patch for KDE since Karmic which allows to change brightness of lcds. Does anyone know where to find it? It doesn't seem to be available upstream. === nigelbabu is now known as nigelb [16:20] I mean this fix: https://bugs.launchpad.net/ubuntu/+source/kubuntu-default-settings/+bug/293213/comments/21 [16:20] Launchpad bug 293213 in qt4-x11 "Qt and kde4libs don't support various multimedia keys" [Medium,Fix released] === yofel_ is now known as yofel [17:36] apachelogger: hey, who the hell can't implement a ctor? [17:39] * Quintasan goes out for some sort of event [17:42] Quintasan: you [17:42] :< [17:43] void Horse::Horse() { std::cout << "I'm a horse, I can't help you" << std::endl; } [17:43] who cares if it's useless? :P [17:47] see, what troubles me is that void right there :P [17:47] doesnt get any more pointless than that [17:47] [18:43] who cares if it's useless? :P [17:47] :P [17:49] well then [17:49] I was talking about cctors [17:50] and that is no cctor :P [17:53] api is at 743 sloc -.- [17:59] apachelogger: you pinged? [18:00] Riddell: I forgot what it was about ... something about Qt's type system probably :) [18:00] Riddell: btw, authing works for the better part, and I am currently implementing the ubunuone REST api (necessary for KCM) [18:01] apachelogger: careful or you'll have it all implemented before SoC begins :) [18:01] ^^ [18:02] * apachelogger wants the UI stuff out of the way so he can dive into akonadi :) [18:36] Riddell: I meant to bring this up at UDS, but I forgot. Can we merge virtuoso this cycle? [18:36] Debian has a dedicated maintainer now, so it shouldn't be a great burden on our part === JontheEchidna changed the topic of #kubuntu-devel to: Welcome to #kubuntu-devel | KDE merges: https://wiki.kubuntu.org/Kubuntu/MaverickKDEMerges | Congratulations to neversfelde for making MOTU \o \o o/ o/ \o/ \o/ [18:44] Patch review page needs set up too [18:47] JontheEchidna: patch review? [19:04] nigelb: Like this, except for maverick: https://wiki.kubuntu.org/Kubuntu/LucidPatchReview [19:08] JontheEchidna: we should merge everything unlesss there's a good reason not to [19:09] Riddell: Ok. I was wondering because of the hoops we needed to jump through for getting the MIR approved, but I suppose that keeping a separate package from Debian would be less secure since we'd have to maintain it ourselves [19:10] Hah, Roman really does look different with long hair [19:11] JontheEchidna: nice page :) [19:15] nigelb: It was really helpful last cycle. :) [19:16] Hmm, my aunt seems anxious for Belgium pictures on Facebook [19:16] I'd better get right on that [19:21] JontheEchidna: um, belgium, pictures, I'm anxious too! [19:21] ;) [20:23] hrm, pkg-kde-tools needs a newer dpkg [21:33] nigelb: http://www.facebook.com/album.php?aid=213233 [21:33] (Dunno if people without FB accounts can see it though) [21:40] JontheEchidna: Even with a FB account I can't see it [21:42] ryanakca: I should add you as a friend then [21:43] hmm, there are like 5 Ryan Kavanaghs [21:56] JontheEchidna: can add me as friend too? [21:58] nigelb: I can't find you on facebook [21:58] unless your profile picture is of the pope [21:58] JontheEchidna: no [21:59] did you try 'nigel babu'? [21:59] ...or if you're a woman from south africa [21:59] yes [22:00] I get this searching for "nigel babu": http://simplest-image-hosting.net/i0-plasma-desktoptw1449-jpg.jpg [22:00] oh, my fault [22:00] account privacy settings [22:00] ah, right [22:01] added you :) [22:01] confirmed [22:02] awesome, thanks [22:09] * nigelb comments furiously. [22:09] I hope JontheEchidna doesn't regret adding me :D [22:09] Naw. ;) [22:42] JontheEchidna: You could try making the album visible to everyone? (If that's possible. I don't use FB all that much, so I'm guessing) [22:43] Ah, nevermind, it is visible to all