[03:37] <RAOF> Damnit, laptop. Stop thermally throttling yourself!
[03:48] <mlankhorst> yeah start making burn holes on raof's lap
[03:51] <RAOF> It's sitting on a table. It's welcome to try and burn that!
[04:09] <robert_ancell> RAOF, bug #1102762 is fixed right?
[04:09] <ubot5`> bug 1102762 in Mir "Support nouveau drivers" [Medium,In progress] https://launchpad.net/bugs/1102762
[04:37] <racarr> RAOF: Does Mesa still live on github?
[04:37] <racarr> I was hoping to see a shiny eglplatform revision but itsnothere
[04:37] <racarr> oh I see, mir-ppa is the branch
[04:54] <RAOF> racarr, robert_ancell: Correct
[05:50] <racarr> are we having the weekly?
[05:51] <RAOF> I presume so?
[05:52] <robert_ancell> racarr, RAOF, yes
[06:00] <RAOF> FIRST!
[06:03] <hikiko> hi
[06:30] <RAOF> Oh, thomi - could we get Saucy builds of mesa out of github, too?
[07:01] <mlankhorst> is mir used on the touch image yet?
[07:37] <tvoss> mlankhorst, it's not officially in the image, but working towards that goal
[07:37] <tvoss> greyback, good morning :)
[07:38] <greyback> tvoss: well howdy
=== alan_g is now known as alan_g|cold
[11:22] <greyback> kdub: hey, good news: unity runs on mir in process: http://ubuntuone.com/58W4ARPX7KKPLdoHHsDrvL
[11:22] <greyback> kdub: but performance isn't buttery, in fact appears a bit slower than out of process
[11:22] <greyback> this is galaxy nexus
[11:30] <greyback> kdub: doing some initial profiling, it appears the buffer swap is slow (often over 30ms per frame)
=== greyback is now known as greyback|lunch
=== mmrazik is now known as mmrazik|afk
=== greyback|lunch is now known as greyback
=== francisco is now known as Guest49422
[13:19] <hikiko> alf___, here?
[13:19] <alf___> hikiko: hi, here
[13:19] <hikiko> hi!
[13:21] <hikiko> I have a problem when I compile, I get this error: Linking CXX shared library ../../lib/libmirserver.so ../../lib/libmirfrontend.a(session_mediator.cpp.o):(.data.rel.ro._ZTVN3mir8frontend15SessionMediatorE[_ZTVN3mir8frontend15SessionMediatorE]+0x38): undefined reference to `mir::frontend::SessionMediator::drm_auth_magic(google::protobuf::RpcController*, mir::protobuf::DRMMagic const*, mir::protobuf::DRMAuthMagicStatus*, google::protobuf::Closure*)
[13:21] <hikiko> ' because I include files from the graphics/gbm, do you know how I could change the CMakeLists.txt so that the linker doesn't complain anymore? :)
[13:21] <hikiko> i mean the cmakelists from the sdl platform
=== mmrazik|afk is now known as mmrazik
[13:25] <hikiko> i tried to make it identical to the gbm cmakelists
[13:26] <hikiko> but I still get thi serror
[13:27] <alf___> hikiko: you need to create a session_mediator_sdl which is going to be basically a copy of session_mediator_android.cpp, and build with it
[13:32] <hikiko> alf___, I can't use the existing gbm one?
[13:32] <hikiko> what's the session mediator?
[13:41] <alf___> hikiko: SessionMediator is the object that (ultimately) handles RPC requests from clients. At this point you don't need to support drm_auth_magic, since it's used by clients only. I am not sure if you can support it at all, since you will need to be the DRM master to do so.
[13:42] <hikiko> I can't be the drm master, the xserver will be :S
[13:42] <alf___> hikiko: perhaps you will just need to forward the request to X then
[13:42] <hikiko> but I need to authenticate to the xserver
[13:43] <alf___> hikiko: but anyway, for the first step (getting render_to_fb to run) it's not needed
[14:09] <alf___> hikiko: ... which makes me wonder if we need mir_connection_drm_auth_magic() at all. No one seems to be using it.
[14:09] <alf___> RAOF: ^^
[14:10] <hikiko> I need to do the magic I guess because when I tried to create my gbm buffers in my example without authentications and inside X I didn't have permissions
[14:11] <hikiko> +then you have to somehow pass the device descriptor to the xserver I guess (not sure)
[14:17] <alf___> hikiko: For GBM at least, we are sending a pre-authenticated DRM fd to the client (e.g. to mesa), so I guess you will need to do the same for SDL, with the difference that you will need to ask X to authenticate it (XF86DRIAuthConnection). In any case, mir_connection_drm_auth_magic() seems to be unused right now. I will talk to RAOF to ensure he doesn't need it and remove it.
[14:19] <hikiko> mmm maybe :) let me find where you do this authentication for gbm
[14:19] <alf___> hikiko: gbm_display_helpers.cpp
=== olli_ is now known as olli
[15:05] <olli> is anybody attending the "Support for convertibles and touch screen on Ubuntu desktop ( Client )" session
[15:05] <olli> Saviq, kgunn
[15:05] <kgunn> checking
[15:06] <Saviq> olli, I was going to go to "Touch System Settings"
[15:06] <olli> Saviq, is that now?
[15:06] <Saviq> olli, yes
[15:06] <Saviq> no
[15:06] <Saviq> in an hour
[15:06] <Saviq> ok, coming to the convertibles
[15:06] <kgunn> sorry, i was going to kylin images...but maybe i should go to that instead
[15:07] <olli> kgunn, Saviq I think we should cover both
[15:07] <olli> it would make sense to me for kgunn to go to kylin
[15:07] <kgunn> ack
[15:07] <olli> as this is a "customer" of ours and Saviq to go to the touch one to keep them straight/honest
[15:12] <kdub> greyback, i'm working on performance stuff today, can take a look at inprocess qml going slowly
[15:12] <greyback> kdub: ok cool
[15:12] <kdub> greyback, that video looks ok though, is it only sometimes that it takes 30ms?
[15:13] <greyback> kdub: overall frames took just a bit over 30ms when adding in the rendering time
[15:14] <greyback> kdub: handy tip is to set QML_RENDER_TIMING=1 before running a qml app, you get frame info out with that
[15:15] <kdub> greyback, what sort of special branches do i need?
[15:15] <greyback> kdub: see racarr's email last night, he gives all the info there.
[15:16] <kdub> greyback, cool, thanks
[15:35] <kdub> busy morning :) status, yesterday worked on a cleanup branch, began swap interval 0 work
[15:41] <mlankhorst> oh thanks for reminding me, drm needs to add support for thjat
[15:41] <mlankhorst> that*
[15:42] <moustafa> Hey! Is anybody else having difficulty installing mir through the PPA due to unmet dependencies?
[15:42] <kdub> moustafa, what package?
[15:43] <moustafa> kdub, 'mir'.  The conflict is caused by libmirserver0 being at 0.0.3bzr687raring0, where mir requires it to be at 0.0.2bzr643raring0
[15:43] <mlankhorst> remove mir
[15:44] <alf___> status: integrating lttng with mir reporters
[15:52] <moustafa> mlankhorst, It can't even install :)
[16:02] <kgunn> moustafa: i had the same issue
[16:03] <kgunn> thomi: ^
[16:03] <thomi> kgunn: we're fixing that today :)
[16:03] <kgunn> moustafa: i was pestering folks last night, but my understanding is, you can build the libmirserver from source and it'll work
[16:04] <thomi> hmmm
[16:04] <kgunn> thomi: thanks ;) it helps lazy folk like me
[16:04] <thomi> isn't libmirserver build by the mir source package?
[16:04] <moustafa> kgunn, well, that would be the obvious way of doing it, but I like being a bit lazy (curse you, PPAs!)
[16:05] <kgunn> thomi: well, based on my interaction with robert, it seemed the source didn't match the ppa (? sheepishly)
[16:06] <thomi> kgunn: yeah, sorry, I assumed you were complaining about the u-s-c issue :)
[16:06] <thomi> moustafa: which series are you looking at?
[16:07] <moustafa> thomi, Raring
[16:07] <kgunn> thomi: me also
[16:08] <thomi> hmmm
[16:09] <thomi> moustafa: right, as mlankhorst mentioned, 'mir' is an old package, that you should remove. If you want the demos, install mir-demos
[16:10] <thomi> moustafa: the packages (listed here: https://launchpad.net/~mir-team/+archive/staging/+packages?field.name_filter=&field.status_filter=published&field.series_filter=raring) look like they've built fine
[16:13] <moustafa> thomi, So, I would require mir-demos to run mir as a system compositor with X?  Or can I just go ahead with mir-demos.
[16:13] <moustafa> thomi, That said, mir isn't even installed (refuses to do so anyway)
[16:14] <thomi> moustafa: hmm, you said "The conflict is caused by libmirserver0 being at 0.0.3bzr687raring0, where mir requires it to be at 0.0.2bzr643raring0", which made me think you were installing the 'mir' package - maybe you could pastebin your shell session?
[16:15] <moustafa> thomi, http://paste.ubuntu.com/5668074/
[16:15] <mlankhorst> thomi: why is there no breaks/replaces on mir then?
[16:16] <thomi> mlankhorst: there should be. I'll do tha today
[16:16] <thomi> moustafa: right, don't install 'mir'
[16:16] <thomi> moustafa: if you want to run the demo shell, that is now in the 'mir-demos' package
[16:17] <thomi> moustafa: if you want to run on top of X... I'm not sure what you need for that. maybe someone like kdub, alf___ or kgunn can tell you
[16:18] <kgunn> hikiko: ^
=== francisco is now known as Guest53340
[16:22] <hikiko> it doesn't run on top of x yet :)
[16:22] <hikiko> moustafa, there's no emulator yet
[16:22] <hikiko> under construction!
[16:25] <moustafa> hikiko, Duly noted!
[16:26] <hikiko> but moustafa I think that some examples
[16:26] <hikiko> work on X
[16:27] <hikiko>  alf___ L
[16:27] <hikiko> ^
[16:39] <kdub> greyback, ping
[16:39] <greyback> kdub: pong
[16:41] <kdub> i've gotten to the point where ipc qml clients are running, how to run qml-phone-shell?
[16:41] <kdub> i'ts failing to run, i'd guess there's some service or something that I haven't started up
[16:43] <greyback> kdub: you running phablet-integrate-mir-no-sw-cursor ?
[16:44] <kdub> greyback, no, i've just jammed some hello world qml files on to test with qmlscene so far
[16:44] <greyback> and they're not working?
[16:45] <kdub> no, they are, its qml-phone-shell that isnt
[16:45] <greyback> ok
[16:45] <greyback> you compiled it with the "build" script I hope
[16:46] <kdub> no, is there a wiki?
[16:46] <greyback> nope, sorry
[16:46] <kgunn> kdub: for building
[16:46] <kgunn> http://unity.ubuntu.com/getinvolved/development/unitynext/
[16:46] <kgunn> assume that's what you were looking for
[16:46] <greyback> ah, news to me
[16:47] <kgunn> greyback: :)
[16:47] <kdub> yeah, i was just thinking 'theres some primitive unity knowledge i'm missing here...'
[16:47] <kdub> thanks kgunn
[16:47] <kgunn> manager== data aggregator :)
[16:47] <greyback> kdub: but most of that info is desktop specific. Really all you need is "./build" and "./run"
[16:47] <racarr>  -m -f ?
[16:48] <kgunn> greyback: you mean ./run_on_device ?
[16:48] <greyback> racarr: worked without for me on last run. "-m" only for desktop
[16:48] <greyback> kdub: are you compiling on the device, or cross-compiling on your PC?
[16:48] <kdub> cross compiling
[16:50] <greyback> kdub: what is easiest for you to do is: "./run_on_device -s" to push code to the phone and compile it there. It will leave the qml-phone-shell executable in /home/phablet/shell/builddir/src/shell/
[16:50] <greyback> kdub: but note that install compiler, etc on the phone
[16:50] <greyback> s/install/installs/
[16:52] <kdub> greyback, thanks, will give it a try
=== mmrazik is now known as mmrazik|afk
[17:00] <racarr> Ah
[17:00] <racarr> status: Met with katie, talked about surface types v. roles
[17:01] <racarr> now eating a bagel and deciding what order to do things in
[17:08] <kdub> so many phablet scripts to learn about
[17:09] <racarr> yeah
[17:10] <racarr> greyback: Wondering if there is anything I can do in particular while you are still on
[17:12] <greyback> racarr: getting your platform-api and qtubuntu branches merged would be handy, so we all can get unity+mir running easily. Think it's much effort?
[17:13] <racarr> greyback: Unfortunately blocked atm
[17:13] <racarr> platform-api is about to (tomorrow, friday?) have a big refactoring
[17:13] <greyback> racarr: :( what by?
[17:13] <racarr> and we are going to land it afterwards
[17:13] <greyback> aha yes, I recall taht
[17:13] <greyback> ok, well best wait for that so.
[17:13] <racarr> the eglplatform.h bits landed though
[17:13] <tvoss> racarr, is QtMir still current or do we do things in QtUbuntu?
[17:13] <greyback> cool
[17:13] <racarr> tvoss: It's all qtubuntu
[17:14] <kgunn> sweet
[17:15] <racarr> tvoss: Though mir qtubuntu is a little different
[17:16] <racarr> due to mir doing keymapping + key input wasn't really working so much in qtubuntu
[17:16] <racarr> (mir client)
[17:16] <racarr> Everyone likes mapping keys on the client
[17:16] <racarr> but I don't understand why
[17:16] <racarr> Espescially because you have to map them on the server too for the input ilter
[17:16] <racarr> filter*
[17:18] <tvoss> racarr, ack
[17:22] <racarr> alf___: Ping?
[17:22] <racarr> p.s. nice underscore
[17:35] <racarr> hmm
[17:35] <racarr> how do you acceptance test that pressing control+c doesn't
[17:35] <racarr> kill the server
[17:39] <racarr> I dont even know :/
[17:39] <racarr> Too many missing pieces
[17:41] <kdub> racarr, what are you trying to do :) shouldn't sigterm/sigint kill the server?
[17:41] <racarr> yes but ctrl+c shouldn't generate
[17:42] <racarr> any signals
[17:42] <racarr> because mir should parse it like anything else and pass it on to applications
[17:43] <racarr> which is easy
[17:43] <racarr> tcgetattr (vtfd), cfmakeraw(attr) tcsetattr(attr)
[17:43] <racarr> and it's easy enough to, well, mock out this in to a file operations bit, test that the virtual terminal uses these interfaces correctly
[17:43] <racarr> test that the display calls the virtual terminals disable_control_sequences methods appropriately
[17:44] <racarr> etc
[17:44] <racarr> but there is no acceptance test
[17:51] <racarr> you just have to mock the kernel ;)
[17:53] <racarr> and with that Im giving up on acceptance testing this for now :p
[17:53] <racarr> it may be possible to use a pty...I dunno
[17:54] <kdub> greyback, what is "phablet-integrate-mir-no-sw-cursor"? a branch somewhere?
[17:55] <greyback> kdub: lp:~robertcarr/unity/phablet-integrate-mir-no-sw-cursor
[17:55] <greyback> it adds code to qml-phone-shell to create a Mir server, add cursor and other bits
[17:55] <kdub> ah, i should read that email more carefully :)
[18:11] <racarr> https://code.launchpad.net/~robertcarr/mir/disable-sequences-and-add-terminate-handler/+merge/164014 exists
[18:11] <racarr> im not in any hurry to land it though
[18:14] <kdub> greyback, http://pastebin.ubuntu.com/5668437/
[18:15] <kdub> not sure what package qpa/qplatformnativeinterface.h comes from...
[18:15] <greyback> kdub: qtbase5-private-dev
[18:15] <greyback> I'm surprised the run_on_device script didn't install that for you
[18:17] <kdub> also, i found i had to install the GL headers in addition to the gles headers, struck me as a bit strange
[18:19] <greyback> kdub: that's probably wrong for phone. In src/shell/CMakeList.txt you should substitute "${OPENGL_gl_LIBRARY}" with "GLESv2"
[18:19] <kdub> anyways, just built successfully, thanks greyback
[18:19] <racarr> DOES IT RUN DOES IT RUN
[18:20] <greyback> dramatic pause!
[18:21]  * kgunn passed out from holding breath
[18:21] <kdub> no, that would be too easy :)
=== mmrazik|afk is now known as mmrazik
[18:24] <greyback> Here are higher-level instructions if anyone wants to run unity+mir on their phone: http://studio.sketchpad.cc/gmY0M6iqeh
[18:25] <greyback> they're a bit messy still, yes. But they work
[18:38] <racarr> wait so it does run?
[18:40] <kdub> no, just exits
[18:40] <kdub> just starting to read through the code though
[18:40] <racarr> the ./run script hides the segfaults
[18:40] <racarr> becaues it does openvt
[18:40] <kdub> still don't know exactly what this thing is
[18:40] <racarr> so I got rid of the openvt bit and ran it in GDB
[18:41] <racarr> to debug in GBM
[18:42] <racarr> greyback: Instructions seem to be missing building platform api + qtubuntu?
[18:42] <greyback> racarr: I've them packaged
[18:42] <greyback> racarr: .deb files are in my chinstrap
[18:42] <racarr> greyback: Oh! I see
[18:43] <racarr> So you're one of those people who does things the right way huh?!?!
[18:43] <racarr> :p
[18:43] <greyback> yes, very much so
[18:43] <greyback> :D
[18:44] <racarr> great. thanks for writing this up :)
[18:45] <greyback> racarr: I'm guessing you plan to have separate platform-api-client and -server packages eventually. And same for qtubuntu?
[18:45] <racarr> greyback: seperate packages yes...I think the final structure is likely to be like
[18:46] <racarr> there is platform-api (which also contains bit like GPS, etc)
[18:46] <racarr> that can (dynamically?) load either mirserver, mirclient, windows forms
[18:46] <racarr> whatever
[18:46] <kdub> do i need the 'client' or the 'server' libqubuntu.so qpa plugin?
[18:46] <racarr> likewise qtubuntu should probably gain just a tiny amount of intelligence
[18:46] <racarr> kdub: server
[18:46] <greyback> racarr: got it. Makes sense
[18:46] <racarr> to auto select loading platform-api server/client
[18:46] <racarr> rather than maintain two qtubuntu
[18:47] <greyback> yep, totally
[18:50] <racarr> ADHD: I was wondering in the shower today. When do we need to have mir running under virtualized environments?
[18:50] <racarr> kgunn: ^
[18:50] <racarr> is it captured somewhere?
[18:51] <kgunn> racarr: what do you mean exactly ?
[18:51] <racarr> then also, I wonder if say porting xf86-video-vmware to a graphics platform is easy XD
[18:51] <racarr> kgunn: I mean, Mir should be able to run under vmware/virtual box
[18:51] <racarr> Has to be able to presumably
[18:51] <racarr> by 14.04
[18:51] <kgunn> racarr: ah, ok....
[18:51] <kgunn> yes
[18:52] <kgunn> and i would think that would be super easy (or am i a nut?)
[18:52] <kgunn> mm
[18:52] <kgunn> maybe not so...kernel stuff
[18:52] <racarr> well, the basics are easy
[18:53] <racarr> but like, in the X driver, they use special interrupts to map a framebuffer through the virtualization layer
[18:53] <racarr> or things like that
[18:54] <racarr> they have libGLs that talk to virtualbox/vmware which then use the host libgl
[18:54] <racarr> things like that
[18:54] <racarr> mode setting, they have special communication (I think it's weird interrupts again)
[18:54] <racarr> for mode interaction between the guest video driver and the virtualization host
[18:54] <racarr> etc.
[18:55] <kgunn> right....i don't know how it all works but i suppose the kernel itself is the virtual bit
[18:55] <kgunn> sharing hw resources like fb
[18:55] <racarr> It seems like (looking at xf86-video-vmware) this is kind of abstracted away a little.
[18:55] <kgunn> i would imagine so
[18:55] <racarr> i.e. they have funny functions like vmware_do_bla_bla that
[18:55] <racarr> handle the complex bit of
[18:55] <racarr> talking to the host and back and forth and all
[18:55] <racarr> but mir needs to be hooked up to it
[18:55] <racarr> if it really is that simple
[18:55] <kgunn> sure
[18:55] <racarr> it's kind of the same sized task as porting mir to SDL
[18:56] <racarr> (but made easier by that)
[18:56] <racarr> but, um
[18:56] <racarr> I just told you literally everything I know about this so
[18:56] <racarr> I don't have the confidence to say that :p
[18:56] <kgunn> :))
[18:56] <kgunn> well, don't get too distracted....other things on the horizon
[18:56] <kgunn> osk
[18:56] <kgunn> app data sharing
[18:57] <racarr> Yes of course, I am not going to work on it now
[18:57] <racarr> I just want to make sure it doesn't get lost
[18:57] <racarr> because I dont think we have a work item, but it's clearly expected
[18:57] <kgunn> lemme check
[18:58] <racarr> and I am worried it will have some secret bit like "actually there is this closed source bit that we have to convince them to modify because our semantics...."
[19:02] <greyback> racarr: probably daft question, but does the phone have concept of a vt?
[19:02] <greyback> "adb shell" is a proper VT?
[19:03] <kdub> greyback, unsure
[19:03] <kgunn> greyback: i think that's as close as it gets (adb)
[19:04] <kgunn> well...i take that back....there are some who got x working on android
[19:04] <kgunn> but it was really custom
[19:05] <greyback> Me too. I'm just curious is there's a way to run mir from ssh so that it works. It's easy on the desktop to create a new vt and run an app in it (openvt), but it would be handy to have something similar for phone
[19:07] <greyback> oh dumb-ass, I can just use "adb shell" itself from the pc
[19:07] <racarr> :)
[19:08] <racarr> not daft XD I was wondering the same thing
[19:08] <racarr> I realized today I don't really remember how terminals/vt/pty all work
[19:08] <racarr> I did in college at one point when I thought reading advanced unix programming was a fun thing to do in your spare time ;)
[19:09] <tvoss> racarr, got the unix bible on my book shelf :)
[19:09] <greyback> :D thanks guys, I'll expect the answer in the morning so :D
[19:10] <racarr> XD
[19:12]  * greyback eod
[19:12] <racarr> Goodnight!
[19:15] <racarr> Lunch for me, nexus 7 upgrade running as we speak so I will try unity mir android when I get back
=== racarr is now known as racarr|lunch
[19:18] <kdub> racarr|lunch, nex7 won't work with default hybris package
=== hikiko-uds is now known as hikiko
[19:32] <racarr|lunch> kdub: Mm right. Do packages to fix it still exist?
[19:32] <kdub> racarr|lunch, no
[19:36] <racarr|lunch> kdub: Can I build them?
=== francisco is now known as Guest83331
[20:19] <hikiko> hello
[20:19] <hikiko> question:
[20:20] <hikiko> I get this linked error when trying to compile my sdl branch:
[20:20] <hikiko> /usr/lib/libmirclient.so.0: undefined reference to `mir::protobuf::EventSequence::~EventSequence()'
[20:20] <hikiko> but when I grep
[20:20] <hikiko> I can't find any reference to EventSequence in the code
[20:23] <bschaefer> hikiko, hmm perhaps its generated at build time?
[20:23] <bschaefer> build/src/shared/protobuf/mir_protobuf.pb.h:1477:class EventSequence : public
[20:24] <hikiko> !
[20:24] <hikiko> I don't have that class
[20:24] <bschaefer> that could be a problem!
[20:24] <hikiko> wtf maybe I have to merge with a more recent version before I proceed :S
[20:25] <bschaefer> possibly, im at rev 687
[20:25]  * bschaefer pulled an hour or so ago
[20:25] <hikiko> yes :) I have to :)
[20:25] <hikiko> lol
[20:25] <bschaefer> :)
[20:25] <hikiko> thank you!
[20:25] <bschaefer> np!
[20:25] <hikiko> +see you tomorrow
[22:52] <kgunn> kdub: racarr|lunch good grief....i just now made my way thru the github shennanigans & have the proper branch
[23:00] <kgunn> kdub: so reading the install file....looks like the best thing to do is native compile this on the device ?
[23:00] <kdub> it probably is the easiest
[23:01] <kdub> it doesn't look like getting that working though is trivial though
[23:32] <racarr|lunch> kgunn: You are trying the seven too?
[23:33] <racarr|lunch> It didn't work for me
[23:33] <racarr|lunch> But I might have done something wrong
[23:33] <racarr|lunch> oh whoops
=== racarr|lunch is now known as racarr
[23:38] <RAOF> Late lunch? :)(
[23:45] <racarr> kdub: linker.c:661| WARNING: `/system/lib/hw/gralloc.tegra3.so` is not a prelinked library
[23:45] <racarr> meaningful?
[23:45] <kdub> racarr, no
[23:45] <racarr> :(