[00:00] kdub_: I've not run that recently, but isn't that expected behaviour now that input is turned on? [00:01] oh, is it? i don't have a problem with --enable-input=true with mir_demo_server on 673 [00:01] Wasn't ctrl-alt-backspace meant to kill the server? [00:04] well, we should always clean up and exit on sigterm :) [00:08] oh, i see what's going on... just a problem with the demo [00:08] not sure about 4.8 though [00:09] Well, yeah, but we're not going to get SIGTERM if we eat the input (like we should!) [00:12] in this case, i think the problem is that we have a sigterm handler registered (in the server) but the threadloop for mir_demo_inprocess_egl doesn't respond to sigterm [01:33] how often do y'all sync from the debian sid repos? [01:53] We, as in ubuntu-mir, don't. [01:53] tehcloud: But it's about daily for Ubuntu as a whole, last I checked. [02:22] robert_ancell: RAOF: saucy packages will be built on next landing [02:22] thomi, thanks [02:23] nw [02:48] hey guys, dependencies withing ppa:~mir-team/staging are all fouled up: unity-system-compositor requires libmirserver0 (= 0.0.2bzr664raring0) but only 0.0.2bzr673raring0 is available [02:50] thomi ^^^: I thought we had unity-system-compositor autobuilt whenever mir got built? [02:51] RAOF: Francis is working on that this week - he promises a hacky solution by the end of the week, sorry [02:51] but... bregma's problem sounds different [02:52] do the dependencies really need to be that strict? [02:52] libmirserver0 is C++, and we don't do anything to ensure ABI. [02:52] It does indeed need to be that strict. [02:53] I can manually re-land u-s-c if you like? [02:53] It's probably not worth it? It'll break the next time mir lands. [02:53] this is true, but if bregma needs it... [02:54] Eh, if it's easy. [02:55] well, it's bedy-bies time for me anyway, so I'll check again in 8 hours or so [02:56] RAOF: or anyone else: do you know what's required to get mir running on the phone? I can run it, but don't see anything on the screen. Is this a case of the phone shell rendering over the mir server, or do I need to do something else? [02:56] the mir_demo_server, that is [02:57] thomi: Last time I followed the instructions (stop surface flinger, chroot in, run mir) they worked? [02:58] RAOF: oh... there are instructions? [02:58] I guess I shoulda read the README [02:58] thomi: http://unity.ubuntu.com/mir/ ? [02:58] http://unity.ubuntu.com/mir/using_mir_on_android.html is what I followed last. [02:59] Of course, you *could* just read the documentation in-tree :) [02:59] Ooops! Documentation skew! [02:59] hmmm? [03:01] RAOF: can you many any sense of this? http://pastebin.ubuntu.com/5646605/ [03:02] thomi: The documentation talks of libmirclient-demos, and a “mir” binary. ☺ [03:02] I mean, I can see what went wrong "cannot create fb display", just not why [03:02] oh, yeah [03:02] Paging kdub_! [03:02] What hardware are you trying it on_? [03:03] the maguro [03:03] surely kdub_ is asleep by now [03:06] Probably. [03:06] The maguro being... Galaxy Nexus? [03:06] yeah [04:01] thomi, what is the autolanding failure in https://code.launchpad.net/~raof/mir/separate-graphics-buffer-and-display/+merge/161939? [04:01] " Approved revid is not set in launchpad (maybe a permission problem?)." [04:33] * thomi looks [04:34] robert_ancell: I believe it's that the MP is either set to WIP, rather than needs review, or that the approved revision is different to the latest revision in that branch [04:35] robert_ancell: if it fails when the MP is set to approved, please let me know [04:35] thomi, ok, will do (it wasn't WIP before RAOF just changed it) [04:35] ahh ok [04:36] my guess is it was approved, and then an additional revision was pushed to the branch [04:53] RAOF, "The EGLStream conceptually operates as a mailbox. When the [04:53] producer has a new image frame it empties the mailbox (discards [04:53] the old contents) and inserts the new image frame into the [04:53] mailbox. The consumer retrieves the image frame from the mailbox [04:53] and examines it. When the consumer is finished examining the [04:53] image frame it is either placed back in the mailbox (if the [04:53] mailbox is empty) or discarded (if the mailbox is not empty)." [04:53] does any of that sound like how you would use a mailbox? [04:54] No :) [04:54] But the behaviour *we* want is in EGL_stream_fifo [04:57] RAOF, and in this case an app is a EGL stream producer and the display server is an EGL stream consumer? [04:57] Indeed. [04:57] Although that's not necessarily exposed to the client. [04:59] "3.10.2.1 No way to connect consumer to EGLStream" ok, so this is a half-spec :) [05:00] Right. [05:00] All the producers and consumers are their own specs. [05:01] You're after http://www.khronos.org/registry/egl/extensions/KHR/EGL_KHR_stream_consumer_gltexture.txt for the consumer side, and http://www.khronos.org/registry/egl/extensions/KHR/EGL_KHR_stream_producer_eglsurface.txt for the producer side. [05:02] Oh, and http://www.khronos.org/registry/egl/extensions/KHR/EGL_KHR_stream_cross_process_fd.txt [05:03] RAOF, and if you have this stuff you don't need the DRM/android buffer IDs that we currently pass? [05:03] Right. They become the buffer passing mechanism. [05:03] that would simplify things down a lot [05:04] Yeah, it'd be pretty neat. [05:04] And currently we require Mesa because of the current buffer passing - with EGL you can use any standard EGL implementation (as long as it has the extensions) [05:05] Yup. Although you also need the ability to do a little bit of extra hooking into the EGL implementation to get a Mir EGL platform. [05:06] RAOF, what sort of stuff? [05:06] AFAIK that's a bit like what Android does. [05:07] Well, at base you need to be able to pass a MirSurface into eglCreateWindowSurface. [05:08] ok [05:09] RAOF, GLX works currently on Mir right? [05:09] (is that both GLX using DRI and traditional GLX?) [05:10] On the mesa drivers, yes. [05:11] I'm not sure what you mean by ‘traditional GLX’. If you mean indirect rendering, then yes. [05:12] yes [05:12] so this is what is required for GLX on android drivers? [05:13] This is a part of what's required, yes. [08:52] thomi: The docs on http://unity.ubuntu.com/mir/ seem behind the times. How often are they pushed? === mmrazik is now known as mmrazik|lunch === mmrazik|lunch is now known as mmrazik === mmrazik is now known as mmrazik|afk === alan_g is now known as alan_g|lunch [13:22] alan_g|lunch: hey...you "surfaced" :) [13:22] quiet day w/o alf === mmrazik|afk is now known as mmrazik === alan_g|lunch is now known as alan_g [13:30] kgunn: yes. Things seem to have been quiet all week. === francisco is now known as Guest49017 === alan_g is now known as alan_g|tea [15:05] good morning [15:08] thomi, ping me when you get up, we'll get your maguro running === alan_g|tea is now known as alan_g [15:21] alan_g, do you remember why we had to disable death tests on android? i'd guess we don't have to do that anymore [15:22] kdub_: There was a problem with the way that gtest forked. I'd guess the same [15:22] kdub_: shall I try enabling them? [15:24] i tried them on the device yesterday, seem to be ok [15:25] cool. (I imagine they've been fine since we junked the NDK) [15:39] kdub_: I'm not sure where to start with the "Composition bypass - Android" task I accepted. As the team Android specialist can you get me started? (If you're busy or need to think about it we can aim for a hangout tomorrow.) [15:41] alan_g, http://bazaar.launchpad.net/~mir-team/mir/trunk/view/head:/src/server/graphics/android/fb_swapper.h#L34 [15:42] basically, we need a BufferSwapper that implements the swapping algorithm, and a path for a client to request composition bypass [15:42] i can hop on a hangout though [15:56] kdub_: alan_g thanks for jumpstarting on comp bypass... [15:57] its so key for mobile === mmrazik is now known as mmrazik|afk [16:50] kdub_: I've updated https://code.launchpad.net/~alan-griffiths/mir/support-multiple-inheritance-in-config/+merge/163156 [16:54] alan_g, still looks good, i'm unsure why changing {} to () helped too though... [16:56] kdub_: I *think* it is a compiler bug (as clang agrees with us) but not completely sure there isn't a corner case that requires a copy constructor. [17:00] kdub_: did you ever ask thomi about the updates docs on http://unity.ubuntu.com/mir/ (they seem to miss a bunch of last week's changes.) === alan_g is now known as alan_g|life [17:01] alan_g|life, yes, i think they're update regularly, but there's some sort of caching somewher [17:05] Hi all kind of in and out today still due to moving [17:05] But tomorrow back to normal [17:05] and by normal I mean I have a new house and office without lots of crazy (all love intended) housemates so hopefully better than normal :) [17:08] racarr_, cool [17:30] kdub_: thanks, I'll be patient (for now) === jono is now known as Guest61463 [20:47] hey guys - will look into the mir docs not being uploaded today. [20:48] kdub_: I'd love to get my maguro up and running, btu I'm snowed under with distractions right now - how much longer before your EOD? [21:00] thomi, a bit of time :) i flashed my maguro today with 119, mir-staging ppa. mir works, but make sure surfaceflinger has never started [21:00] just to be safe :) [21:01] kdub_: ok, I'll start flashing mine now, and maybe when I'm done I can ping you agaibn [21:01] *again [21:01] like, 'adb shell stop' will stop the surfaceflinger process, but i don't trust it to clean up fb resources [21:01] so what's the better way? [21:02] well, you can muck around with the android startup files, i usually just mv /system/bin/surfaceflinger somewhere that the android init scripts can't find it :) [22:01] hey guys - I wonder if I could get someone to review this trivial MP please? https://code.launchpad.net/~thomir/mir/fix-documentation/+merge/163231 [22:01] it's need to fix the documentation upload issues [22:19] kdub_: I'm still getting the same error,and I don't seem to be able to move the surfaceFlinger binary since it's on a read-only filesystem [22:19] am I missing something? [22:21] adb remount [22:22] Its probably ok if sf and all sf clients are dead [22:27] that worked better - rebooting... [22:31] kdub_: ummmm... something bad happened. now it doesn't boot? [22:31] hmmmm [22:31] * thomi tries pulling the battery [22:31] Well the shell wont boot up [22:31] But the adb connection should work [22:32] sure, it wasn't doing anything though. Pulling the battery seems to have fixed it [22:32] for a second there I thought I'd bricked it somehow [22:32] Nah, they're tough to brick :) [22:34] well, mir_demo_server seems to run, so I'm one step closer :) [22:35] wooo! glmark running on the phone [22:35] Yay! [22:35] kdub_: seems like it's limited to 60FPS, just like the desktop though :( [22:36] I get 59 or 60 FPS for everything :( [22:36] Right, it is swapinterval 1 [22:36] I'll pretend I understand... can you fix it easily? [22:37] My next task is a bit of performance analysis... [22:37] Probably will have to add swapinterval 0 support as part of that [22:38] ok === francisco is now known as Guest79778 [22:45] robert_ancell: got a second? [22:45] thomi, sure [22:46] robert_ancell: I wonder if you'd mind reviewing my MP above ^^ ? [22:46] thomi, sure [22:46] I want to see what's causing the mir docs to get out of date on the server - I suspect it's the caching proxy IS uses rather than anything our side [22:46] thanks [22:49] thomi, what do we need to do to retrigger CI on https://code.launchpad.net/~mterry/lightdm/autologin-in-background/+merge/162670? I tried the "click here to trigger a rebuild" but it didn't do it [22:49] * thomi checks [22:50] robert_ancell: the link worked for me. It's a bit awkward though - if you're not logged in to the jenkins server it'll ignore you [22:50] how rude! [22:50] robert_ancell: it's running now though: http://s-jenkins:8080/job/lightdm-ci/31/console [22:50] yeah [22:51] guess what? jenkins is the worst CI server. evar [22:51] (IM not so HO) [23:06] kdub, woah, massive patch :) [23:13] Medium size :)