[01:29] duflu, ping [01:29] bschaefer: Hello [01:30] duflu, Hey, so before i head off for the night i wanted to about racarrs comment on my v/h scroll branch [01:30] https://code.launchpad.net/~brandontschaefer/mir/lp.1233089-fix-v-h-scroll-v2/+merge/189750 [01:30] duflu, increasing the MAX_INPUT_POINTER_SIZE just incase we run into this problem again... [01:32] though it wont help in changing the size of point_cords, but the more room it gives us a bit more options... [01:33] bschaefer: Padding makes sense. But more than 10 pointers does not :) [01:33] duflu, very true, its at 16 atm, which if pretty large [01:33] err yeah [01:33] duflu, how would be pad it then? Just add some dummy floats? [01:34] bschaefer: Yeah some dummy ints inside pointer_coordinates, and more on the end of the struct [01:34] duflu, do you have a number in mind :) [01:34] named "unused" etc [01:35] bschaefer: It's not the end of the world if we run out. So no more than 8, max [01:35] 4 in each? [01:35] Maybe [01:35] well ill add that, and i can push changes tomorrow if more discussions happen [01:37] duflu, pushed :), and thanks for helping with that MP (i've a better understanding of ABI breakage..and more ways around now!) [01:37] bschaefer: Thank you too [01:38] np! [05:15] mhall119, you have got libhybris or libhybris-dev installed on your system. Please just remove it. [05:33] tvoss: bug 1232962 :( [05:33] :) [05:33] bug 1232962 in libhybris (Ubuntu) "hybris packages should fail to install if an Android filesystem is not detected" [Wishlist,New] https://launchpad.net/bugs/1232962 [05:33] duflu, yup [06:24] mlankhorst: Any movement on bug 1195811? [06:24] bug 1195811 in linux (Ubuntu) "nouveau: Abnormally high FPS and tearing (no vsync) with nouveau" [High,In progress] https://launchpad.net/bugs/1195811 [06:35] duflu: I poked upstream about it, they'll look at it, but power management is a higher priority :/ [06:42] mlankhorst: Did you mention it directly affects power usage? :) [06:43] well if you grab the patches from my tree you'll have a lot less issues tbh [06:44] http://cgit.freedesktop.org/~mlankhorst/linux/log/ just this one "drm/nouveau: enable vblank support on nv50 and fermi cards" [06:44] on top of recent kernel [06:46] mlankhorst: Actually I'm not using nv-anything right now. Just curious how it's going [06:50] tsk:P [07:48] tvoss: Confirmed fix-r1049-regressions improves performance on Nexus 4 too [07:48] Slightly. Most noticeably in egltriangle -f [07:49] duflu, ack. did you try with unity8? [07:49] tvoss: No [07:49] duflu, can you please confirm? I can confirm that it improves performance on maguro [07:50] tvoss: Does unity8 have any special incantations or just run the binary? [07:51] duflu, just boot the phone I would even say [07:51] tvoss: Not sure how to package x-compiled packages... [07:52] -packages +binaries [07:52] duflu, why not just flash a current image? :) [07:52] tvoss: I just did [07:59] mlankhorst: I am going to resume nested Mir work today. Do the i965 experiments in the egl-platform-mir-egl-image-i965-experiment have any effect, or should I just revert them? [07:59] mlankhorst: i.e., https://github.com/afrantzis/mesa/commit/b325ffb333875477a83e42bbfc595bb33e4df19a [07:59] This is weird. How does the unity8 binary not link even indirectly to libmir*? [08:00] duflu: it uses mir through Qt [08:02] alf_: Which library? ldd usually shows all indirectly loaded so's too [08:02] duflu: it's a Qt platform plugin, loaded at runtime [08:03] Oh [08:03] cool [08:03] duflu: I don't remember the exact path of the plugin, probably somewhere under /usr/lib/.../qt5/.. [08:07] alf_: all my patches should be applied on top of that branch [08:07] but the i965_context.c is needed probably, dno, it's more sane anyway [08:09] mlankhorst: ok, I will keep the changes for now, and we can figure it out later [08:14] alan_g, alf_, tvoss: I think we lack enough common ground to bother having this call today. Probably best to reject the proposals, move on, and ideally someone propose some nice alternatives [08:18] duflu: (alf_ tvoss) - I'd like to reach a consensus rather than leaving a disagreement simmering. But if that's how you feel... [08:18] duflu, no time for keeping the discussion around any longer. The argument of cleanup has been raised repeatedly in previous discussions as preventing us from moving forward rapidly. Thus, we need to resolve the issue *today* [08:22] tvoss: OK, figured it out. But we bumped the server ABI :P [08:32] duflu: hangout? [08:32] alan_g, trying [09:38] tvoss: Confirmed the branch (built against saucy Mir) works nicely with Unity8. It's at least as smooth as the released version [09:38] duflu, ack. got a maguro, too? [09:38] tvoss: No. I wish... sometimes [09:39] tvoss: I remember kdub saying when that version was first proposed (as the "switch" branch) he said it helped maguro. But that was way back [09:40] duflu, ack [09:40] alf_, can you see if your question in https://code.launchpad.net/~vanvugt/mir/fix-r1049-regressions/+merge/191776 is addressed? [09:42] tvoss: looking [09:42] alf_, thx [09:44] duflu: Is the lock/unlock mechanism in the test fragile? That is, could the test fail if we are unlucky enough with the timings? [09:45] alf_: It's the other way round... the lock/unlock *help* the test to fail more often when the fix is absent :) [09:48] alf_: did it work? :p [09:51] mlankhorst: sorry, haven't got to it yet, will do so after lunch [09:51] duflu: ok [10:03] alan_g: Where a header moves from public to "protected" (libmirserver internal), can we #include "../component/" for now to avoid logic changes on this first step? [10:04] duflu: we *shouldn't* need that. Where does it happen? [10:04] alan_g: Where shell::Surface calls surfaces::Surface [10:06] duflu: I think that specific case is best handled by changing the logic - we can live with one hard to shift class for a little longer. [10:07] I think shell::Surface can become base class for surfaces::Surface with very little effort. [10:07] alan_g: If you'd block on that (and the other occurrence render_surfaces) then I'll defer the whole proposal. It requires more time than just tonight [10:08] duflu: ack - I know that one is gnarly - I'll try to get to it this afternoon [10:09] alan_g: BTW their only relationship right now is that shell::Surface points to the surfaces::Surface [10:09] According to me scrawlings [10:10] In fact, that relationship singularly divides the graph in two [10:10] duflu: Yeah, but it is a very trivial proxy [10:55] alan_g: Two privatize MPs updated. The other is hidden until further notice. [10:56] Ironically, the one most approved has to be deferred the longest [10:56] duflu: ack, will look [11:36] does anyone know if libmirserver is leaking some boost objects? [11:36] (we have some mismatch due to boost) [11:38] didrocks, libmirserver does not expose boost via its public interfaces as far as I know [11:39] tvoss: see #ubuntu-eng-ci about templates [11:39] ooopss #ubuntu-ci-eng rather === greyback is now known as greyback|food [11:43] didrocks, not seeing a discussion there and don't have backlog :) [11:43] terminate called after throwing an instance of 'boost::exception_detail::clone_impl >' [11:43] what(): Could not open hardware module [11:43] 13:36:20 xnox | didrocks: templates do leak ABI/API mismatches, via transitive dependencies. [11:43] 13:36:31 xnox | didrocks: but that's "standard C++" feature. [11:43] tvoss, thats what i get in the unity8 log on the latest trustyx image [11:44] ogra_: trustyx? sounds, sexy ;) [11:44] ogra_, oh well, exceptions are leaked, yes [11:44] didrocks: Nothing has been done (yet) to prevent boost symbols being exported [11:44] tvoss, while unity8 was rebuiolt for boost where seem to be other üplaces wheer a rebuild is needed too [11:44] ok, let's try a rebuild of platform-api and unity-mir then [11:44] s/where/there/ [11:44] Is that what you mean? [11:45] alan_g: yeah, mir was transitionned to new boost, but we have that exception above [11:45] so can be that unity-mir and platform-api (mirserver deps) needs a rebuild as wll [11:45] well* [11:45] didrocks, my bad, I forgot about the exception [11:45] no worry, let's see if it helps :) [11:46] tvoss: the question is if the exception is failing due to boost ABI missmatch, or the exception is real and unity8 is failing to open hardware module =) [11:46] xnox: TBH, I would prefer #1 :) [11:46] didrocks, i'll refrain from doing a local build then, tell me if the debs are in the PPA and ill test them, i'll leave my phone in the broken state for that [11:46] ogra_: I'm just building on my phone [11:46] oh. ok [11:47] ogra_: well, if I can boot [11:47] ogra_: I have the android bot spinning when flashing [11:47] thats normal [11:47] it will reboot fine and leave you with the google logo then [11:47] (and adb) [11:49] ogra_: hum, are you a nm expert? [11:49] ogra_: I guess I need network and connect to wifi without any ui :p [11:50] oh,. you flashed with --no-backup ? [11:50] try phablet-network [11:51] (with the phone connedcted to a lappie) [11:51] ogra_: trusting you, launching random commands :) [11:51] ogra_: I didn't use --no-backup, it's just nm didn't come up [11:51] oh [11:52] Network setup complete [11:53] # ifconfig [11:53] lo Link encap:Local Loopback [11:53] and only lo [11:53] Network connection failed to become active. [11:53] ah :) [11:53] ok, starting the process by hand [11:53] good :) [11:53] works ? [11:53] yep :) [11:54] good [11:54] nice trick, thanks ogra [11:54] * didrocks apt-get source and try === alan_g is now known as alan_g|lunch [12:16] ogra_: did you just try sudo start unity8? [12:18] hmm, no [12:19] phablet@ubuntu-phablet:~$ sudo start unity8 [12:19] [sudo] password for phablet: [12:19] start: Unknown job: unity8 [12:19] heh [12:19] falls back to the system init it seems [12:19] ogra_: you do you start those upstart jobs? [12:20] ogra_: just so that I'm running the same way as you, how did you start it? [12:20] you just look at usptart logs? [12:21] hum, I'm not even getting to the user session (there is no .local/.cache) [12:21] sudo -i phablet -i [12:21] start unity8 [12:21] thats all i do [12:22] and then check the upstart log [12:22] ogra_: yeah, I have no session started [12:22] (or see it come up) [12:22] # sudo -i phablet -i [12:22] initctl: unable to determine sessions [12:22] oh weird [12:22] and no .* [12:22] so try starting lightdm first [12:22] as root [12:23] still the same :/ [12:23] (but lightdm started) [12:24] ogra_: so, for you, the session did start at least? [12:24] lightdm did apparently [12:25] (upstart claimed it was running when i tried to start it) [12:25] instead of poking around manually, did you just try a reboot with the rebuilt packages ? [12:26] didrocks: where are you testing unity from? such that I can check which boost's it's using. [12:26] didrocks: and if needed revert those r-deps to boost1.53 until we know it works. [12:26] xnox: I guess there is another issue TBH [12:27] xnox: I started lightdm and not sure I have a seat [12:27] didrocks: that can't be good =/ [12:27] didrocks, the log should tell you [12:28] # less /var/log/upstart/systemd-logind.log [12:28] New session c3 of user phablet. [12:28] so, I have one [12:28] [+0.34s] DEBUG: Seat: Session authenticated, running command [12:28] [+0.34s] DEBUG: Registering session with bus path /org/freedesktop/DisplayManager/Session0 [12:28] [+0.35s] DEBUG: Session pid=884: Running command /usr/sbin/lightdm-session ubuntu-touch-session [12:28] [+0.36s] DEBUG: Session pid=884: Logging to .xsession-errors [12:28] but no user session started it seems [12:28] tail /var/log/lightdm/lightdm.log [12:28] (is what that was) [12:28] * ogra_ wonders if he sees an old log [12:29] root@ubuntu-phablet:/# ls -l /var/log/lightdm/lightdm.log [12:29] -rw------- 1 root root 1713 Oct 22 13:27 /var/log/lightdm/lightdm.log [12:29] no, that was about the time i flashed [12:29] so its from the current session (i didnt reboot yet) [12:29] xnox: it's image 2 in trusty-proposed btw [12:29] didrocks: ack, let me flash that. [12:30] thanks === greyback|food is now known as greyback === alan_g|lunch is now known as alan_g [12:55] * mlankhorst pokes alf_ with a testing stick [12:55] mlankhorst: it works fine so far :) [12:56] alf_: is that nested+gbm? [12:57] alan_g: yes [12:57] Progress! \o/ [13:07] thought so [13:08] the lookup extension to lookup a bo from inside the screen context is still broken, but I have no idea where it's used [13:13] EGLImageTargetTexture2D, EGLImageTargetRenderbufferStorage probably [13:15] tvoss: should I remove all libhybris-* packages? [13:15] mhall119, yup, ideally [13:15] ok [13:21] hm probably not [13:53] alf_: raii fixed. ;) (tvoss spotted the obvious error) [13:53] alan_g, \o/ [13:55] alan_g: tvoss: great! === alan_g is now known as alan_g|afk === alan_g|afk is now known as alan_g === alan_g is now known as alan_g|tea === alan_g|tea is now known as alan_g === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [15:22] alf_, hello! You're working on nested Mir mode? === pete-woods is now known as pete-woods-away [15:32] whoops, got disconnected. alf_, if you replied I missed it. [15:33] mterry: you didn't and he is [15:33] alan_g, cool, I have things to discuss then, when he gets back [15:33] * mterry is paranoid with this bad connection [15:34] time that we get reliable tethering with ubuntu phone ;) [15:34] Heh === dandrader is now known as dandrader|lunch === pete-woods-away is now known as pete-woods === alan_g is now known as alan_g|EOD === dandrader|lunch is now known as dandrader [17:31] racarr, ping [18:28] dandrader: what did you need ? [18:30] kgunn, talk about https://code.launchpad.net/~dandrader/mir/uniqueTouchIds/+merge/191695/comments/441723 [18:31] dandrader: ah...kind a need racarr for that [18:38] Hi....yes sorry [18:38] Kind of sick [18:38] *stares blankly at screen* [18:47] racarr, so, wanna have a chat about that branch? [18:51] dandrader: Not now....sorry the longer I stay out of bed the more sick I am feeling and I think I am going [18:51] back to rest [18:51] racarr, ok, np === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader === sam113101 is now known as SAM113101