[11:20] <apachelogger> so
[11:20] <apachelogger> question of the day
[11:20] <apachelogger> should apachelogger upgrade to one-eye-rick?
[11:22] <apachelogger> oh, we are oversized again
[11:30] <bulldog98_> apachelogger: yes
[11:30] <apachelogger> what if it breaks?
[11:30] <bulldog98_> apachelogger: fix it :P
[11:30]  * apachelogger is too stupid
[11:32] <bulldog98_> apachelogger: you don’t seem to be stupid
[11:33]  * apachelogger almost toasted his router yesterday with openwrt :S
[11:34] <bulldog98_> apachelogger: we are more stable than openwrt :)
[11:34] <apachelogger> yeah
[11:35] <bulldog98_> at least I hope so
[11:35] <apachelogger> until ubuntu-x starts uploading stuff again
[11:36] <bulldog98_> apachelogger: do apt-pinning for x :)
[11:38] <bulldog98_> apachelogger: and who cares about x use wayland instead :P
[11:38] <apachelogger> G
[11:57] <bulldog98_> apachelogger: can the icecc in the pbuilder access a scheduler running on the computer running the pbuilder?
[11:57] <apachelogger> sure it can
[11:57] <tsimpson> it communicates over the network, so sure
[12:04] <bulldog98_> apachelogger: then why is says my pbuilder icecc that it’s gone build local?
[12:05] <apachelogger> cause you aint using the hook properly
[12:06] <bulldog98_> apachelogger: so what is needed? the ICECC_VERSION and my PATH are set
[12:06] <apachelogger> read the readme
[12:08] <bulldog98_> apachelogger: I did all the stuff that’s in the README (the first two steps, the others aren’t possible anymore)
[12:29] <bulldog98_> how do I change the shell for root in a pbuilder?
[13:11] <apachelogger> JontheEchidna: pingpinpingping
[13:11] <apachelogger> JontheEchidna: the muon needs to grow dist upgrade functionality
[13:37] <bulldog98_> !search /etc/console-setup/cached.kmap
[13:38] <bulldog98_> !help
[13:38] <bulldog98_> !help search
[13:38] <bulldog98_> !
[13:38] <bulldog98_> ~search
[13:38] <kubotu> incorrect usage, ask for help using 'kubotu: help search'
[13:38] <bulldog98_> kubotu: help search
[13:38] <kubotu> search <string> (or: google <string>) => search google for <string> | wp <string> => search for <string> on Wikipedia | unpedia <string> => search for <string> on Uncyclopedia
[13:38] <bulldog98_> kubotu: help
[13:38] <kubotu> help topics: 10 core modules: auth, basics, config, filters, httputil, irclog, remote, unicode, userdata, wordlist; 56 plugins: alias, autoop, autorejoin, bans, bar, botsnack, chanserv, debug, dns, eightball, excuse, factoids, googlefight, greet, hangman, host, identica, insult, iplookup, karma, keywords, lart, lastfm, map, markov, modes, nickrecover, nickserv, note, poll, q, quote, reaction, remind, ri, roshambo, rot, rss, salut,
[13:38] <kubotu> script, search, sed, seen, shiritori, spotify, time, topic, translator, tumblr, twitter, uno, urban, usermodes, wheeloffortune, wserver, youtube; 31 plugins ignored: use help ignored plugins to see why; 1 plugin failed to load: use help failed plugins to see why (help <topic> for more info)
[15:03] <JontheEchidna> apachelogger: as in, upgrade from one version of Kubuntu to another?
[15:03] <apachelogger> yus
[15:03] <apachelogger> well
[15:03] <apachelogger> detecing the presence
[15:04] <apachelogger> and invoke od-release-upgrade
[15:04] <apachelogger> s/od/do
[15:04] <JontheEchidna> kded/distupgradeedvent
[15:04] <JontheEchidna> the muon-notifier kded module checks for this already
[15:04] <apachelogger> ah cool
[15:05] <apachelogger> now if only the update-manager wasn't complete rubbish and could actually check for devel series
[15:08] <JontheEchidna> the impending hurricane has cancelled the first day of class tomorrow
[15:10] <JontheEchidna> probably because they don't want all the residents moving in today during a hurricane ^^
[15:55] <shadeslayer> ScottK: soprano 2.7.0+dfsg.1-0ubuntu2 does not fix kalzium as well
[16:26]  * Quintasan is back
[16:26] <Quintasan> apachelogger: ping
[16:26] <Quintasan> yofel: ping
[16:27] <yofel> pong, somewhat
[16:27] <Quintasan> yofel: Did you try Riddell's proposed fix to runtime?
[16:27] <yofel> nope (he proposed one?)
[16:29] <Quintasan> Yes he did, but now that I'm back I will try it in a second
[16:46] <apachelogger> well then
[16:46] <apachelogger> this update only screwed me over mostly
[16:46] <apachelogger> also it managed to do it worse than any before
[16:46] <apachelogger> hooray
[16:47] <yofel> really? what's broken now?
[16:47] <apachelogger> fglrx results in black screen
[16:47] <apachelogger> with ati driver no plymouth is shown and of course nice and cosy graphics artifacts are to be enjoyed
[16:48] <apachelogger> of course the xrandr stuff is still royally broken so my dual screen layout is all fckd up
[16:49] <apachelogger> oh
[16:49] <apachelogger> and there is this flash install thing
[16:49] <apachelogger> that now installs all of x86
[16:49] <apachelogger> thanks to qapt not being able to handle multi arch
[16:51] <apachelogger> RuntimeError: the PyQt4.QtCore module is version 1 but the PyKDE4.kdecore module requires version -1
[16:52] <apachelogger> http://i.imgur.com/cZCXq.png
[16:52] <apachelogger> truth be told, I expected an icon there
[16:53] <Quintasan> Grrrr
[16:53] <Quintasan> Now we need rebuild of kdelibs and runtime and it will most likely work
[16:55] <apachelogger> uhhh
[16:55] <apachelogger> rekonq
[16:55] <apachelogger> not wanting me google cookies
[16:55] <apachelogger> that is cool
[17:03] <apachelogger> chop chop make[2]: *** No rule to make target `/usr/lib/libQtTest.so', needed by `phonon/libphonon.so.4.5.56'.  Stop.
[17:04] <apachelogger> row row row your boat gently down the stream
[17:05]  * apachelogger kisses rekonq goodbye as it decided to hang itself
[17:09] <apachelogger> ok
[17:09] <apachelogger> someone explain that error to me
[17:10] <Quintasan> apachelogger: multiarch
[17:10] <apachelogger> that is not an explainatin of the error
[17:10] <apachelogger> actually
[17:10] <apachelogger> on second thought
[17:10] <apachelogger> I dont give a rats ass why the error is there
[17:10] <apachelogger> I want it gone
[17:10] <apachelogger> like now
[17:10]  * bulldog98 suggest apachelogger not to use flash at all :)
[17:10] <yofel> apachelogger: rebuild of phonon should fix it I guess
[17:10] <Quintasan> apachelogger: Fix it yourself if you know how
[17:11] <apachelogger> yofel: if that was not what I was doing
[17:11] <yofel> apachelogger: ah, then that's probably qtzeitgeist still not being rebuilt
[17:13] <apachelogger> ah
[17:13] <apachelogger> the madness
[17:17] <apachelogger> ahhh
[17:17] <apachelogger> broken opengl
[17:17] <apachelogger> what more does a man need
[17:18] <Quintasan> bber?
[17:18] <Quintasan> beer even
[17:41] <shadeslayer> apachelogger: ever written code in Objective C?
[17:42] <apachelogger> read
[17:42] <apachelogger> I do not write code that only makes sense on proprietary patent supporting vendor locked in systems
[17:43] <shadeslayer> hehe, just asking :P
[17:43] <shadeslayer> i
[17:43] <apachelogger> also I need to reboot
[17:43]  * shadeslayer hits his keyboard
[17:43]  * apachelogger sighs about the crap state of oneiric
[17:43] <shadeslayer> apachelogger: yeah, quite broken it is
[17:43]  * shadeslayer blames shadeslayer
[17:44] <shadeslayer> no sound again it seems
[18:07] <apachelogger> hm
[18:07] <apachelogger> shadeslayer, Quintasan: you know, I absolutely fail to get telepathy-kde to do anything useful
[18:10] <shadeslayer> apachelogger: depends on what you're trying to do
[18:10] <apachelogger> show my contacts
[18:10] <shadeslayer> and also depends on how you installed
[18:10] <apachelogger> or have the kcm remember my accounts for starters ^^
[18:10] <apachelogger> shadeslayer: ppa
[18:10] <shadeslayer> apachelogger: you need gnome-keyring for saving accounts iirc
[18:10] <shadeslayer> Quintasan: ^^ did you add that to depends?
[18:11] <shadeslayer> apachelogger: telepathy can't store passwords in kwallet yet
[18:11]  * apachelogger giggles
[18:11] <shadeslayer> they're working on integration for ksecretservice as of now
[18:11] <apachelogger> epic
[18:11] <shadeslayer> yup
[18:12] <apachelogger> I suppose that also explains all the rubbish it installed
[18:12]  * apachelogger autoremoves
[18:12] <shadeslayer> apachelogger: http://community.kde.org/Real-Time_Communication_and_Collaboration/Troubleshooting#TroubleShooting
[18:12] <Quintasan> absolutely fabulous
[18:12] <apachelogger> now
[18:12] <apachelogger> you know
[18:12] <shadeslayer> apachelogger: what else? :)
[18:12] <apachelogger> another famous piece of KDE 4 technology has a troubleshooting wiki
[18:12] <Quintasan> I can install nvidia-glx but it doesnt work
[18:12] <Quintasan> !#$%=&*(
[18:13] <apachelogger> also another closely related one
[18:13] <apachelogger> they are called akonadi and nepomuk
[18:13] <apachelogger> if you google them you will come to understand why I find it a bit worrying when software needs a troubleshooting documentation
[18:14] <apachelogger> Quintasan: define does not work
[18:14] <Quintasan> apachelogger: no screens found,  fails to load module and tells me to look into logs but nothing there
[18:14] <apachelogger> I have the funniest stuff with fglrx here
[18:15] <apachelogger> if I do a normal boot up it will kernel panic at X start or something
[18:15] <apachelogger> if I boot in failsafe and fire up kdm manually everything works as expected
[18:15] <apachelogger> also... is it just me or does oneiric boot take insanely long?
[18:16] <Quintasan> EE failed to load module "nvidia" (module-specific error, 0)
[18:16] <Quintasan> sup
[18:17]  * apachelogger must remember to not listen to bulldog98
[18:17] <bulldog98> apachelogger: sorry for me everything works fine
[18:18]  * apachelogger thinks bulldog98 should do bug triage
[18:18] <bulldog98> Quintasan: you have to blacklist nouveau
[18:18] <Quintasan> shadeslayer: also, who on earth made KDE Telepathy using gnome-keyring?
[18:19] <shadeslayer> Quintasan: they use telepathy as the backend
[18:19] <shadeslayer> which uses gnome-keyring to store stuff
[18:19] <shadeslayer> apachelogger: yep, takes *alot* of time here
[18:19] <Quintasan> shadeslayer: Uhm, I lol'd here
[18:20] <apachelogger> Quintasan: it is gnome people writing a technology, kde people coming along, adopting the technology and willingly take all the shit from aforementioned gnome people
[18:21]  * Quintasan also can't get his second monitor to work
[18:21] <Quintasan> ffs
[18:21] <shadeslayer> they're working on a ksecretservice backend
[18:21] <apachelogger> let's install opensuse
[18:21] <shadeslayer> so, there's some progress there
[18:21] <Quintasan> apachelogger: I really don't understand why start telepathy-kde when still depending on gnomeshit
[18:21] <apachelogger> I have friends there
[18:21] <shadeslayer> sure
[18:21] <apachelogger> we surely woud be most welcome
[18:22] <apachelogger> Quintasan: that is the beauty of it ... KDE people bowing to the superior might of GNOME
[18:22] <apachelogger> shadeslayer: there is no need for a backend
[18:22] <Quintasan> >superior
[18:23] <Quintasan> >GNOME
[18:23] <apachelogger> kss implements the secret storage spec
[18:23] <apachelogger> which gnome-keyring implements too
[18:23] <apachelogger> one can use both with one implementation
[18:23] <apachelogger> that is the entire point
[18:23] <apachelogger> BUT
[18:23] <Quintasan> I think Sput said that seeing GTK code makes him want to poke his eyes out with a rusty fork
[18:23] <apachelogger> kss has been in development for forever and 3 weeks
[18:23] <bambee> evening
[18:23] <apachelogger> and I am not entirely sure it will land in 4.8 either
[18:23] <apachelogger> AND
[18:23] <apachelogger> AFAIK the spec is still in draft stage
[18:24] <JontheEchidna> lost power for 20 minutes there...
[18:24] <apachelogger> JontheEchidna: hows the weather?
[18:25] <JontheEchidna> ironically, the power went out right as the hurricane past us :<
[18:25]  * apachelogger wonders if his comment on kde bug 267319 was too provocative
[18:25] <JontheEchidna> it never got too bad. It was a bit windy and there was  a lot of rain
[18:25] <yofel> Quintasan: oh, that error, boot with nomodeset then nvidia will work
[18:25] <Sput> Quintasan: and I still stand by that utterance
[18:25] <JontheEchidna> now there's a nice breeze, and less rain
[18:26] <apachelogger> JontheEchidna: ah, not much of a hurricane then, that is good :)
[18:26] <JontheEchidna> yeah, it was downgraded to a tropical storm by the time it got to me
[18:26] <apachelogger> Sput: surely your new employer will have some gtk stuff to do :P
[18:26]  * apachelogger only has good memories of gtk and glib
[18:27] <Sput> apachelogger: surely I wasn't hired for that :)
[18:27] <JontheEchidna> apachelogger: that guy is an ass :(
[18:27] <apachelogger> Sput: difficult times my friend :P
[18:27] <Sput> mmmh. gnome keyring seems to be optional in telepathy-mission-control
[18:27] <apachelogger> JontheEchidna: yus, always has been
[18:28] <Sput> not sure if they can also use kwallet somehow
[18:28] <apachelogger> Sput: or maybe it is just badly engineered :P
[18:28] <Sput> my telepathy-mission-control is built without gnome keyring support, and KDE manages to remember my passwords
[18:29] <apachelogger> unencrypted storage in gconf then
[18:29] <apachelogger> which would probably make sense
[18:31] <Sput> yeah looks like it, at least my kwallet doesn't seem to be storing account pws
[18:33]  * apachelogger really cannot manage to find the energy to read JontheEchidna's mail on multiarch
[18:33] <JontheEchidna> lol
[18:33] <apachelogger> actually I wonder what is with the tiredness these days
[18:33] <apachelogger> must be side effect from being sober
[18:33]  * bulldog98 has already done that
[18:33] <apachelogger> rather horrible
[18:33] <apachelogger> bulldog98: yet you did not give input :P
[18:33] <bulldog98> apachelogger: I was the guy reporting the problem, so that’s input :P
[18:35] <apachelogger> true
[18:36] <bulldog98> apachelogger: I would have given a mail, but kmail is broken (doesn’t send mails atm)
[18:37] <bulldog98> and it seems like I can blame ScottK or yofel for that
[18:38]  * bulldog98 is in favor of option 3 cause we keep binary compitabilty
[18:38]  * bambee wonders if kmail works sometimes
[18:39] <bulldog98> bambee: it will with release 42
[18:42] <bambee> Will it work as well as other releases ? :P (no, I am kidding)
[18:42] <bulldog98> normaly kmail works quite well
[18:53] <sheytan> apachelogger: ping ping
[18:53] <sheytan> can I have the kdepase page again?
[18:53] <sheytan> I couldn't compile phonon last time, some errors
[19:20] <apachelogger> sheytan: it is gone
[19:20] <apachelogger> check your shel history
[19:20] <apachelogger> shell even
[19:20] <apachelogger> or maybe it was shellll
[19:20] <apachelogger> sheeel
[19:21] <apachelogger> something like that anywayz
[19:23] <sheytan> i did, it's gone ;/
[19:23] <sheytan> if you could, mail me it
[20:28] <apachelogger>     KIO::Integration::AccessManager::putReplyOnHold(reply);
[20:28] <apachelogger> either that thing is broken or rekonq is using it wrong
[20:28] <apachelogger> it causes quite the hang
[20:28] <apachelogger> when downloading a file
[20:28] <apachelogger> (or trying to)
[20:28] <apachelogger> what I find best about it is how that can block the UI
[20:29] <apachelogger> *thumbs up*
[20:29] <bulldog98> apachelogger: thread support is planed, but nobody steped up for that
[20:59] <CIA-1> [kubuntu-default-settings] Harald Sitter <apachelogger@ubuntu.com> * apachelogger@ubuntu.com-20110828205905-o0uf4v3691bkrfz4 * (debian/changelog share/config/kdeglobals) Set rekonq as explicit browser application in kdeglobals. Otherwise konqueror will be used as default when both are installed.
[21:01] <apachelogger> perhaps we should chop kde-workspace's history
[21:01] <apachelogger> it has become rather fat
[21:04] <apachelogger> yofel: ping
[21:06] <yofel> ?
[21:07] <apachelogger> yofel: why for the love of kent beck does the install default wallpapers buttons install the wallpapers from workspace and artwork?
[21:07] <apachelogger> I find this the ultimate overkill
[21:07] <apachelogger> plus
[21:07] <apachelogger> JontheEchidna: qapt aint asking me whether I want to install that rubbish
[21:08] <apachelogger> it just asks for auth and I have no idea what for
[21:08] <JontheEchidna> the wallpapers?
[21:08] <apachelogger> yus
[21:08] <yofel> well, what do you want to install then? Formery the button was installing stuff from artwork, then one of the artwork folks complained that we install the artwork stuff without installing the workspace wallpapers first
[21:09] <JontheEchidna> you did click the "install wallpapes" button, so presumably you wanted to
[21:09] <yofel> adding 2 buttons is nonsense, and adding a dialog that asks you which to install isn't something I know how to do
[21:09] <apachelogger> JontheEchidna: yeah but right now that drags in 10003482^50 trazillion TiB
[21:10] <apachelogger> it would be jolly nice to tell me before hand, because I have 56k and could rather not effort to wait that long
[21:10] <apachelogger> yofel: no, it installed the workspace one
[21:11] <yofel> apachelogger: well, _now_ it installs both
[21:11]  * apachelogger files the bukx
[21:11] <yofel> before it installed the one from artwork
[21:11] <apachelogger> because you broke it all
[21:11] <yofel> huh?
[21:11] <apachelogger> at the time the feature was introduced there were no wallpapers in artwork
[21:11] <yofel> I never wrote that patch
[21:11] <yofel> ah
[21:12] <yofel> well, what's that button *supposed* to do then?
[21:12] <yofel> and how is lord user supposed to know how to install the wallpapers that are shipped with kde?
[21:12] <yofel> even in artwork
[21:13] <apachelogger> package manag0r
[21:13] <apachelogger> the wallies in workspace are the latest  hot ones, so we want those
[21:13] <apachelogger> not the bazillion wallies before the current ones
[21:13] <yofel> yeah, like a kde user will look in the package manager to add wallpapers...
[21:14] <yofel> that's why that button is there in the first place
[21:22] <apachelogger> yofel: yes, to give the default
[21:22] <apachelogger> not every wallpaper ever 
[21:24] <apachelogger> bug 836276
[21:25] <apachelogger> yofel: so what do we do with the button?
[21:26] <yofel> hm, I agree with the descriptions. We somewhat decided to leave the renaming of the package to the debian folks though
[21:26] <yofel> as for the button...
[21:27] <yofel> I didn't want to break the behaviour the button had before, so I just added it. If you think installing the artwork wallpapers isn't worth the download size then drop it
[21:27] <apachelogger> roger
[21:28] <apachelogger> who wants to bring in a new snapshot of qtwebkit?
[21:28] <apachelogger> the current one is utter rubbish
[21:28] <apachelogger> or maybe rekonq is
[21:28] <apachelogger> on that note
[21:28] <apachelogger> we need a newer snapshot of rekonq
[21:29] <apachelogger> oh right
[21:29] <apachelogger> kde-wallpapers is now a seperate source package, no?
[21:29] <yofel> yep
[21:30] <apachelogger> you shoudl change the bug then :P
[21:30]  * apachelogger installs qtwebkit-dbg before rekonq goes down again
[21:30] <apachelogger> funniest thing
[21:30] <apachelogger> if rekonq went down once, it will again when restoring
[21:30] <apachelogger> so I think this is a qtwebkit issue really
[21:32] <yofel> bah, didn't pay enough attention to the packages
[21:33] <apachelogger> man
[21:33] <apachelogger> webkit is terrible
[21:33] <apachelogger> 189 mib of dbg symbols
[21:33] <apachelogger> that is just insane
[21:34] <apachelogger> do we not lzma them, or something?
[21:34] <apachelogger> --with pkgkde_symbolshelper,lzma
[21:34] <apachelogger> should
[21:34] <apachelogger> :S
[21:34] <apachelogger> imagine how big that is without lzma!!!!
[21:38] <apachelogger> #6  primarySimpleFontData (f=0x7f3d7ecabe68, this=0x0) at platform/graphics/FontFallbackList.h:61
[21:38] <apachelogger> mhhh
[21:38] <apachelogger> "this be null, go away"
[21:58] <apachelogger> JontheEchidna: ping
[21:58] <JontheEchidna> pong
[21:58] <apachelogger> JontheEchidna: cant you get the architecture without the name from apt?
[21:58] <apachelogger> or is it literally appended to the name?
[21:59] <apachelogger> in the binary cache
[22:00] <JontheEchidna> the arch is not part of the name in the binary cache, I don't think
[22:01] <JontheEchidna> which is why apt's fullname function has to append it + return an std::string
[22:01] <JontheEchidna> it can't simply point at the name in the binary cache
[22:01] <apachelogger> right
[22:01] <apachelogger> BUT
[22:01] <apachelogger> why is there no explicit interface to query the arch?
[22:02] <JontheEchidna> hmm
[22:02] <apachelogger> neither of the approaches you presented seems very object oriented... the arch would just be a property of a package, not of the package's name
[22:03] <apachelogger> if a consumer were to require a fullname they could still concat themselfs
[22:03] <apachelogger> which IMHO makes loads more sense than forcing users to go the other way around and split the fullname
[22:05] <apachelogger> because for muon you'd not want the names to be displayed with arch anyway, but perhaps a dropbox where the user can select the arch of the package to install (or inform the user that only a x86 version of this package is availabe and that all deps will thus be fullfilled using x86 blah blah)
[22:05] <apachelogger> (in the muon UI anyway)
[22:05] <JontheEchidna> apachelogger: would it be possible for you to install a foreign-arch package with dpkg or apt-get, and then paste the output of apt-cache show foo?
[22:05] <apachelogger> for the software center you'd only need it for dep resolving anyway
[22:05] <JontheEchidna> it might help me, since I'm not on a 64-bit system
[22:07] <apachelogger> JontheEchidna: http://paste.ubuntu.com/676845/
[22:07] <JontheEchidna> hmm, apt-cache show says Architecture: amd64 :s
[22:07] <JontheEchidna> oh, but when you specify :i386 it says i386
[22:08] <apachelogger> that is because it will prefer the x64 version of course
[22:08] <apachelogger> flash might make more sense
[22:08] <apachelogger> sec
[22:09] <apachelogger> JontheEchidna: http://paste.ubuntu.com/676846/
[22:09] <apachelogger> -installer has a x64 version
[22:09] <apachelogger> BUT
[22:09] <apachelogger> -downloader only can do x86, so it shows the only available arch by default
[22:09] <apachelogger> which is i386
[22:10] <JontheEchidna> ok, that's good
[22:10] <JontheEchidna> it would be trivial to add an arch() function to Package
[22:11] <apachelogger> right, then you just need to look at the host arch I suppose and prefer that whenever possible
[22:11] <JontheEchidna> hmm, I think my lappy is 64-bit capable, actually. I should just reinstall w/ 64bit
[22:12] <apachelogger> for 11.10 simply filtering duplicated entries based on best match for host arch should be sufficient
[22:12] <apachelogger> later you can have a gui integration in muon to install/remove one specific arch
[22:12] <apachelogger> which will probalby be a bit tricky
[22:12] <apachelogger> UIwise
[22:12] <JontheEchidna> yeah, since it's tricky to know which packages can be multiarch'd
[22:13] <JontheEchidna> oh, actually there is the MultiArch: field
[22:13] <apachelogger> well, name occurs more than once -> add checkbox with foreach entry, arch()
[22:13] <apachelogger> uh
[22:13] <apachelogger> even better
[22:14] <apachelogger> anyhow, I think doing the actual UI is the really tricky part there ^^
[22:14] <JontheEchidna> yeah
[22:14]  * JontheEchidna downloads 64-bit iso
[22:15] <JontheEchidna> blast, we're oversized. I'll have to use a dvd :/
[22:15] <apachelogger> no usb stick?
[22:16] <JontheEchidna> not big enough, no
[22:16] <apachelogger> uff, I should send you some :P
[22:16] <JontheEchidna> all I has is this 128 MB one I use for school papers, etc
[22:16] <apachelogger> Quintasan: not only do I have devices coming out my ears, I also have usb sticks coming out my nouse :P
[22:16] <JontheEchidna> lol
[22:17] <apachelogger> The custom templates folder for identity "Ubuntu" does not exist (anymore); therefore, the default templates folder will be used.
[22:17] <apachelogger> way to go kmail
[22:17] <JontheEchidna> I can't use kmail. It's not able to show my kubuntu-devel folder anymore
[22:17] <apachelogger> tooo big
[22:17] <apachelogger> cleanup :P
[22:17] <JontheEchidna> only 8k mails in thar
[22:18] <JontheEchidna> plus it has this habit of duplicating mails
[22:19] <JontheEchidna> apachelogger: while I'm downloading, could you search for libc6 in synaptic and tell me if it has duplicate entries? (It's using solution 1 in my mail)
[22:20] <apachelogger> awww
[22:20] <apachelogger> I only ripped that piece of fluff out earlier today :(
[22:20] <JontheEchidna> lol
[22:20] <apachelogger> oh
[22:20] <apachelogger> and my vbox is broken
[22:20] <apachelogger> darn it
[22:21] <JontheEchidna> if there's more than one it means that apt has two PkgIterators for both the native and multiarch versions
[22:21] <JontheEchidna> and QApt::Package is a OO class wrapping a PkgIterator
[22:22] <JontheEchidna> well, it has to look at several other apt classes to get all the info you'd want from a package, but that's because libapt is an un-OO piece of crap
[22:22] <apachelogger> lol
[22:23] <apachelogger> JontheEchidna: well, IMHO that is how it is supposed to be
[22:23] <apachelogger> they are 2 different packages after all
[22:24] <JontheEchidna> initially I think it'll be ok to show both
[22:24] <apachelogger> so it is essentially up to the API consumer to implement multiarch support
[22:25] <apachelogger> i.e. the consumer would decide which to filter, or at least tell libqapt what policy to implement (return all, return only host, return only host except when no host is available, return arch xy)
[22:25] <apachelogger> or you implement policy handling right away and default to return - only host except when no host is available -
[22:26] <apachelogger> that way even without change API consumers get a sane package listing with multiarch systems
[22:29] <JontheEchidna> I'm trying to work out how I'd do this:
[22:29] <JontheEchidna> When initialized, QApt::Backend creates a Package object for every PkgIterator in the apt depcache
[22:30] <JontheEchidna> it then stores all of those in a QList
[22:30] <JontheEchidna> (a qlist of pointers)
[22:31] <JontheEchidna> The Muon Package Manager's PackageModel is populated with Backend::availablePackages() which returns a copy of that list
[22:31] <apachelogger> that will need to become more expensive
[22:31] <JontheEchidna> that's ok I think. sorting within the PackageModel is the bottleneck there anyways
[22:32] <JontheEchidna> so if I create some flags, I can make a Backend::packages(ArchTypeFlag) that will filter out archs based on ArchTypeFlag
[22:33] <JontheEchidna> and return a list of only native, only foreign, or both
[22:33] <apachelogger> you'd probably also want specific foreign
[22:33] <apachelogger> but for now I'd keep it to the bare minimum necessary for sane behavior
[22:33] <JontheEchidna> yeah
[22:34] <apachelogger> thing is, how would you implement the algorithm
[22:34] <JontheEchidna> the amount of archs dpkg/apt, et al. support is not guaranteed anyways
[22:34] <JontheEchidna> so if we suddenly add an arch we'd need a new flag
[22:34] <apachelogger> you could just have a property
[22:35] <apachelogger> QList<arches>
[22:35] <apachelogger> arches being a QString
[22:35] <apachelogger> or QBA perhaps
[22:35] <JontheEchidna> bleh, that means doing a strcmp when filtering :(
[22:35] <JontheEchidna> well, the QString equivalent, but yeah :P
[22:35] <apachelogger> you'll need strcmp anyway
[22:36] <apachelogger> or how would you implement the algorithm?
[22:36] <JontheEchidna> so mr consumer calls packages(NativePackages)
[22:36] <apachelogger> because my first thought is to have a map<ql1s, package*>
[22:36] <apachelogger> in the thing that makes the qlist
[22:37] <apachelogger> then for each iterator you check if the ql1s is already in the map
[22:37] <JontheEchidna> we go to our store of every package, go through each, and if package arch matches native arch, add to the list we return
[22:37] <apachelogger> if so you check your policy
[22:38] <JontheEchidna> with the map we only incur the cost of checking the arch once
[22:38] <JontheEchidna> right?
[22:39] <JontheEchidna> at the cost of storing a qmap with 35,000 nodes :P
[22:39] <apachelogger> you throw the map away afterwards
[22:39] <apachelogger> once built you only need the resulting qlist
[22:40] <apachelogger> also it might be a map with 35,000*a nodes :P
[22:40] <apachelogger> oh
[22:40] <apachelogger> actually
[22:40] <apachelogger> if policy is return all -> use existing algorithm
[22:40] <apachelogger> otherwise use the qmap
[22:40] <apachelogger> phew
[22:40] <apachelogger> now that just saved our cpu from setting on fire \o/
[22:41] <JontheEchidna> so you'd go through each package in the canonical list, map package to arch, then make a qlist from the packages with the correct arch?
[22:41] <apachelogger> nope
[22:41] <apachelogger> way too expensive
[22:41] <apachelogger> so
[22:41] <apachelogger> you iter over packages in libapt
[22:42] <apachelogger> foreach iter:
[22:42] <apachelogger> if policy is showmeall -> add to final qlist
[22:42] <apachelogger> else -> invoke map based algorithm
[22:42] <apachelogger> looks like this:
[22:42] <JontheEchidna> oh, I should add that the iterating over libqapt iterators is only done once when Backend::init() is called
[22:42] <apachelogger> get latin1name
[22:42] <apachelogger> check if latin1name is already in map
[22:42] <apachelogger> if not -> add
[22:43] <apachelogger> else -> invoke policy check
[22:43] <apachelogger> looks like this:
[22:43] <JontheEchidna> ::init() populates a QList<Package *>, and this list doesn't change
[22:43] <apachelogger> uha
[22:43] <apachelogger> uhaaaa
[22:43] <apachelogger> JontheEchidna: then you need your own container structure
[22:43] <apachelogger> similar to a map
[22:44] <apachelogger> struct {QLatin1String string, Package *pkg}; where pkg is in the master list
[22:44] <JontheEchidna> or
[22:45] <JontheEchidna> hmm, nvm
[22:45] <apachelogger> well, you need a more complicated container, as you need to store the whole list and then build a filtered list
[22:45] <apachelogger> and I still did not finish explaining my approach to the filtering :P
[22:46] <apachelogger> so, if there is already a package with the same name in the container, then you invoke a policy check
[22:46] <apachelogger> which basically goes: does current Package*->arch() match policy better than Package* in the container
[22:47] <apachelogger> e.g. in the container we have an i386 Package, but host is amd64 -> overwrite the entry with our new Package * when amd64 is preferred
[22:48] <apachelogger> if the policy is listAllForeign you'd skip the amd64 package
[22:49] <apachelogger> if you want poth you simply add it to the container (so now you have two Package* with the same latin1name in your container)
[22:49] <apachelogger> so a lot of strcmp but still relatively cheap
[22:50] <apachelogger> (as a matter of fact it might just make more sense to trade the QL1S instance for the pointer dereference)
[22:50] <apachelogger> so you'd only have a list as container
[22:51] <apachelogger> but I don't think you'd get away from the n*n iteration to filter
[22:52] <JontheEchidna> if only APT's multiarch impl. wasn't so hacky :(
[22:52] <JontheEchidna> they probably did it the way they did so they wouldn't invalidate the binary package cache format
[22:52] <apachelogger> yus yus, then again qstr stuff is probably more efficient than any std::string thing anyway ^^
[22:53] <apachelogger> so I think their stuff is just fine from a qapt POV
[22:54] <apachelogger> in fact I could not think of a much better approach for the APT impl
[22:54] <apachelogger> JontheEchidna: about your syantpic question -> no duplicates from what I can tell
[22:55] <JontheEchidna> hmm
[22:55] <apachelogger> also it does not indicate multiarch ness or anything
[22:55] <apachelogger> very useless tool :S
[22:55] <apachelogger> oh
[22:55] <apachelogger> JontheEchidna: nvm
[22:55] <JontheEchidna> it doesn't say libc6:i386 or anything?
[22:55] <apachelogger> libc6
[22:55] <apachelogger> and then there is libc6-amd64
[22:55] <apachelogger> mind me, the vbox is x86 it seems ^^
[22:56]  * apachelogger installs natively
[22:56] <JontheEchidna> my iso is half done downloading
[22:59] <apachelogger> ah yes
[22:59] <apachelogger> JontheEchidna: proper duplication
[22:59] <JontheEchidna> ok, good
[22:59] <apachelogger> libc6 and libc:i368
[22:59] <apachelogger> eh libc6:i386
[22:59] <apachelogger> also there is libc6-i386
[22:59] <apachelogger> :O
[22:59] <apachelogger> wtf
[23:00] <JontheEchidna> that's the pre-multiarch solution, I think
[23:00] <apachelogger> yeah
[23:00] <apachelogger> looks like it
[23:01] <apachelogger> JontheEchidna: so kmail will, I sent a quick summary of this as reply to your mail
[23:01] <JontheEchidna> cool, thanks
[23:47] <JontheEchidna> brb, reinstalling kubuntu
[23:56] <JontheEchidna> Default desktop doesn't look to shabby: http://i.imgur.com/Jpa6Q.png
[23:59] <apachelogger> try it at a non-widescreen resolution
[23:59] <apachelogger> the wallpaper almost made me throw up