[00:16] anyone around? :D [00:19] apachelogger: ok, so that parser is complete, just one problem, DTSTART and DEND don't work for some reason [00:19] plz fix [00:43] meh [00:44] apachelogger: QDateTime issue fixed as well [00:44] yer parser is ready [00:45] night guys [00:57] * afiestas is close to tarball // tag 1.2.1 [00:57] together with a patch for libbluedevil (one line patch) [07:16] shadeslayer: http://vladimirputin.soup.io/post/154482096/Image [07:26] booooooo Quintasan! [10:20] shadeslayer: woohoo, if you give me access I can improve it a bit :P [10:20] * apachelogger only got up so did not yet look at it in detail [10:25] morning [10:52] apachelogger: you already have [10:52] cool [10:52] shadeslayer: btw Q_ASSERT(iCalFile->open(QIODevice::ReadOnly | QIODevice::Text)); I'd not call the thing a library with something like that :P [10:52] apachelogger: there's one thing that needs doing, handling \n characters at the end of a descriptio [10:53] +n [10:53] huh? [10:53] apachelogger: huh? why not? ( i'm relatively new to library development ) [10:53] shadeslayer: because a library must never ever go down [10:53] well [10:53] ah [10:53] unless there is an issue that it can definitely not prevent or recover from [10:54] I mean, for internal libs it makes no difference but for public shared ones it does [10:54] bah, that \n is for vjournal [10:55] so nvm that :P [10:55] *shrug* [10:55] even so [10:55] readline would simply come up with an empty line [10:55] yeah [10:55] shadeslayer: another thing: it is very uncommon to do all the work in the ctor [10:56] shadeslayer: one would provide a static function for that most of the time [10:56] ok, we could just move stuff into a static function [10:56] right [10:56] Parser::parse(File) [10:56] yeah [10:56] plus we need that anyway since I also need a ctor for QBA [10:57] QBA? [10:57] only the on-disk cache is read from file, incoming updates form remote are processed from the networkreply directly [10:57] shadeslayer: qbytearray [10:57] ah ok [10:57] ah kewl [10:57] Quintasan: hahah :D [10:58] shadeslayer: so, two options... a) one ctor and multiple parse(QFile), parse(QBA) ... b) multiple ctor and one parse [10:58] former seems bettar [10:58] oh [10:58] Quintasan: ping [10:58] apachelogger: pong [10:58] apachelogger: btw how fast is this yer device [10:59] shadeslayer: what device? [10:59] apachelogger: N950 [10:59] what kind of fast? [10:59] how fast does it process the ical file [10:59] because it was insanely fast on my machine [10:59] but since we need it to be fast on a ARM device ... [11:00] parsing is not so much the problem... [11:00] problem is that currently I have multiple lists which in JS means multiple memcopies [11:01] IIRC the main oneiric ical had something like 220 sessions [11:01] so that means like 3*220 memcopies or so [11:01] and memcopies are evil! [11:01] heh [11:02] even so it only hung for like 1 second [11:02] neat [11:02] so for ondisk cache reads actually starting the app is slower than parsing [11:02] and for remote updates the actual file transfer was slower than parsing [11:03] apachelogger: btw is the QBA going to contain the entire iCal starting with BEGIN: VEVENT ... or just one particular field like DESCRIPTION [11:03] O_O [11:03] shadeslayer: the ical as read from UDS [11:03] ah [11:03] yeah [11:03] your parser will trip over that [11:04] ical files have a sort of manifest section before the first event [11:05] apachelogger: i ignore that :P [11:06] oh yeah [11:06] sorry misread the code there [11:06] although i shouldn't, but seeing how i couldn't figure out what i should do with that, i left them [11:06] bad code... [11:06] I shall refactor :P [11:06] feel free too :P [11:06] shadeslayer: you should const the read line IMHO [11:06] more compiler love + no changes by accident [11:06] hmm ... yeah, we don't want to change the values [11:07] apachelogger: are you refactoring the entire thing right now? [11:07] sorta [11:08] alright i'll wait for you to push yer changes [11:09] shadeslayer: another thing ...the event class is likely to be derived, whereas the parser is less likely to [11:09] shadeslayer: Any news about the touchpad or I missed it? [11:09] so former needs a virtual dtor and latter does not [11:09] Quintasan: nothing so far [11:10] Quintasan: they're going to take some time and i'm pretty sure it won't be available before another month [11:10] and as soon as they make it available its going to be gone [11:10] they've said they can't even gurantee pricing [11:11] wtf dbus is taking all my cores [11:11] shadeslayer: some thought on formartting btw ... functions always before members in the header [11:12] i'll keep that in mind :) [11:13] apachelogger: does the event even require a dtor? [11:13] i'm not allocating memory manually anywhere in the event , which is why i didn't have a dtor there [11:14] shadeslayer: a library class *always* needs a dtor :P [11:14] even if its empty? :P [11:14] the compiler has free hand at where he puts the dtor and what it makes of it [11:15] if you do not declare a dtor the compiler might make a virtual one because one of your classes derives the class internally [11:15] suddenly stuff starts segfaulting for no apparent reason [11:15] heh [11:18] apachelogger: we could make it a bit more faster by using continue in parseICalBlock [11:19] as soon as it finds the field, it sets stuff in the event and then proceeds to the next line instead of checking it again [11:19] yes [11:22] shadeslayer: you should fix your copyright line btw :P [11:23] where [11:24] everywhere, year misisng [11:24] ah [11:27] that loop needs help [11:27] done [11:27] } else if (line.startsWith(QLatin1String("DTSTART:"))){ [11:27] event->setStartDate(QDateTime::fromString(line, "'DTSTART:'yyyyMMdd'T'hhmmss'Z'")); [11:27] formatting fail :P [11:27] yes? what about it? [11:27] get a proper editor [11:27] i'm using qtcreator [11:27] :P [11:27] learn to use it then [11:28] heh :P [11:29] so [11:29] that loop needs to go away [11:30] 1 if that checks for dtstart, dtend and categories, any other value that is end:vevent is then parsed as key:value and insertered into the property map [11:30] that said [11:30] instead of having hard coded properties the event should probably use a property map [11:30] so it is easily enhancable [11:37] you broke something [11:37] i can't parser anymore [11:37] s/parser/parse [11:37] [~/projects/libqcalparser/build]$ ./qcalparser /tmp/uds-p.ical [11:37] [~/projects/libqcalparser/build]$ [11:38] no [11:38] your library was badly designed [11:38] QCalParser *parser = new QCalParser(file); [11:38] that now passes for QCalParser(QObject*) [11:38] ah ok [11:39] so you need to change that to construct and then parse [11:39] shadeslayer: how do we do property management? [11:39] apachelogger: QVariantMaps ? [11:39] 2 options: dynamic property injection via QtMetaObject OR QVariantMap [11:39] because QML has better support for QVariants than other Qt classes i hear [11:39] former is again bettar [11:39] shadeslayer: qml also expects properties [11:40] so if we use a QVM we need to manually declare the properties in the header still [11:40] actually [11:40] we need to declare them and write getters [11:40] so dynamic injection via the metaobject might in fact be the only sane solution [11:41] * apachelogger tests this for a bit [11:41] * shadeslayer tries to understand what apachelogger just said [11:41] lol [11:41] I'll show you code :P [11:41] that would actually be better :P [11:43] shadeslayer: http://paste.kde.org/123079/ [11:43] the qtmetaobject of the qobject already has a QVM for the properties [11:43] so you can inject new properties by simply setting a new property [11:43] without having to declare ANYTHING [11:44] O_O [11:44] sweet [11:44] every time I tell people about this a jahava kitten dies [11:44] hahaha [11:45] apachelogger: so, we just inject new properties in our QCalEvent [11:46] when the setters are called [11:46] and change getters to return that property [11:48] if it were working :O [11:49] its not ? :O [11:49] not in qml [11:49] aw :( [11:52] * apachelogger tries to debug [11:59] Note that dynamic properties are added on a per instance basis, i.e., they are added to QObject, not QMetaObject. [11:59] great [12:00] shadeslayer: so we can use the dynamic properties, but we still need to provide Q_PROPERTY macros and reader functions [12:00] (latter can be put into a macro too though [12:00] ) [12:01] hmm [12:01] or..... [12:01] we could write our own metaobject which allows for this use case ^^ [12:03] actually [12:03] shadeslayer: we could just not use properties [12:04] Q_INVOKABLE QVariant getValue(QString key) [12:04] actually [12:04] scratch the get [12:04] * apachelogger has used too much crappy qtmobility api [12:04] heh [12:04] apachelogger: and we could just reuse the current API to deal with that [12:05] deal with what? [12:05] shadeslayer: we'd still setProperty() [12:05] Q_INVOKABLE QVariant getValue(QString key) [12:05] oh [12:05] only that we provide a Q_INVOKABLE which gives access [12:05] ah [12:05] i get it [12:05] Q_INVOKABLE QVariant value(QString key) { return getProperty(key); } [12:05] * apachelogger did it again [12:06] heh [12:07] did i mention KCombBox is weird [12:07] because of the K [12:07] probably [12:07] shadeslayer: well, make a choice on the event matter... either is fine with me [12:08] actually having Q_PROPERTY hardcoded props allows for easier use in qml as qtcreator will autocomplete ^^ [12:08] for ultimate scalability one will need a value() function anyway [12:08] ok, lets go with Q_PROPERTY then [12:12] k [12:12] shadeslayer: btw QDateTime eventStartDate(); [12:12] yes? [12:12] that is a C func name, not a cpp one :P [12:13] QDateTime startDate() [12:13] how? [12:13] we know that it is a event because it is an Event class :P [12:13] event->eventStartDate()? [12:13] vs [12:13] ah heh [12:13] event->startDate() [12:13] that reminds me [12:13] shadeslayer: scope is broken there [12:13] the event is held by the parser [12:13] i think everything is currently using eventFoo [12:14] yeah, well, fixing [12:14] heh :P [12:14] Q_PROPERTY(QDateTime dtStart READ dtStart) [12:14] QDateTime dtStart() const { return getProperty("DTSTART"); } [12:14] hm [12:15] something is not cool about this [12:15] that does not look right [12:16] looks weird tbh [12:16] but why? [12:16] how does it know to return the dtStart property [12:17] when we call it with getProperty("DTSTART") [12:22] qvariant? [12:23] whut [12:35] doesnt work [12:35] no idea why [12:35] it works for a qstring but not for a datetime [12:35] http://paste.kde.org/123109/ [12:37] this sucks [12:37] shadeslayer: so either we make our own QVM OR we create our own metaobject which allows this stuff OR we simply do not use Q_PROPERTY [12:38] actually, this is a bit weird [12:38] you could return a datetime string [12:38] no, I need actual date objects in qml [12:39] QML can parse thos? [12:39] *those [12:39] slow [12:39] for no good reason [12:39] heh [12:43] * apachelogger tries with own QVM [12:45] i'll bbl [12:46] oh [12:46] I know what is wrong [12:47] * apachelogger is a bad maco coder [12:47] worky worky \o/ [12:55] fabo: Check_for_buffer_overflow_in_Lookup_MarkMarkPos_CVE-2011-3193.patch is already part of qt 4.7.4 [12:56] ** RESERVED ** This candidate has been reserved by an organization or individual that will use it when announcing a new security problem. When the candidate has been publicized, the details for this candidate will be provided. (http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2011-3193) [13:45] apachelogger: new API done? [13:45] and/or want my help? [13:46] WIP [13:46] rewriting parseBlock righ tnow [13:46] QString is a right PITA [13:46] what s the problem with parseBlock [13:47] or are you changing it to use the new API [13:49] new api [13:49] http://paste.kde.org/123121/ [13:50] dude [13:50] that is wrong [13:51] event->setProperty(key, QDateTime::fromString(value, "'DTSTART:'yyyyMMdd'T'hhmmss'Z'")); [13:51] what about poor DTEND [13:51] your forgot about him [13:52] otherwise ... this new API will handle other keys as well, awesome [13:53] shadeslayer: take a closer look at the if [13:54] i did [13:54] apachelogger: how would it work for DTEND [13:54] ah [13:54] :) [13:54] now I see [13:54] your code was rubbish :P [13:54] but it worked :P [13:54] event->setProperty(key, QDateTime::fromString(value, "yyyyMMdd'T'hhmmss'Z'")); [13:54] ah [13:54] no [13:55] wont work [13:55] :D [13:55] or maybe [13:55] shadeslayer: why not? [13:55] cuz you don't ignore DTSTART ... uh ... maybe it will [13:55] not sure [13:55] lol [13:55] need to look at docs again [13:57] ah ok, i understand now [13:57] apachelogger: value is just the part after DTSTART: right? [13:57] not the entire line itself [14:00] yes [14:00] hence there is a key and a vlue [14:00] *value [14:01] right, didn't read closely [14:01] * shadeslayer should probably blog about the stuff he did over the weekend [14:33] * apachelogger thinks he is happy with the event now [14:36] shadeslayer: pushd [14:36] please have a look [14:38] looking [14:51] hi kubuntu devs. gtk2-engines-oxygen doesn't work for all gtk apps in oneiric. please fix it :) [14:51] apachelogger: looks good, but, you don't need #include in qcalparser.cpp [14:51] yeah, noticed that [14:51] nikola: got a patch? [14:51] more worrying is that it does not work [14:51] bugreport: https://bugs.launchpad.net/ubuntu/+source/gtk2-engines-oxygen/+bug/853264 [14:51] Ubuntu bug 853264 in gtk2-engines-oxygen (Ubuntu) "gtk2-engines-oxygen doesn't work for all gtk apps " [Undecided,New] [14:51] shadeslayer: no sorry [14:52] apachelogger: it doesn't work? [14:52] I get an undefined qlist in qml [14:53] maybe we need to export our QCalEvent as a property or sth? [14:53] so that QML recognises it [14:53] even when I just use a qobject [14:54] file:///home/me/src/git/uds-build-desktop-Qt_4_7_4__System__Release/src/qml/ubuntudevelopersummit/main.qml:65: TypeError: Result of expression 'model[0]' [undefined] is not an object. [14:54] wtf [14:56] 0.o [14:56] "A list of QObject* values can also be used as a model. A QList provides the properties of the objects in the list as roles." [14:57] possibly different versions of documentation ? [14:57] :P [14:57] I have been using qobj* list before [14:57] oh [14:58] most confusing [14:58] QVariant(QList) [14:58] intersting [15:00] apachelogger: btw instead of using a deliminatorPosition we could also use QString::section [15:02] point? [15:02] no point, i'm just saying :) [15:03] I mean, section() has to seek the string to find the seperator and consequently a section [15:03] and, shouldn't that be a QLatin1Char [15:03] so since we need lhs and rhs it makes more sense to find the seperator position within the string [15:04] and then take everything left and right of that [15:04] (since we know the exact position of seperator qstring will not have to seek for the seperator anymore) [15:04] right [15:06] so [15:06] I think qml is just bugged again [15:06] huh [15:07] apachelogger: use a C++ backend to set data in the QML app? [15:07] I am emitting it [15:07] using setContextProperty or whatever [15:07] and apparently qml it to stupid to convert to declarative type in a handler [15:07] shadeslayer: setcontextproperty? what is this? kindergarten? [15:07] Oo [15:08] you do not set stuff like that as a context property [15:08] a KDE i18n handler you could set ther [15:08] heh :P [15:19] shadeslayer: I think I know what to do [15:19] but first I need a break [15:19] alright [15:19] i'm off to dinner myself [15:20] apachelogger: oh btw, one thing that we can extract form the ical file is the time zone [15:21] there's a field for that in the [15:21] +beginning [15:21] X-WR-TIMEZONE:UTC === yofel_ is now known as yofel [15:52] apachelogger: upload digikam already [15:54] debfx: There were two security issues that fabo was working on. One is already fixed in 4.7.4, the other is not. [16:20] apachelogger: see email if you're around [16:21] why [16:21] I am at war [16:21] apachelogger: just do it [16:22] what mail [16:22] Your Amazon.co.uk order has dispatched [16:22] yay [16:22] shadeslayer: how did you know that? :O [16:22] lol [16:22] not that [16:23] apachelogger: see the one from wordpress [16:23] * shadeslayer needs to start prepping for his US visa [16:23] ah [16:24] that must wait [16:24] because I need booze [16:24] * shadeslayer switches apachelogger's booze with coffee [16:25] STUPID qdeclarativelistproperty [16:27] cant use qsharedpointers [16:27] terrible [16:29] apachelogger: That's why I find school absolutely exasparating. I just learned we have a test on thermodynamics tomorrow [16:29] I mean what the f... [16:29] I do not even know what that is [16:29] lol [16:29] * apachelogger wonders if one could store it in a sharedpointer so he does not need to know about it [16:30] Not really [16:30] dang me running [16:30] It's basically useless info so I am not entirely sure why would you want to store it anywhere [16:30] Quintasan: up for reviewing me blog post? [16:30] shadeslayer: Send it this way, might take a break as well [16:31] shadeslayer: so.... I think I will have to do major refcktoring of the uds app to accomodate qcal [16:31] Quintasan: alright [16:31] particularly to add threading into the picture [16:31] oh it makes me sweat already [16:31] Quintasan: done [16:31] apachelogger: threading in QML ? [16:31] :O [16:31] yes [16:31] * Quintasan goes reading [16:31] workerscript [16:31] but no [16:31] are you shitting me [16:31] in cpp [16:31] ah ok [16:31] that makes sense [16:32] the data retrieval and storage becomes thread0red [16:32] threading in QML would probably make my mind implode [16:32] Quintasan: so, actually thermodynamics is a very important thing [16:33] if you have too fast intercourse you can start a fire, and no one wants that [16:33] so I consider it in everyones interest that everyone knows about this and consequently does not cause accidential fires ever night [16:33] Quintasan: also, review my digikam packaging at : https://code.launchpad.net/~rohangarg/kubuntu-packaging/digikam : if its alright, i'll upload to kubuntu-packaging and then someone can upload it [16:33] it aint even merged [16:33] good ld [16:34] howwas I supposed to do stuff [16:34] G [16:34] apachelogger: you were supposed to review it, give me the OK and then i would i have pushed to kubuntu packagers :P [16:34] shadeslayer: you really want a different background for the presence screenshot [16:34] shadeslayer: did you do a merge proposal? [16:34]