**** BEGIN LOGGING AT Sun Oct 07 02:59:58 2012 Oct 07 05:38:58 the command to configure webos for arm is ./mcf -p 0 -b 0 qemuarm right? Oct 07 05:39:09 I can't seem to find the doc anymore.... Oct 07 05:46:58 yep Oct 07 05:47:35 okay thanks... it used to be in the readme on git, but its either not there or I'm just blind... Oct 07 05:49:38 they removed the section: https://github.com/openwebos/build-webos/commit/f879b820ce2a65400f669d139a66efca63a927a9 Oct 07 05:49:56 hey invalidopcode do you have a touchpad handy with preware installed? Oct 07 05:50:32 HaiKaiDo: my touchpad is currently in two pieces sorry :P why? Oct 07 05:50:58 well its not directly relaated to webos ports or anything Oct 07 05:51:11 but im on the hunt for something that should be in preware Oct 07 05:51:16 panda-z: why did they do that? Oct 07 05:51:19 but its not showing up on my palm pre Oct 07 05:51:37 HaiKaiDo: I might have a pre3 handy... want me to check? Oct 07 05:52:24 well its an ooold app and i dunno if itll showup for you but sure give it a shot if you dont mind Oct 07 05:52:25 its called Oct 07 05:52:33 dunno Oct 07 05:52:47 quick system tasks Oct 07 05:52:52 forgot what it was called XD Oct 07 05:53:10 maybe they have a better way to build arm image Oct 07 05:53:20 HaiKaiDo: actually, my pre3's battery is dead too.. :/ Oct 07 05:53:26 panda-z: hm.. okay Oct 07 05:53:27 D'oh Oct 07 05:53:29 xD Oct 07 05:53:48 EricBlade: do you know why the arm instructions were removed from the open webos readme? Oct 07 05:54:20 invalidopcode: if i had to make a guess, i'd guess it's because arm didn't work yet Oct 07 05:54:26 HaiKaiDo: neither are my daily drivers so I'm not very good about keeping them charged Oct 07 05:54:33 gotcha Oct 07 05:54:40 EricBlade: ah okay Oct 07 05:54:41 I wish the pre 3 worked on sprint Oct 07 05:55:33 I don't have mine activated at all Oct 07 05:59:03 anyone have a pre3 that is running a load average of less than 8.x? Oct 07 06:00:21 i just unboxed one, started it up, put it in dev mode, and it's running at 8. my regular one is running at 8-10, but my pre2 runs at around 0.35 or so Oct 07 06:04:15 okay... I'm out. night folks Oct 07 07:02:50 hi Oct 07 07:05:29 well there's a fun random compile error: "long long long is too long" Oct 07 07:09:07 better too long then "still too short" :) Oct 07 07:10:15 EricBlade: qemuarm build works a bit better now, db8 actually builds now with default settings Oct 07 09:17:10 Is this the right place for asking questions regarding the build-enviroment, when trying to build? Oct 07 09:19:01 for building what? Oct 07 09:19:48 there are serveral and I want to get you to the right place. Oct 07 09:20:24 building for the galaxy nexus, I'm following this guide: http://webos-ports.org/wiki/Galaxy_Nexus_Build_Setup but things are not working out properly Oct 07 09:20:37 make update and try again Oct 07 09:20:51 there are some new fixes in the last our Oct 07 09:20:53 hour Oct 07 09:21:56 mil0ck: if it still doesn't work for you, pastebin the error please Oct 07 09:23:48 I've tried doing the whole process again but I'm still getting the same error, gonna pastebin soon Oct 07 09:27:22 http://pastebin.com/Mw1TTDct this is the pastebin, and keep in mind, I not an expert in these kinds of things Oct 07 09:27:48 that's OK, I'm :) Oct 07 09:28:00 mil0ck: install gawk Oct 07 09:28:16 env: gawk: No such file or directory Oct 07 09:28:32 apt-get install gawk if you have ubuntu or other debian derivate Oct 07 09:29:44 then call "make setup-webos-ports" Oct 07 09:29:57 don't forget: Oct 07 09:29:58 $ cd into-your-build-directory/webos-ports-env/webos-ports Oct 07 09:29:58 $ . setup-env Oct 07 09:30:14 and only _after_ this do MACHINE=tuna bb webos-image Oct 07 09:30:30 working on it right now:) thanks Oct 07 10:40:25 Hi ! Oct 07 10:41:04 After one day of struggle (webos-webkit takes a *lot* of memory to link !), I've got my arm image ! Oct 07 10:42:18 congratulations Oct 07 10:42:41 Now, time to hack it a little bit :) Oct 07 10:45:19 linking qtwebkit now... mem is heavily loaded. Oct 07 10:45:35 panda-z: yes... mine has been up to 2.5GB Oct 07 10:46:01 it made crash my linux once (I forgot to activate swap, and had only 2.7GB in total...) Oct 07 10:46:33 Tofe: ah, my swap is almost full as well Oct 07 10:47:04 I had 5GB if swap ;) It needed about 1.2GB out of it Oct 07 10:47:12 of* Oct 07 10:47:42 well well well. where is that eglfs module now... Oct 07 10:48:53 it's not built ? it uses only the framebuffer ? Oct 07 10:51:02 WebKit is now built (3h:09m:00s). Oct 07 10:51:08 mmh the RPATH for the ARM Qt libs is wrong, but well, that's not critical. Oct 07 10:52:07 Ok, let's try with my 2.6.38 kernel and no opengl. Oct 07 10:55:54 why my boost built libs don't match target arch? weird Oct 07 11:11:42 someone pointed that out here, yesterday, I think Oct 07 11:11:53 I'll check mine Oct 07 11:14:04 my libboost_regex is fine Oct 07 11:17:10 both qemux86 and qemuarm generate x64 libs (my linux is arch x64) Oct 07 11:33:08 ok well. Just putting in the 2.6.38 modules and booting on my Iconia's 2.6.38 kernel didn't seem to work at all. Oct 07 11:38:47 an ARMv7 kernel can peacefully boot an ARMv5 init, isn't it ? Oct 07 11:39:15 should Oct 07 11:39:51 if only I had something on the A500 screen... Oct 07 11:50:51 Maybe I'll try with a more recent kernel, but I don't see why it would improve the situation Oct 07 12:41:18 Okay. none of my kernels work :( Oct 07 12:43:58 is there a way to tell bitbake to use a specific version of the kernel ? Oct 07 13:09:22 Tofe: ye Oct 07 13:09:23 s Oct 07 13:09:32 you do this in your machine confguration Oct 07 13:10:28 Tofe: https://github.com/webOS-ports/meta-smartphone/blob/webOS-ports/master/meta-samsung/conf/machine/tuna.conf Oct 07 14:03:11 morphis: wonderful, thanks Oct 07 14:20:59 morning all. anyone else having issues with github accepting their public key? Oct 07 14:21:19 or have had issues with this in the past? Oct 07 14:22:17 this is keeping me from building the oe version of owos Oct 07 14:23:31 i followed the instructions but i keep getting this error: Key is invalid. It must begin with 'ssh-rsa' or 'ssh-dss'. Check that you're copying the public half of the key Oct 07 14:23:59 the key begins with ssh-rsa and ends with my email address Oct 07 14:24:57 no pb for me, and I did it twice yesterday Oct 07 14:26:25 be sure you copy the whole id_rsa.pub file, without any transformation (line return, so on) Oct 07 14:26:43 did you past the hex key from the terminal or the alpha numeric key from the id_rsa.pub file? Oct 07 14:26:55 id_rsa.pub Oct 07 14:28:00 darn, i did that also. i followed the instructions from here: https://help.github.com/articles/generating-ssh-keys Oct 07 14:28:44 me too, except for the over-complicated xclip stuff Oct 07 14:29:11 I just did "cat id_rsa.pub", and seleted the output and pasted it in github Oct 07 14:32:01 Tofe: thanks! that did it. must have pasted it wrong a few times. Oct 07 14:32:27 :) good Oct 07 14:33:13 morphis: the samsung example is quite complex ! do I have to go through all that webos-ports layer stuff ? Oct 07 14:36:49 of course, it would be a great example for how to create a new port, but right I just want to see if my stuff boots :) Oct 07 14:48:01 hi again. Oct 07 14:52:28 don't want to sound stupid.. but after "MACHINE=tuna bb webos-image" what is the final build result and where should it reside? there is a whole lot of stuff build... but nothing looks like an image file or a linux-root. Oct 07 14:56:10 Garfonso: have a look in the tmp-eglibc directory Oct 07 14:56:29 yes, that's big. ;) Oct 07 14:56:45 is it in the sysroots folder? Oct 07 15:00:06 that you'll have to find out yourself, I've only build specific components as IPKs so far Oct 07 15:02:17 hm. ok. there are a whole lot of ipks now.. mostly in the work folder. Oct 07 15:02:59 Is there a place describing what are the minimal steps needed to create a specific webos-port build ? Oct 07 15:04:56 I only know the list and description of the repositories. Oct 07 15:06:01 mmh. okay. Oct 07 15:06:19 but I really don't know much (yet). ;) Oct 07 15:07:12 I could as well try a GNexus build, it's not so far from what I'd like to get Oct 07 15:07:25 Tofe: are you building for a gnex or for another device (just curious)? Oct 07 15:07:48 another, I'm trying to make it work on the Iconiatab A500 (Tegra2) Oct 07 15:08:23 I've got a working kernel 2.6.38, and also another one 3.0 Oct 07 15:08:45 (with the corresponding NVidia blobs) Oct 07 15:08:49 thanks, i think i remember that from yesterday. i'm trying to make it work on a motorola photon Oct 07 15:09:36 so far, it looks like my assumption that the arm build would work with a different kernel was quite wrong, unfortunately Oct 07 15:10:19 darn Oct 07 15:10:52 what seems to be the problem? Oct 07 15:10:52 Tofe: you mean the armv7 rootfs doesn't work with your kernel? Oct 07 15:11:31 well, the rootfs seems to be an armv5, isn't it ? Oct 07 15:13:18 but I don't have much details: the tablet doesn't output any text at all apart from "booting kernel"; so I don't know exactly what step went wrong (init ? glibc ? mounting ?...) Oct 07 15:14:06 booting a homemade OS on that tablet is still a bit of a hack, even if xda made great work on it Oct 07 15:14:49 is there a way to run an android linux kernel in safe mode? Oct 07 15:14:56 do you have a chance to set console variable to kernel boot param? Oct 07 15:15:15 penk: if I rebuild the kernel, yes :) Oct 07 15:15:41 last time I played with Iconia tab it's still using recovery trick to side-load custom kernel Oct 07 15:17:26 Exactly. I has just improved a little bit: now, with the ICS bootloader, it is possible to setup a second kernel, apart from first kernel and recovery. And it is possible to set it as the default boot :) Oct 07 15:17:57 With fastboot, it is also possible to boot a kernel+initrd without even flashing it Oct 07 15:18:04 nice Oct 07 15:18:42 Yes, exactly, so I'm playing with that and it works quite well: ubuntu, Mer/Nemo, Mer/Plasma, all of that work more or less Oct 07 15:24:00 funny enough, we do have the source code of the 3.0 iconia kernel on github; I think it should be possible to integrate it in bitbake, but I'm still feeling a bit nooby with that stuff :) Oct 07 15:26:04 is there a guide somewhere on how to boot a custom image on an andriod device? Is this very different for the devices? Oct 07 15:26:24 i have the same question Oct 07 15:26:31 quite different usually, yes Oct 07 15:26:44 however xda-developer regroup a lot of them Oct 07 15:27:31 the big thing being always to hack the bootloader to make it boot on an unsigned image Oct 07 15:27:45 i'm definitely a noob on this. i've built and played with kernels for the desktop. never on a mobile device. Oct 07 15:28:45 i figure there is only one way to learn.. Oct 07 15:33:25 Ok, let's retry with linux-yocto-3.0 ... Oct 07 15:37:25 crossing my fingers for you Oct 07 15:39:29 hopefully bitbake it really smart, it detected the .conf change and is only doing what's needed Oct 07 15:45:55 Tofe: I hope that works. I've got an A500 over here I can mess with if it does. Oct 07 15:47:29 hope too... interestingly, if bitbake only recompiles the kernel, it won't be a good sign. It would mean that no dependancy has been declared on the kernel, i.e. the other kernels should have worked Oct 07 15:56:18 I'm beginning to doubt: is armv5 code able to run on armv7 architecture ? Oct 07 15:58:31 why would oe owos build to armv5 instead of armv7? Oct 07 15:58:54 don't know, maybe qemuarm needs that Oct 07 15:59:58 right now I'm failing to see where this "armv5" target is specified Oct 07 16:09:43 Tofe: 'grep -r 'armv5' ./'? Oct 07 16:10:07 ShiftyAxel: it's ok, I found it in the qemuarm.inc machine configuration Oct 07 16:12:00 What I will try, before putting in place a more robust "webos-ports" solution, is to hack a bit qemuarm.inc to make it include cortexa9.inc instead of the armv5 one. Also I'm switching to yocto-3.0, just in case. Oct 07 16:12:30 that means recompiling a hell lot of stuff, of course... Oct 07 16:22:27 mmmh it goes a bit too fast, I fear that bitbake didn't understand that the target architecture has changed... Oct 07 16:26:41 ;) Oct 07 16:41:43 Tofe: you have to remove tmp-eglic when changing architecture of a machine Oct 07 16:43:51 oops. Oct 07 16:44:24 switching architecture isn't a simple change Oct 07 16:44:32 which means to rebuild everything ;) Oct 07 16:44:33 agreed :) Oct 07 16:44:47 I just don't want to re-fetch everything Oct 07 16:45:08 Tofe: refetch isn't needed Oct 07 16:45:14 just keep your downloads directory Oct 07 16:45:20 and sstate-cache of couse Oct 07 16:45:39 so I remove all "work" and "work-shared" and that's it ? Oct 07 16:46:27 no Oct 07 16:46:31 complete tmp-eglic Oct 07 16:46:47 you can rename it if you want to switch back later Oct 07 16:46:51 I'm trying hard to find it Oct 07 16:47:21 you're usng webos-ports-setup or build-webos? Oct 07 16:47:27 what is saved in sstate-cache? Oct 07 16:47:28 build-webos Oct 07 16:48:05 in sstate-cache, lot of armv5vte compiled stuff Oct 07 16:48:12 Garfonso: when OE is build stuff is caches the state of a component (recipe, meta-data, package data, etc) so you don't have to recompile it if nothing has changed after you removed your tmp-eglic dir Oct 07 16:48:26 Garfonso: we're using it in webos-ports to give people a fast start Oct 07 16:48:51 you setup up everything, use sstate and can build the image with patch cache items from our build servers in a short time Oct 07 16:49:02 s/patch// Oct 07 16:49:02 morphis meant: you setup up everything, use sstate and can build the image with cache items from our build servers in a short time Oct 07 16:49:34 helps to reduce build time for big components like webkit Oct 07 16:49:50 oh, big, big....... Oct 07 16:49:58 :) Oct 07 16:50:05 Tofe: whats your target device? Oct 07 16:50:37 in practice, it is a tegra2/arm cortex A8 device Oct 07 16:50:45 (iconia tab a500) Oct 07 16:50:52 A9, sorry Oct 07 16:51:17 ok, so you might be able to use a lot of what we build for the galaxy nexus Oct 07 16:51:32 yes, that's what I saw from your link earlier Oct 07 16:51:49 I just didn't know at all where to begin in order to reuse that Oct 07 16:51:57 just ask :) Oct 07 16:52:37 ok, well, let's go :) Should I clone webos-ports, first ? Oct 07 16:53:29 Tofe: http://www.webos-ports.org/wiki/Galaxy_Nexus_Build_Setup Oct 07 16:53:52 ah and we wont support any other machines than the galaxy nexus in the near future Oct 07 16:54:14 but will help you to get it working as best as we can Oct 07 16:54:22 no pb at all Oct 07 16:54:25 ok Oct 07 16:55:22 my-build-directory is build-webos, right (if I want to reuse some of the fetched software) Oct 07 16:55:42 Tofe: safe the downloads folder Oct 07 16:55:50 thats the only one you can share with webos-ports Oct 07 16:55:54 everything else you can remove Oct 07 16:56:00 ok, I'll do that Oct 07 16:56:08 morphis: I did follow all steps in the wiki and I think it did build ok... but I'm not sure where it build what... if I had a galaxy nexus (maybe I'll get one to tinker around with next week) what would be the next steps? Oct 07 16:56:31 Garfonso: after you have the ext2 image and kernel? Oct 07 16:58:29 do I have an ext2 image? I'm a bit lost after the MACHINE=tune bb webos-image ... Oct 07 16:58:46 it did compile very long and in the end said something about 3000 jobs and no failures. Oct 07 16:59:00 but I did not really figure out what it build... Oct 07 16:59:13 yeah, I'm a noob in this area.. sorry. ;) Oct 07 16:59:33 ah ok Oct 07 16:59:45 Garfonso: everything it has build will be in tmp-eglic/deploy Oct 07 16:59:52 ipk packages + images + kernel Oct 07 17:00:08 Garfonso: quite few people have more than one week of experience in that area, I guess :) Oct 07 17:00:29 ahh... that's the first step I was missing. Oct 07 17:02:12 Tofe: yes, thats why we're providing a foundation for all porting stuff with webos-ports Oct 07 17:02:21 and caring about the basic needs Oct 07 17:02:45 Garfonso: we're not providing any instructions yet how you get the build images working on your device Oct 07 17:02:55 Okay, I'm on the edge of starting bitbake for the "tuna" machine, maybe I should have a look at the tuna config Oct 07 17:03:03 webos-image-tuna.ext2 would be the ext2 image you talked about, right? Oct 07 17:03:19 Garfonso: yes Oct 07 17:03:35 and zImage a kernel, I guess. Oct 07 17:03:39 yes Oct 07 17:03:50 cool. :) Oct 07 17:05:10 Tofe: You wouldn't mind documenting what you're doing for other people to follow? Oct 07 17:05:16 ok, and with that files, what would be the next steps for a galaxy nexus? Is there some how to somewhere? Oct 07 17:05:27 xda developers maybe? Oct 07 17:05:31 modulusshift: I'll try, yes Oct 07 17:05:42 Thanks. Oct 07 17:07:40 Garfonso, Tofe: we're talking about how much instructions we will provide yet to help you to try things out Oct 07 17:08:02 as of right now we don't tell anything how you get the ext2 image and kernel onto the device Oct 07 17:08:29 (you might be able to figure this out if you read the OE recipes carefully but we're not offically talking about this yet) Oct 07 17:09:28 Why the silence? I doubt anyone would find their way to this channel without the understanding we might brick a device. Oct 07 17:09:36 one can brick a gnex, right? Oct 07 17:11:41 for my part it's ok; I'm going to produce a rootfs compatible for the Tegra2, putting the rootfs on a sdcard (or on the device itself) and booting it with the corresponding kernel from xda-developers is a bit outside the field Oct 07 17:12:25 modulusshift: as I said we're talking about this and trying to find out what is the best way to get the instructions out Oct 07 17:12:53 as we need to take care about getting people enabled to help us getting porting done and avoid people trying out things and breaking anyhting Oct 07 17:13:19 modulusshift: we want this but we have to find the right way Oct 07 17:14:11 morphis: what are these "_virtual" preferred providers ? Are they actual compiled software ? Oct 07 17:14:52 Tofe: there is for example and virtual provider for the linux kernel Oct 07 17:14:57 yeah, I understand your point of view... Oct 07 17:15:01 Alright. It's still really early in the project. I've always admired the polish the webOS community has had in this sort of thing. Best to get it right to begin with. Oct 07 17:15:05 if you're targetting different devices you have different kernels Oct 07 17:15:21 but you don't want to depend on a specific kernel in other recipes Oct 07 17:15:30 thats why you create a virtual provider for it Oct 07 17:15:37 and set it per machine/distro Oct 07 17:15:57 modulusshift: yes Oct 07 17:16:28 modulusshift: if we were in the situation to provide a full featured webos experince yet we would do it Oct 07 17:16:35 but it's too early Oct 07 17:17:03 pre-pre-alpha, not usable and most people will be disappointed from the current state Oct 07 17:17:21 (from the user perspective) Oct 07 17:17:35 openwebos isn't that far yet Oct 07 17:17:47 it isn't really ready for porting to real devices Oct 07 17:18:06 too much missing, too much isn't support for real devices Oct 07 17:18:23 qemu* and desktop may work atm but real target devices are completely different Oct 07 17:19:03 Be on the lookout for stable ports in the coming months, huh? Oct 07 17:19:09 great, the nexus uses a 3.0 kernel branch Oct 07 17:21:56 modulusshift: stable ports ... good word, hard to define and we're far away from that Oct 07 17:22:16 we're just at the beginning Oct 07 17:22:49 sorting things out, getting in touch with HP on the opensource level (github pull-requests), development of new components, interfacing with android hardware, ... Oct 07 17:23:26 and we need every help we can get Oct 07 17:23:44 wow, the binary blobs of the nexus are quite getting in the way, I see why it was so hard to port it :) Oct 07 17:23:45 it doesn't make sense to develop for more than one device yet Oct 07 17:23:51 Tofe: :) Oct 07 17:24:29 However it is quite handy for me, as the situation is very similar with the IconiaTab Oct 07 17:24:32 modulusshift: there are too much open questions to answer and problems to solve so we'll just targetting one device for a longer time until we have solve the initial problems Oct 07 17:24:59 devices will not go away and there will be a lot new devices in the coming months and hardware specific things we getting a lot better Oct 07 17:25:26 so let us get the basics right so porting to other devices will be done in a reasonable time later for a device of your choice Oct 07 17:26:22 Stskeeps: you tried running test_egl and test_glesv2 at the same time? Oct 07 17:27:14 no, it usually doesn't work nicely Oct 07 17:28:10 two things can't use framebuffer at same time Oct 07 17:28:31 hm, I am bit confused Oct 07 17:28:44 luna-sysmgr uses a QPA platform module for graphics Oct 07 17:28:50 which is based on eglfs Oct 07 17:28:54 right Oct 07 17:29:00 on startup it forks Oct 07 17:29:02 eglfs is practically 'full screen egl' Oct 07 17:29:06 ok Oct 07 17:29:16 main process is the luna-ui and second one for the cards Oct 07 17:29:33 i must admit i've only seen the QWS implementation of luna Oct 07 17:29:35 maybe I should show you some code? Oct 07 17:29:47 yes, there's one thing I didn't understand in there: if the whole Qt stuff is based on the gles libs of the nexus why bother with a X server ? Oct 07 17:30:07 Tofe: who uses a X server? Oct 07 17:30:13 morphis: yes, that'd help Oct 07 17:30:50 morphis: it *looks like* it is included in the rootfs of the "tuna" machine Oct 07 17:30:54 Stskeeps: ok, here it forks https://github.com/webOS-ports/luna-sysmgr/blob/master/Src/Main.cpp#L706 Oct 07 17:31:00 Tofe: its not Oct 07 17:31:01 at least there are providers for it Oct 07 17:31:10 just some components have some x dependency Oct 07 17:31:16 ah, oki Oct 07 17:31:17 but that would not include a full x server Oct 07 17:31:24 I do hope so :) Oct 07 17:31:34 qt for example depends on some x header files Oct 07 17:31:44 oki Oct 07 17:31:55 Stskeeps: in that line it forks the second process but sets the right QPA module before Oct 07 17:32:19 Stskeeps: so both processes will use the same module and initialisation for it Oct 07 17:32:22 morphis: okay, can you try to force ::setenv("QT_QPA_PLATFORM", "palm-soft", 1); Oct 07 17:32:25 at first? Oct 07 17:32:42 or force sw rendering, for the child Oct 07 17:32:51 thats what we're doing already Oct 07 17:33:02 works fine Oct 07 17:33:16 alright, but you'd like to move on to be using egl? Oct 07 17:33:28 yes Oct 07 17:33:41 for that I changed the config flags for luna-sysmgr to use opengl Oct 07 17:33:54 that will switch the card rendering process to use egl and gles Oct 07 17:34:16 yeah, that's where you need to be careful :) Oct 07 17:34:29 Stskeeps: https://github.com/webOS-ports/luna-sysmgr/blob/master/sysmgr.pro#L679 Oct 07 17:34:32 can you show me how a 'client' behaves when setting up egl? Oct 07 17:34:37 they have different implementations Oct 07 17:34:39 ok, I just did minor modifications to the tuna machine, and let's go Oct 07 17:34:52 Stskeeps: can get the log, yes Oct 07 17:34:55 ok Oct 07 17:35:36 Stskeeps: it fails here: https://github.com/webOS-ports/qt/blob/master/src/plugins/platforms/palm/qeglfsscreen.cpp#L121 Oct 07 17:35:40 (in the second process) Oct 07 17:36:02 eglGetDisplay returns EGL_NO_DISPLAY Oct 07 17:36:08 yes, figured as much Oct 07 17:36:14 do you know what display it passes? Oct 07 17:36:21 DEFAULT_DISPLAY Oct 07 17:36:30 you mean for EGLNativeDisplayType? Oct 07 17:36:48 yeah Oct 07 17:37:06 alright, so, naturally this won't work and i'm fairly sure that it wasn't the way it was done on touchpad for instance :) Oct 07 17:37:24 as you're trying to init yet antoher screen Oct 07 17:37:28 on the same screen Oct 07 17:38:08 did you ever read about the stuff that was done for wayland on raspberry pi? Oct 07 17:38:19 no Oct 07 17:38:37 maybe I should have a look at the qt sources for the touchpad Oct 07 17:39:09 * Stskeeps ponders Oct 07 17:39:31 i have a nice drawing i need to find.. where did i put it Oct 07 17:39:35 ok Oct 07 17:39:46 Stskeeps: should I show you the code for the remote rendering too? Oct 07 17:39:55 that too Oct 07 17:40:44 Stskeeps: ok, first start is the configuration: https://github.com/webOS-ports/luna-sysmgr/blob/master/sysmgr.pro#L679 Oct 07 17:40:53 I just selected opengl as config flag Oct 07 17:41:17 which will use HostWindowDataOpenGL.cpp, RemoteWindowDataSoftwareQt.cpp and RemoteWindowDataOpenGLQt.cpp Oct 07 17:41:31 all to find here: https://github.com/webOS-ports/luna-sysmgr/tree/master/Src/webbase Oct 07 17:41:57 but when opengl is selected (https://github.com/webOS-ports/luna-sysmgr/blob/master/Src/webbase/RemoteWindowData.cpp#L26) Oct 07 17:42:04 it uses RemoteWindowDataOpenGLQt Oct 07 17:44:03 Stskeeps: and runtime log will come in some minutes Oct 07 17:44:08 a quick glance, it'll use opengl to render into a shm buffer which is then asked to be shipped to screen? Oct 07 17:46:13 yes, somehow Oct 07 17:46:32 hey ! it's back to mono-process ! Oct 07 17:46:46 they use some ipc mechanism to do this Oct 07 17:49:39 morphis: so, at the moment, if you use sw rendering for the sub webkit subprocess, and opengl for lunasysmgr , all good? Oct 07 17:49:51 yes Oct 07 17:50:09 ok Oct 07 17:50:13 that works fine but will reduce the framerate to a bare minimum when opening just one card Oct 07 17:50:21 so it's unusable Oct 07 17:50:29 yeah, shm is nasty.. Oct 07 17:51:07 morphis: can you try to make sure the sub webkit process won't be trying to do a eglCreateWindowSurface? Oct 07 17:51:09 I am unsure if it's a shm, they are using a PIpcBuffer Oct 07 17:51:13 and it 'just' renders into a buffer? Oct 07 17:51:33 will try to find a call to eglCreateWindowSurface Oct 07 17:51:59 but should not eglGetDisplay succeed? Oct 07 17:53:03 well, on a normal, sane system yes it would, but.. Oct 07 17:53:03 :P Oct 07 17:53:21 * Stskeeps ponders idly Oct 07 17:53:29 I meant it should succeed when calling from a second process when first one called it already Oct 07 17:53:52 well the original process still has the display open Oct 07 18:00:02 ok Oct 07 18:00:10 so that should proceed fine in theory Oct 07 18:00:24 I am just a little bit lost with this EGL/GLES stuff :) Oct 07 18:00:29 never did something with it before Oct 07 18:01:29 it's really black magic at times Oct 07 18:03:37 isn't there some way to pass something else than EGL_DEFAULT_DISPLAY? Oct 07 18:03:50 it's hard coded in the palm platform module Oct 07 18:04:11 do you know if lunasysmgr non-webkit subprocess is supposed to be rendering to the screen/ Oct 07 18:04:12 like new QEglFsScreen(EGL_DEFAULT_DISPLAY) Oct 07 18:04:18 no, it's not Oct 07 18:04:26 as far as I know Oct 07 18:04:35 ShiftyAxel may be the better one to ask here Oct 07 18:04:49 he is more experienced with the luna-sysmgr Oct 07 18:04:59 lunasysmgr being the parent that is Oct 07 18:05:05 yes Oct 07 18:05:08 ok Oct 07 18:05:27 but as I read the sources its just rendering to a buffer which is then shared with some IPC mechanism Oct 07 18:05:37 so I doubt it's rendering directly to screen anywhere Oct 07 18:05:44 ok Oct 07 18:05:45 one thing I do know is that there's a function that sets direct rendering for a window, if that's of any help Oct 07 18:05:53 it gets called when a card is put into fullscreen Oct 07 18:06:01 and then unset when minimizing Oct 07 18:06:06 there is a directrendering config option as well Oct 07 18:06:36 yeah, that's just a way to render directly without making buffers all the time i think. Oct 07 18:06:46 ie, get fb0 or something.. Oct 07 18:08:53 morphis: could you try to figure out what Android NDK apps use in order to get a window to render in? Oct 07 18:09:33 ie, c/c++ android apps Oct 07 18:10:47 Stskeeps: log file: pastie.org/4928474 Oct 07 18:10:56 Stskeeps: can do Oct 07 18:11:14 log with indication of eglCreateWindowSurface Oct 07 18:11:31 will extend libhybris so it shows the current process id in the log states too Oct 07 18:12:32 where in that log does it do the fork? Oct 07 18:12:50 good question Oct 07 18:12:56 will add a log entry for this too Oct 07 18:13:01 but tomorrow morning Oct 07 18:13:06 now it's enough for today Oct 07 18:13:14 alright Oct 07 18:13:35 hopefully openwebos will go towards wayland too, it just makes life easier Oct 07 18:13:35 :P Oct 07 18:14:13 they plan to switch qt5 .. lets see what this is about :) Oct 07 18:14:40 wise choice Oct 07 18:14:41 :P Oct 07 18:14:54 i'm dealing with a fonts issue at the moment, much fun Oct 07 18:15:00 there's really some magic in text rendering Oct 07 18:16:58 today, my laptop does not work again. i think it is toast Oct 07 18:17:14 fxspec06: what happened? Oct 07 18:45:45 ok, i keep forgetting that my irc program minimizes to the tray when i click the x button. i apologize if anyone was trying to reach me. Oct 07 18:46:17 i ran into a problem on the build. i'm putting together a pastebin now Oct 07 18:50:31 looks like there was a problem with building openjade. i ran the make command again. same error. here is the rather large pastebin: http://pastebin.com/ghYKKWGg Oct 07 18:53:19 if someone with a little more knowledge would look at this for me i would appreciate it. Oct 07 19:18:11 what is the filesystem of your "Iomega-HDD"? Oct 07 19:20:13 is it FAT32 or NTFS or something? Oct 07 19:20:27 that might be the reason why the chmod fails. Oct 07 19:20:30 @k4ever Oct 07 19:20:56 it is ntfs Oct 07 19:21:35 yeah.. that does not work nice with chmod. Oct 07 19:21:44 I had that issue in a different project. Oct 07 19:22:12 is there a way to bypass the chmod check? Oct 07 19:23:35 the only 40gb plus free space i have is on windows partitions Oct 07 19:29:15 k4ever: NTFS will not work for OE Oct 07 19:29:18 most likely Oct 07 19:29:22 you'll need to use ext4 Oct 07 19:30:14 ugh, ok. i have a slow 100gb usb drive i could reformat to ext4. i will do that then try again. Oct 07 19:34:28 if not, i have 37 gb left on my laptop. build calls for at least 40 gb.. Oct 07 19:37:55 37gb won't be enough... I did try that, too. ;) Oct 07 19:38:22 moved to two different drives, until I found one that had enough space left. Oct 07 19:38:50 one had 38gb and failed with one of the last jobs.. :( Oct 07 19:43:43 darn. well i may have to shut down to fix this new drive. Oct 07 19:43:53 see you guys in a few minutes Oct 07 19:45:00 what bot is WoceBot? Oct 07 20:37:19 dunno Oct 07 20:37:26 ~info Oct 07 20:58:01 wocebot: wocebot? Oct 07 20:58:08 wocebot: infobot? Oct 07 20:58:19 * dwc- slaps WoceBot around a bit Oct 07 20:58:23 CTCP VERSION reply from WoceBot: infobot 1.5.3 -- linux Oct 07 21:03:30 * WoceBot whacks dwc- with the cluebat, courtesy of shiftyaxel Oct 07 21:03:38 you made WoceBot angry! Oct 07 21:35:57 folks Oct 07 21:36:18 morning Oct 07 21:39:05 morning bhuey Oct 07 21:39:11 morning Oct 07 21:48:46 laptop dead again Oct 07 23:17:38 hi all. Oct 07 23:18:40 i thought i would only be gone for a few minutes 4 hours ago, but my old 100gb usb hd was corrupted. Oct 07 23:19:57 after wasting several hours on it, i finally decided to resize the ntfs partition on my 1tb drive to allow an 80gb ext4 partition. Oct 07 23:20:37 then i had to fight with linux mint/ubuntu to allow me to mount it with read/write access. Oct 07 23:20:54 you got to like this stuff to do this. Oct 07 23:23:43 the build just started. it is on an 80gb ext4 drive so hopefully i won't run into any other issues Oct 07 23:30:35 has anyone got the oe version to work using an android kernel? Oct 07 23:31:50 specifically a 2.6.x kernel? Oct 07 23:40:39 he sure hits and runs... Oct 07 23:54:22 sometimes, i swear, you two are the same people Oct 08 00:11:51 LarrySteeze, how is you Oct 08 00:23:11 what's the good word today people? Oct 08 00:48:14 Jack87: you live! Oct 08 00:48:30 LarrySteeze, indeed Oct 08 00:48:53 LarrySteeze, I live on the east coast now.. no longer in utah Oct 08 00:49:02 wow! where on the east coast? Oct 08 00:49:12 LarrySteeze, Rhode Island Oct 08 00:49:17 not far from me then Oct 08 00:49:40 provi area? Oct 08 00:49:40 LarrySteeze, you told me to get a hold of you if i ever made it this way haha. but i forgot where you are at.. Boston? Oct 08 00:49:46 yup Oct 08 00:49:59 I am about 15min south of providence Oct 08 02:31:04 hey peeps **** ENDING LOGGING AT Mon Oct 08 02:59:58 2012