[02:12] Hm. I really need to fix keymapping before using the unity8 desktop session :) [05:01] RAOF: you and your dvorak.. [05:02] :P [06:47] Bah. [06:47] C++, why you no map (list)? [06:58] RAOF: Wha/ [06:58] ? [06:58] duflu: I want to iterate over the result of applying a function elementwise to an existing iterator. [06:59] Without constructing an unnecessary container. [06:59] This would typically be called “map’ [06:59] Or possibly “apply” [07:00] RAOF: Temporary results in a loop you don't wish to modify to accept an apply functor? [07:01] * duflu feels dirty for knowing what that means [07:04] I'd like to construct a list of SharedLibraries from a directory iterator [07:05] Now, I can do this by iterating over the directory, constructing the SharedLibrary, and shoving it into the list. [07:05] But I'd *prefer* to just pass an iterator to the list constructor. [07:11] RAOF: I think most real-world problems involve picking and choosing from lists, and not a blind 1:1 mapping. As such, there's little re-use value in C++ providing that for you. [07:11] Besides, C++ is already massive in spec [07:12] Well, that's very simple [07:12] map(filter(thing, predicate), functor) :) [07:12] Basically I wish iterators were substantially more powerful, because it's really awesome to be able to operate on them. [07:13] I wish iterators had a common base class and weren't a pure pattern [07:14] hey, anyone aware of efforts to make freeglut run on mir? I think its just the input part that uses XInput. The rest seems to be ok already. [07:14] * mzanetti was trying to compile VoltAir on the phone [07:16] mzanetti: I remember the simplicity of GLUT fondly. Would be nice but not aware of any efforts [07:17] mhm... [07:17] duflu: _Yes_ === renato is now known as Guest25241 [08:23] alan_g: No, my blocking comments weren't addressed. On the other hand, https://code.launchpad.net/~raof/mir/fix-unnecessary-virtual-package/+merge/227862 is simple and resolves them :) [08:24] RAOF: sorry. I got the incantation I used from both fginther and sil2100 so I let the majority rule. [08:25] alan_g: Yeah, that's totally reasonable. [08:25] I clearly need to poke sil2100 and fginther with a cluebat :) [08:30] RAOF: the more I get to understand the debian/control file the more I feel we've got wrong. Is that normal? [08:31] Not normal, no. [08:31] Hm. Now that I look at it... [08:34] * alan_g goes to look at a problem he does understand [08:35] Excellent plan! [08:36] * RAOF → dinner === tvoss is now known as tvoss|test === tvoss|test is now known as tvoss === alan_g is now known as alan_g|afk [11:40] anpok: hey just checking in, any progress with https://bugs.launchpad.net/mir/+bug/1346952 [11:40] Launchpad bug 1346952 in Mir 0.5 "[qtcomp] Random crash in Mir input when running AP tests: [terminate called after throwing an instance of '...' what(): assign: File exists] when constructing a mir::AsioMainLoop::FDHandler" [Critical,New] === alan_g|afk is now known as alan_g === alan_g is now known as alan_g|lunch [12:06] hey folks [12:15] alan_g|lunch, hey, I'll join you for the interview with Jawad, sit in on your part and then do mine when you'll be relieved ;) === greyback is now known as greyback|lunch [12:27] greyback|lunch: debugging it - so yes reproducing the issue.. === pete-woods1 is now known as pete-woods|away [12:57] Saviq: ack === alan_g|lunch is now known as alan_g === greyback|lunch is now known as greyback === dandrader is now known as dandrader|afk === pete-woods|away is now known as pete-woods === dandrader|afk is now known as dandrader [14:31] Good morning! === alan_g is now known as alan_g|tea === alan_g|tea is now known as alan_g [14:44] racarr_, evening :) [14:44] I need some help compiling mir [14:45] groot_: ok! Lets give it a go! What are you trying so far? [14:46] racarr_, I'm trying to follow the cross-compile section of this http://unity.ubuntu.com/mir/building_source_for_android.html [14:47] just changed saucy to utopic [14:48] also downloaded mir tarball from here http://bazaar.launchpad.net/~mir-team/mir/0.5/revision/1792?start_revid=1792 [14:50] hope it's ok. I'm now confused with cross-compiler. Shoul I try install g++-arm-linux-gnueabihf/utopic ? [14:51] Is your desktop trusty? [14:51] I'm not 100% sure you can crosscompile from trusty to utopic [14:51] kdub_: ^? [14:51] unfortunately yes [14:51] hmm [14:51] doesn't work as well as you'd like it to sometimes [14:51] I dunno maybe someone will say they have dsone it and it works but my guess i\s no...I didn't think you could install the utopic G++ [14:52] groot_: If you don't want to upgrade your system...probably building on the phone itself is your best option [14:54] racarr_, wouldn't that take more time ? Also how can I compile in phone (native compile?)? Cross-compile seemed easier. [15:00] kdub_, how can I compile now? I want to test your fix :) [15:00] its more of an experiment [15:01] but native on-the-phone compile is slow, but you don't have to set up cross compiling [15:01] groot_: rCompile in phone../.isn't the fastest lol [15:01] from the phone, you first have to apt-get build-dep mir [15:01] then you just get the mir build directory and build it like normal [15:01] mkdir build; cd build; cmake .. -DMIR_PLATFORM=android; [15:01] make [15:01] oh [15:01] standup [15:02] racarr_, I don't have net connection in phone :( [15:03] racarr_: question on Mir's client buffer queue scheme: say client is starting up and Mir allocates it 3 buffers. Client draws its first frame and swaps. Does the compositor render that frame, or the 2 other yet-to-be-drawn-into frames have to be rendered first [15:04] greyback: The compositor should render that frame I think [15:04] the only requirement is that there is always a buffer available for the compositor [15:04] groot_: Hmm presumably we can get it for you with [15:05] adb or something [15:05] racarr_: ok ta [15:06] racarr_, I'll do it. Give me the commands serially :D [15:07] groot_: I am just thinking...there is no obvious waya (to me) gimme just a minute...I need to listen to some other stuff too lol [15:07] racarr_, no problem. Take your time. [15:13] groot_: I dont really know...there must be some trick to get USB networking working [15:13] but I dont know it or the magic incantation to reveal it on google :p [15:14] so I can't really suggest anything better than downloading the deps and copying them over to phone... [15:14] you need a net connection lol [15:14] No wifi? :( [15:14] galf_: how did you switch to 4.9 with arm-linux-gnueabihi-g++ [15:15] alf_: ^ :) [15:17] anpok: added '-4.9' to cmake/LinuxCrossCompile.cmake [15:17] ah :) [15:21] racarr_, ohh :(. I don't know whether wifi is running or not. Is there a way I can access it through terminal ? I use router in my house so connecting will not be a probelm. [15:22] groot_: if you run phablet-network on the host system it will copy over your [15:22] wifi config to desktop [15:22] if you are using that [15:22] if not...you will have to use wpa_supplicant or some such to connect to wifi via terminal [15:22] greyback: right it will draw that latest ready frame [15:22] connecting to wifi on desktop then using [15:23] phablet-network sounds easiest [15:23] AlbertA: thanks for confirming [15:23] greyback: it will not wait to fill the queue [15:29] racarr_, I'll try now and let you know. [15:32] groot_: :) [15:32] lol....46 files modified for touchspots [15:32] that... [15:32] doesn't seem quite right [15:37] racarr_, I don't use wifi in my desktop, desktop has broadband connection which is connected through router. So, there'll be config, right ? === dandrader is now known as dandrader|lunch [15:50] groot_: Yeah... [15:50] you will have to google like...connecting wifi terminal, sort of stuff :p I don't remember the exact options, etc [15:50] if you have wifi encryption you will need to use [15:50] wpa_supplicant [15:50] its pretty easy [15:57] groot_, you want nmcli [15:58] oh that sounds true :p [16:00] tvoss, how to use it ? [16:01] nmcli --help gives you an overview [16:01] groot_, should be something nmcli device wifi connect WIFI_NAME [16:02] dednick, When I pass the FD to the helper I set MIR_SOCKET to "fd://%d", right? [16:02] tvoss, alright I'll give it a try. [16:03] tedg, yup [16:04] tvoss, Hmm, okay. [16:04] Is there a way to get QtUbuntu to give more info than this: QUbuntu: Could not create application instance [16:10] tvoss, connected to wifi, but speed is slow :) [16:11] racarr_, how to compile mir? please give me the commands serially :) [16:11] tedg: passing to the trust helper? or the prompt provider? [16:12] groot_, I cannot help with the speed [16:12] groot_, in the mir root directory, call dpkg-buildpackage [16:12] groot_, that likely will fail with a list of missing build dependencies that you have to install [16:12] tedg: you need to pass the trusted socket to the helper, and the new_fd_for_prompt_provider sockets to the providers. [16:12] dednick, Yes, to the prompt provider. [16:13] dednick, So passing that socket to the provider as "fd://%d" [16:13] tedg: yup [16:15] tvoss, where's the mir directory? It's not in the root directory. [16:15] groot_, you have to branch mir by calling bzr branch lp:mir [16:16] tvoss, I've the files in pc. Push them will do ? [16:16] yeah [16:16] groot_: ^ [16:17] then; cd ; mkdir build; cd build; cmake .. -DMIR_PLATFORM=android; make [16:17] wait patiently [16:17] :) [16:19] dednick, Good, know any reason that QtUbuntu couldn't create an application instance then? [16:22] racarr_, alright :) [16:25] racarr_, this error happened http://paste.ubuntu.com/7842738/ === dandrader|lunch is now known as dandrader [16:31] tedg: check unity8 logs? [16:32] anything saying rejected? [16:32] tedg: did you start the trust helper with trusted socket? [16:32] No, it's not saying rejected. [16:33] Think I must have the wrong number. [16:36] racarr_, I think I've to install 14.10 on desktop :P [16:37] groot_: It looks like you forgot the apt-get build-dep mir part [16:39] racarr_, you're right!!! I forgot you mentioned that. I'll give this another go. [16:42] brb in ~20 [16:52] racarr_, still got error " You must put some 'source' URIs in your sources.list". List doesn't contain the package === alan_g is now known as alan_g|EOD [17:13] groot_: apt-get update [17:13] should take care of it [17:18] racarr_, OK, I'll try again tomorrow. I've to go now. I'm not gonna give up :). Thanks for helping me. [17:20] camako: kgunn: so there's a bug that already has a fix on devel... [17:20] https://bugs.launchpad.net/mir/+bug/1347789 [17:20] Launchpad bug 1347789 in mir (Ubuntu) "Crash in USC during mir::compositor::BufferQueue::BufferQueue::operator()" [Undecided,Confirmed] [17:20] camako: kgunn: should that be backported to the 0.5 series? [17:25] AlbertA: yep...we will need thta [17:25] that [17:41] here's a fun one [17:41] https://bugs.launchpad.net/mir/+bug/1347828 [17:41] Launchpad bug 1347828 in Mir "glmark2 [desktop] effect=shadow:windows=4 failure on mx3" [Undecided,New] [17:46] kdub_: For me that doesn't even qualify as a fun in the sense of like this is a nice hairy one bug [17:46] and more of just a "off to the crying closet" [17:46] type bug [17:46] failed acquire CCB space [17:46] lol [17:47] lol [17:47] not to mention ScheduleTA "didnt work properly" [17:47] yes, very informative :) [17:47] E/IMGSRV ( 3519): :0: KEGL_SGXDestroyRenderSurface: Timeout failed on waiting for previous render op [17:48] is kind of interesting [17:48] kdub_: Perhaps it has something to do with fences [17:48] i.e. like [17:48] surface is being destroyed [17:48] before the actual rendering of the last frame [17:48] finishes [17:49] E/IMGSRV ( 3519): :0: FreePDSFragBuffers: PDS fragment buffer for render surface still in useE/IMGSRV ( 3519): :0: FreePDSFragBuffers: PDS fragment buffer for render surface still in use [17:49] something, im' thinking some resource problem perhaps around contexts [17:49] but we'll just see [17:49] :) [17:50] good luck :D [17:57] aahhh the good old KickTA... [18:03] :( I want to write... [18:04] constexpr size_t size = someMirGeometrySizeWidth*height*MIR_BYTES_PER_PIXEL(pixel_format) [18:05] (where the geometry and pxiel format are constant as well) [18:06] so that I can then write uint32_t pixels[pixel_size] {} and use automatic storage so that {} can be used to initialize the whole thing to black [18:06] I mean...lol it's purely a matter of conveniencee [18:06] but unfortunately can not use [18:06] IntWrapper with [18:06] constexpr [18:06] because it's not a http://en.cppreference.com/w/cpp/concept/LiteralType [18:07] oh maybe it coudl be though [18:07] I don't really understand the deep details of constexpr lol [18:25] lunnnch [18:26] random question, does/will mir support SDL? [18:26] josharenson: Yes! [18:26] woo hoo! [18:26] courtesy of bschaefer [18:27] who can help with issues and such im sure. [18:27] 95% sure it's SDL2 only [18:27] whatcha up to? [18:27] thanks, no issues, just wondering if I'll be able to play old games [18:31] josharenson, you wont be able to play SDL 1.2 atm (until some layer gets put into place) [18:31] josharenson, but anything for SDL2 will work with mir! === dandrader is now known as dandrader|bbl [18:32] bschaefer thats awesome, thanks [18:32] (if you really want to mess with SDL1.2 + mir you'll need this branch: https://code.launchpad.net/~brandontschaefer/+junk/sdl1.2-mir) === chihchun is now known as chihchun_afk [18:33] they aren't accepting patches for 1.2 soo yeah [18:33] josharenson, np! [19:22] kdub: How should I implement mga::Buffer::write(void*,size_t) [19:22] I'm open to the idea that I shouldn't [19:22] lol [19:22] I just can't remember which android thing has which [19:23] for the touchspots, which end up being kind of like single buffered software clients [19:23] / cursor [19:25] oh I see. gralloc_module_t :) [19:35] racarr_, I guess I'd want to avoid it [19:35] is there a way to verify expectations added to a mock at the end of a scope? [19:36] rather let the test fail unless expectations are met.. [19:36] racarr_, you mean like mg::Buffer::write()? [19:36] anpok, you can verify explicitly [19:36] like Mock::VerifyAndClearExpectations() [19:43] kdub_: Yeah [19:43] to upload cursor images in to the buffer [19:47] racarr_, the pain there is all the pixel formats [19:47] so like, on the client side, we have a way to map it and return the address, but spare the client side buffer the pain of regulating all the pixel format stuff [19:49] like, the intention is to? [19:50] kdub_: hmm it never fails [19:50] I guess I should look at how the client side works on android === kdub_ is now known as kdub [19:52] racarr_, like, whats the outer loop? to write to a cursor on the server side? [19:53] the outer loop? [19:54] like, the larger picture of what having Buffer::write would accomplish [19:54] oh yes [19:54] to write to a cursor on server side. [19:55] write seems like a good choice because 1. they are always small buffers. 2. on mesa you can't even create GL contexts from the cursor buffer. [19:55] ah [19:59] so, its probably a little easier to structure like [19:59] the platform has some writer object [20:01] I thought about that for a bit but coudln't think of [20:02] alternate implementations, is it easier? [20:02] yeah, just because iirc, the mga::Buffer doesn't have access to the gralloc mapping stuff anymore [20:03] whereas the platform could just keep the gralloc as a singleton [20:04] or... even I guess that there could be a mg::Buffer::write(std::vectorconst&) and android could do a gl draw under the hood? (and mesa would just do the memcpy) [20:05] kgunn: do you know a good person I can ask to triage this unity8 crash? https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1347886 [20:05] Launchpad bug 1347886 in unity8 (Ubuntu) "/usr/bin/unity8:6:__assert_fail_base:__GI___assert_fail:operator:std::_Function_handler:operator" [Undecided,New] [20:07] racarr_, I guess I'm not feeling strongly that one is better than the other :) [20:08] other than it probably has to be behind the platform abstraction, and we should have stronger types than void* for the pixels we want to use [20:08] * kdub dreams of bool mir::PixelFormat::ABGR8888::premultiplied() [20:11] robotfuel: i'd like alf_ to take a look at that tomorrow.... [20:12] it looks like it might be mir related [20:15] robotfuel: do you know which image that is/was ? [20:16] kgunn: yesterday's proposed image [20:16] robotfuel: ok so something like ~143 [20:16] robotfuel: any chance we understand/know what the device was doing ? or was this just random monkey ? [20:17] kgunn: version_detail: ubuntu=20140722.1,device=20140717.1,version=144 [20:17] kgunn: it was the random monkey :/ [20:20] kgunn, robotfuel what about https://code.launchpad.net/~mir-team/mir/fix-1335481-0.5/+merge/227497 ? [20:20] seems somewhat in the same ballpark [20:22] kdub: it looks very close +1 [20:25] kdub: sorry lol...alt tabbed and then....distraction [20:25] kdub: I've added that to the bug, thanks [20:26] yeah I was thinking maybe android could do the GL draw under the hood....its probably easiest to use some sort of writing thing I guess though...with the android platform [20:26] and yeah....there are enough instances of void *pixels now [20:26] ill start looking in to something better lol [20:28] Hahahahahaha [20:28] I just tested the touchspots on the phone in [20:28] demo server shell [20:28] and they have [20:28] window decorations [20:28] forgot about that [20:29] hmm what am I going to do there [20:33] kgunn: it looks like this one is also a mir issue https://bugs.launchpad.net/ubuntu/+source/+bug/1347901 [20:33] Launchpad bug 1347901 in ubuntu-system-settings (Ubuntu) "/usr/bin/system-settings:6:__gnu_cxx::__verbose_terminate_handler:__cxxabiv1::__terminate:std::terminate:__cxxabiv1::__cxa_throw:boost::throw_exception" [Undecided,New] [20:35] robotfuel: ah crap i can't see it... [20:35] robotfuel: is there something i can do to see these bugs by default [20:36] * kgunn vaguely remembers mail from duflu [20:37] kgunn: my copy paste glitched https://bugs.launchpad.net/ubuntu/+source/ubuntu-system-settings/+bug/1347901 [20:37] Launchpad bug 1347901 in ubuntu-system-settings (Ubuntu) "/usr/bin/system-settings:6:__gnu_cxx::__verbose_terminate_handler:__cxxabiv1::__terminate:std::terminate:__cxxabiv1::__cxa_throw:boost::throw_exception" [Undecided,New] [20:37] that should work [20:38] kgunn: "void mir::client::rpc::MirSocketRpcChannel::send_message(const mir::protobuf::wire::Invocation&, const mir::protobuf::wire::Invocation&)" is what I see in the trace [20:39] it doesn't look like it was able to complete the whole stack trace :/ [20:41] robotfuel: following the link, looks there may be enough tho... kdub that one looks like its eminating from android platform possibly ?? [20:43] kgunn, I was looking at https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1347886 [20:43] Launchpad bug 1347886 in unity8 (Ubuntu) "/usr/bin/unity8:6:__assert_fail_base:__GI___assert_fail:operator:std::_Function_handler:operator" [Critical,Confirmed] [20:46] and this one https://errors.ubuntu.com/problem/3f606cde0c8ce2787f2a2ae3ed4b1a73738d975e [20:47] the stack passes through the android code, but it looks like it throws somewhere around the ipc stuff, I'd say that looks like a different one [20:47] oh, actually, i guess if the server was asserting and failing [20:48] then maybe the client is just saying "wheres my server" in that stack trace [21:32] kgunn, Silo 6 with the QtComp doesn't seem to start for me. Does it need to be rebuilt after the gcc build? [21:32] tedg: make sure you install ubuntu-touch [21:32] also add proposed to your sources list [21:33] tedg: i put a note at the top of the wiki....its been frustrating a lot of folks [21:33] https://wiki.ubuntu.com/Unity8/QtComp [21:33] i updated the instructions [21:33] silo 8 is gumming up the works [21:35] Ah, k [22:13] ok touchspots are 95% done except the fact that they receive [22:14] window decoprations [22:14] going outside for a little bit (been at desk since 7 lol...ordered lunch in) [22:14] then will be back to work on that === dandrader|bbl is now known as dandrader === renato is now known as Guest43905 [23:06] back [23:24] thomi: Are touchspots black on orange or purple on orange or black on purple or black or [23:24] red on yellow [23:24] :p [23:24] racarr_: huh? [23:25] bzr diff [23:25] oops [23:25] I mean has anyone already [23:25] picked the colors [23:25] nope [23:25] something bright I guess [23:25] ok [23:25] ubuntu orange + black outline [23:26] sounds good to me [23:26] that'll match those fancy videos the design team produce :)