**** BEGIN LOGGING AT Sat Jan 30 02:59:57 2010 Jan 30 03:00:32 destinal-work: just tring to get evas to build with gl support Jan 30 03:00:55 PuffTheMagic: it'll build. it will just display garbage. thats as far as xcomp got Jan 30 03:01:24 raster: well what i mean is im trying to get it to build against my gentoo cross userland that i've been using Jan 30 03:01:29 and i dont have all the GL headers in place Jan 30 03:01:33 it seems Jan 30 03:01:36 or the wrong ones Jan 30 03:03:17 ooh ok Jan 30 03:03:31 for webos u will need gl-es2 headers Jan 30 03:03:48 (GLES2/gl2.h) Jan 30 03:03:52 for starters Jan 30 03:04:08 + sdl with gl-es/egl support built too Jan 30 03:05:10 ahh so i need gl2.h but renamed to gl.h? Jan 30 03:05:25 cause the check is for gl.h Jan 30 03:06:08 ehh i dont seem to matte rwhat i put in GL/ or EGL/ it never passes the tests Jan 30 03:06:17 the SDL header its detecting though Jan 30 03:07:02 or does it require pkg-config Jan 30 03:14:01 raster: do you have any idea how the gl.h check works cause i know I have a GL/gl.h Jan 30 03:14:16 PuffTheMagic: no. u need gl2.h Jan 30 03:14:23 i have those 2 Jan 30 03:14:28 PuffTheMagic: evas gles support uses gles headers and conventions Jan 30 03:14:46 the script never detects a GL/gl.h file Jan 30 03:14:52 GLES2/gl2.h is the gles2.0 header Jan 30 03:14:53 i dont see a check for gl2.h Jan 30 03:15:00 there is Jan 30 03:15:13 alsi libGLESv2.so Jan 30 03:15:14 is the lib Jan 30 03:15:30 EGL/egl.h is .. egl Jan 30 03:15:58 u do have to explicitly --enable gl-x11 snfd --enable gles flavor and --ebale which kind of gles flavor Jan 30 03:16:06 (sgx or s3c6410) Jan 30 03:16:08 see --help output Jan 30 03:16:15 i beleive you that gl2.h is the header i need but i enabled --enable-gl-sdl --enable-gl-flavor-gles --enable-gles-variety-sgx Jan 30 03:16:24 ok Jan 30 03:16:27 thats right Jan 30 03:16:32 but it never checks for gl2.h Jan 30 03:16:34 well i have never done the gl-sdl Jan 30 03:16:34 when i run configure Jan 30 03:16:37 only gl.h Jan 30 03:16:37 gl-x11 for me Jan 30 03:16:38 yeah we lack x11 Jan 30 03:16:40 but otherwise the same Jan 30 03:17:17 nb Jan 30 03:17:21 it never looks for gl2.h Jan 30 03:17:24 it assumes it Jan 30 03:17:30 it looks for the lib Jan 30 03:18:04 well why is it looking for gl.h Jan 30 03:18:05 mind u Jan 30 03:18:06 and failing Jan 30 03:18:11 if gl2.h is what i need Jan 30 03:18:16 looks like xcomp has extra headers in his env Jan 30 03:18:18 egl is still Jan 30 03:19:02 oh no Jan 30 03:19:05 thhats th x111 checks Jan 30 03:19:13 AC_CHECK_HEADERS([SDL/SDL_opengles.h EGL/egl.h], [have_egl="yes"]) Jan 30 03:19:23 then...\ Jan 30 03:19:24 AC_CHECK_LIB(GLESv2, glTexImage2D, [have_glesv2="yes"], , -lEGL -lpthread -lm) Jan 30 03:23:37 im not telling it to build x11 though Jan 30 03:25:11 its not checking Jan 30 03:25:20 it checks for egl and then glesv2 lib Jan 30 03:25:29 something with this GLES stuff must not be too crosscompile friendly Jan 30 03:25:29 if u have that... u'll have a gles2 sdk Jan 30 03:25:39 no need to check for gl2.h beyond being paranoid for checking Jan 30 03:25:58 ask xcomp Jan 30 03:26:00 he builds it Jan 30 03:26:06 as such gl-es is cross-build freindly Jan 30 03:26:19 i;ve built gles2 x11 engine many times - crossbuild Jan 30 03:26:49 i cant get past configure and i know i have the rigth headers and its not finding anything gl properly Jan 30 03:27:00 u neeed the libs too Jan 30 03:27:12 yeah i have it Jan 30 03:27:18 but it fails that check too Jan 30 03:28:08 rsync is pissing me off Jan 30 03:28:16 why is it copying entire files? Jan 30 03:28:30 its as if the gl checks arent using ${PREFIX} or what ever it is to set where to look for the headers Jan 30 03:28:48 "rsync -avz /foo/bar /foo/bar2" Jan 30 03:29:22 if /foo/bar contains FILEA and /foo/bar2 also contains FILEA, and they are the same, shouldn't rsync skip it? Jan 30 03:29:43 notwith that commands u gave it Jan 30 03:29:45 i dont think Jan 30 03:30:01 then what would? Jan 30 03:30:20 man rsync Jan 30 03:30:28 u want update or something like that Jan 30 03:30:34 PuffTheMagic: u should set CFLAGs, LDFLAGs etc. etcc. too u know Jan 30 03:30:41 the usual cross-compile stuff Jan 30 03:30:44 I did that too Jan 30 03:30:47 and no no checks every checkk $PREFIX Jan 30 03:30:53 thats non-standard to do so Jan 30 03:30:58 rsync -avzu /foo/bar /foo/bar2 Jan 30 03:31:01 raster: that should be already handled properly in the ebuild for evas Jan 30 03:31:02 and its still copying the entire file Jan 30 03:31:09 it "has" been cross compiling fine Jan 30 03:31:15 and it finds SDL_gles.h Jan 30 03:31:22 can't help u Jan 30 03:31:23 or SDL_opengles.h Jan 30 03:31:27 it does work on cross-build Jan 30 03:31:40 xcomp managed to build and run on his pre as he at least knows it displays garbage Jan 30 03:31:51 unlress he used the desktop gles sdk only + sdl Jan 30 03:31:53 if he was using sb2 that is not cross-compile Jan 30 03:32:05 it is cross-compile Jan 30 03:32:11 the compiler is actually x86 Jan 30 03:32:16 and it produces arm code Jan 30 03:32:16 :) Jan 30 03:32:24 that is cross-compiling by definition Jan 30 03:32:25 :) Jan 30 03:32:27 sb2 setup using qemu-arm to build it Jan 30 03:32:29 natively Jan 30 03:32:46 sb2 uses binaries for x86 for most core tools Jan 30 03:32:54 like bsh, ls, rm, tar, make, gcc, etc. Jan 30 03:33:01 its can run bth archs Jan 30 03:33:06 PuffTheMagic: you need the headers from the powervr sdk Jan 30 03:33:12 so your poutout - if arm can run like x86 Jan 30 03:34:19 xcomp: i assume he already had those... as thats a pretty obvious pre-requisite Jan 30 03:34:23 PuffTheMagic: copy the contents of the SDKPackage/Builds/OGLES2/Include directory into your staging include dir Jan 30 03:34:27 (or they are provided from the palm-pre sdk) Jan 30 03:34:36 xcomp: ok Jan 30 03:34:38 they're not in the palm dsk Jan 30 03:34:40 *sdk Jan 30 03:34:44 you have to download from imgtec Jan 30 03:34:54 well thats one useless-arse-sdk Jan 30 03:35:03 raster: they haven't released the native sdk yet :) Jan 30 03:35:14 i'm sure those headers will be there when they release it Jan 30 03:35:44 I'm going to go ahead and stage them in cross-compile though, since they are sgi license Jan 30 03:35:45 they'd better be :) Jan 30 03:36:58 xcomp: this is pretty crazy 115mb download for some text headers :D Jan 30 03:37:04 yeah :) Jan 30 03:38:24 xcomp: so it runs but it down run? Jan 30 03:38:27 Hi all :) Jan 30 03:38:33 Anyone know what Pseudoland in the emu is?!? Jan 30 03:38:45 PuffTheMagic: it currently displays garbage for me Jan 30 03:38:54 xcomp: any idea why? Jan 30 03:39:15 don't know yet ... I haven't done any debugging today and probably won't be able to until sunday Jan 30 03:39:18 xcomp: what u testing with expedite? Jan 30 03:39:22 yeah Jan 30 03:39:49 gl_sdl on x11 works fine Jan 30 03:39:59 hmm Jan 30 03:40:29 but gl_sdl on sdl Jan 30 03:40:30 :D Jan 30 03:40:55 you mean gl_sdl on fb :) Jan 30 03:41:03 specifically, gles_sdl on fb Jan 30 03:41:08 yeah Jan 30 03:41:52 xcomp: what init flags is evas passing to SDL_init Jan 30 03:42:19 cause i just saw fullscreen, hw_surface, noframe, and alpha Jan 30 03:42:24 but hw and gl arent the same Jan 30 03:42:39 it passes SDL_OPENGLES Jan 30 03:42:48 thats defined in EVAS_SDL_GL_FLAG Jan 30 03:43:11 ohhmm Jan 30 03:43:13 ok Jan 30 03:43:19 nm Jan 30 03:43:22 im thinking of evas_sdl_software Jan 30 03:43:53 xcomp: what the link the api docson that again? Jan 30 03:44:23 i dunno, I just go poking around google and the code itself Jan 30 03:45:04 but i've gone over the flags ad nauseum Jan 30 03:45:23 i think i've convinced myself that the flags are not the issue Jan 30 03:50:02 hey PuffTheMagic ping Jan 30 03:50:14 sup Jan 30 03:50:41 PuffTheMagic: What you're drinking has a large effect on your results in hacking.. I've got Dogfish Head India Brown Ale, which is just about perfect but I think would be better balanced with a little less hops Jan 30 03:51:11 destinal: i've got wine right now actually Jan 30 03:51:28 destinal: coding while intoxicated? Jan 30 03:51:52 xcomp: PWI Jan 30 03:52:05 Hrm, can I create a second Palm profile with the same email address? Jan 30 03:52:11 PuffTheMagic, how much you know about this mjson stuff Jan 30 03:52:53 umm, a little bit why? Jan 30 03:52:59 Because I'm lost Jan 30 03:53:05 what u doing? Jan 30 03:53:07 I got a json whos' value is an array Jan 30 03:53:13 and I don't know how the hell to get at it Jan 30 03:53:24 get frddie Jan 30 03:53:28 should sort it out Jan 30 03:53:38 * xcomp runs Jan 30 03:53:38 egaudet: I think there's good examples in the libpurple-adapter Jan 30 03:53:40 all I got for gets is string, int, bool Jan 30 03:53:50 egaudet: look at the mjson patch its got really good examples Jan 30 03:54:02 for doing more complex stuff Jan 30 03:54:05 k Jan 30 03:54:14 egaudet: or u could just get the array as a spring Jan 30 03:54:20 and break it up your self Jan 30 03:54:26 doesn't get it as a string Jan 30 03:54:52 egaudet: you can get any payload as a string instead of a json object Jan 30 03:54:55 I was going to just iterate the damn json structure myself but if there's a better way then I want to do that Jan 30 03:55:08 ooh turn the payload into the string hmm Jan 30 03:55:11 yeah in the patch comments Jan 30 03:55:42 could u scanf it? Jan 30 03:56:12 is there a function like that for strings Jan 30 03:57:54 ooh, ~$100 ARM9 laptop in china http://armdevices.net/2010/01/29/android-laptop-review-hivision-pws700ca/ Jan 30 03:59:38 jauderho: we do need erlang for webos (it's already in optware) for ubuntu one support Jan 30 04:00:22 ubuntu one was written in erlang? Jan 30 04:01:53 rwhitby: wow Jan 30 04:02:03 first I've heard of this Jan 30 04:04:16 destinal: I just listened to the FLOSS weekly podcast #99 which interviewed the author Jan 30 04:04:22 (on the plane) Jan 30 04:06:45 rwhitby: very cool Jan 30 04:07:08 looks like canonical is going to be doing a music store probably with this tech as the dropboc Jan 30 04:07:14 yep Jan 30 04:07:14 s/boc/box/ Jan 30 04:07:15 destinal meant: looks like canonical is going to be doing a music store probably with this tech as the dropbox Jan 30 04:07:39 rwhitby: so why not pre? I like the way you think Jan 30 04:07:45 when I heard it I thought "hey, we've got erlang, so maybe we can get a ubuntu one client on the Pre quickly" Jan 30 04:08:51 the client? Jan 30 04:09:25 the client is erlang? Jan 30 04:10:19 can we port dropbox to the pre? Jan 30 04:13:10 that probably has a higher user base compared to ubuntu one Jan 30 04:15:13 xcomp: i extracted those headers from the SDK just like you said and ./configure still does not see them Jan 30 04:15:29 what's the error? Jan 30 04:15:51 gkatsev: dropbox is open source? I didn't think so. Jan 30 04:16:21 yeah, i dont think so. but if it is possible to port it, would be cool Jan 30 04:16:28 I have a quick question. Currently, my pre is running 1.2.1 (disabled automatic updates), I want up upgrade (1.3.x) by cleaning out everything with the WebOS doctor. Jan 30 04:16:41 dropbox has an opensource gui, but closed source core Jan 30 04:16:41 rwhitby: looks like .py and .pyc files Jan 30 04:16:59 xcomp: http://dpaste.com/152434/ Jan 30 04:17:16 gkatsev: we'll do it after we port Skype ... Jan 30 04:17:25 lol Jan 30 04:17:27 The question I have is how can I save the phone calllog, sms messages, bookmarks before I wipe everything? Jan 30 04:17:39 PuffTheMagic: did you get EGL/egl.h from that directory? Jan 30 04:17:41 that would be cool as well Jan 30 04:17:50 xcomp: yeah Jan 30 04:18:00 and it's in staging/armv7/usr/include/EGL Jan 30 04:18:00 Z-Dawg: saving is easy - rsync /var and /media/internal to another computer Jan 30 04:18:08 as well aass everything else -- eglext.h, eglplatform.h Jan 30 04:18:08 rwhitby: heh, we could probably just run skype (with some special environment setup and binaries from that one arm phone) Jan 30 04:18:32 you need everything in the Include directory or it will fail Jan 30 04:18:39 xcomp: something about those configure checks are not playing well with my CC env Jan 30 04:18:51 cause it should at least see the file Jan 30 04:18:56 rwhitby: Do you know the files in particular under /var that are required? Jan 30 04:19:00 and maybe not pass another test Jan 30 04:19:02 rwhitby: or maemo bins actually I forgot they had skype Jan 30 04:19:15 PuffTheMagic: even if the file is there, if a file it depends on/includes is missing it will fail Jan 30 04:19:31 that's why I'm asking whether you copied the entire contents of the Include/ directory over Jan 30 04:19:35 Z-Dawg: I do not. That's why I said saving was easy ;-) Jan 30 04:19:44 yea i copied it all Jan 30 04:20:06 PuffTheMagic: ok, then export CFLAGS="-I/usr/local/include -g -O2" and try configure again Jan 30 04:20:13 rwhitby: hehe..Just want the exact files and leave out any other junk :-) Jan 30 04:20:35 xcomp: what is that gonna solve Jan 30 04:20:39 i have nothing in that include dir Jan 30 04:20:45 Z-Dawg: you do that in the restore step Jan 30 04:20:55 PuffTheMagic: aren't you in sb2? Jan 30 04:21:00 xcomp: no Jan 30 04:21:16 i am doing a "real" cross-compile Jan 30 04:21:23 rwhitby: True Jan 30 04:21:23 PuffTheMagic: hey now Jan 30 04:21:51 PuffTheMagic: then change /usr/local/include to wherever your staging include dir is Jan 30 04:21:56 and put it in CFlags Jan 30 04:22:30 xcomp: my include dirs should be fine considering it works for everything else Jan 30 04:22:32 real mean edit the gcc binary by hand to change the include path, none of this sb2 sissy stuff Jan 30 04:22:35 its just the GL stuff that dont work Jan 30 04:23:02 PuffTheMagic: not necessarily ... the rest of the checks are not cflags dependent Jan 30 04:23:08 they all use pkg-config Jan 30 04:23:41 rwhitby: he doesn't want to use sb2, hence his woes. :) Jan 30 04:23:55 I really hate this json stuff I think I might go to strings Jan 30 04:24:09 There's nothing to explain WTF to do with a JSON_ARRAY Jan 30 04:24:11 xcomp: PuffTheMagic doesn't have woes, he is a cross-compiling hacker extroadinaire Jan 30 04:24:19 heh Jan 30 04:24:34 but as of right now, he can't build evas :) Jan 30 04:24:42 except i think the GL stuff is using pkg-config Jan 30 04:24:48 egaudet: creating or parsing arrays? Jan 30 04:25:59 egaudet: use the example in the comments Jan 30 04:26:24 egaudet: you basically have to walk the json object Jan 30 04:26:27 like xml Jan 30 04:26:27 PuffTheMagic: not the evas check stuff Jan 30 04:26:45 no pkgconfig check there Jan 30 04:27:10 ok I guess a JSON_ARRAY is very intuitive Jan 30 04:27:21 because it's an.... array of json lol Jan 30 04:27:39 egaudet: look at it this way, would you rather use XML? :P Jan 30 04:27:55 PuffTheMagic, the example had no arrays :p Jan 30 04:28:38 heck no, I just wish mjson told me "hey I know the whole world likes to be counter-intuitive, but literally it's an array just go to the next one" Jan 30 04:28:40 rats ... raster slips away just as I remember what I wanted to talk to him about all day Jan 30 04:29:09 egaudet: hehe Jan 30 04:29:41 xcomp: is the EGL check relative to ${DATADIR}? Jan 30 04:30:00 PuffTheMagic: has absolutely nothing to do with it Jan 30 04:30:09 it's a simple AC check Jan 30 04:30:18 it tries to compile a simple program with the file included Jan 30 04:30:22 well i dont get how it could fail Jan 30 04:30:25 like it is Jan 30 04:30:26 if it fails, it comes back and says no Jan 30 04:30:42 well something is wonky about how that check is working Jan 30 04:30:44 PuffTheMagic: I would suggest reading config.log and finding out exactly what went wrong Jan 30 04:31:04 that check should not be absolute path Jan 30 04:31:24 PuffTheMagic: if EGL/egl.h is in the include path at the time of configure Jan 30 04:31:27 it will pass Jan 30 04:31:28 otherwise, it wil lfail Jan 30 04:31:33 that's why I recommended setting CFLAGS Jan 30 04:31:52 rwhitby: Looks like the SMS and Phone call log are in /var/luna/data/dbdata/PalmDatabase.db3 Jan 30 04:32:11 my cflags are prefectly fine Jan 30 04:32:49 -I/usr/armv7l-softfloat-linux-gnueabi/usr/include Jan 30 04:33:02 which is where it is Jan 30 04:33:07 So, what's the catchy google-searchable name for this new theming stuff? Jan 30 04:33:17 PuffTheMagic: ok ... then only config.log can tell you what went wrong Jan 30 04:33:22 rwhitby: Premed Jan 30 04:33:38 Themes decorate Pre's and Pixi's. A Pre is like a river pebble. A pixi is a fairy. Jan 30 04:33:57 PuffTheMagic: you're doing pre-med now? Jan 30 04:34:45 "premed" fails the google searchable test Jan 30 04:36:05 xcomp: configure:23142: checking GL/gl.h usability Jan 30 04:36:05 configure:23142: armv7l-softfloat-linux-gnueabi-gcc -c -O3 -g -pipe -march=armv7-a -mtune=cortex-a8 -mthumb -fomit-frame-pointer -I/usr/armv7l-softfloat-lin Jan 30 04:36:07 ux-gnueabi/usr/include/ -I/usr/armv7l-softfloat-linux-gnueabi/include/ -mfpu=neon -mfloat-abi=softfp -D_GNU_SOURCE conftest.c >&5 Jan 30 04:36:10 conftest.c:69:19: error: GL/gl.h: No such file or directory Jan 30 04:36:16 i still dont get my it keeps looking for that Jan 30 04:36:38 PuffTheMagic: what are your config options? Jan 30 04:36:39 the technology is cascading style sheets Jan 30 04:36:57 cascading themes .. Jan 30 04:37:25 xcomp: ./configure --prefix=/usr --build=i686-pc-linux-gnu --host=armv7l-softfloat-linux-gnueabi --mandir=/usr/share/man --infodir=/usr/share/info --datadir=/u Jan 30 04:37:25 sr/share --sysconfdir=/etc --localstatedir=/var/lib --disable-cpu-mmx --disable-cpu-mmx --disable-cpu-sse --disable-software-xlib --disable-software-16-x11 Jan 30 04:37:25 --disable-xrender-x11 --enable-software-sdl --disable-gl-x11 --disable-directfb --disable-fb --disable-software-xcb --disable-xrender-xcb --enable-image-loa Jan 30 04:37:25 der-gif --enable-image-loader-png --enable-image-loader-jpeg --enable-image-loader-tiff --disable-image-loader-xpm --disable-image-loader-svg --disable-cpu- Jan 30 04:37:28 altivec --enable-buffer --enable-image-loader-eet --disable-image-loader-edb --enable-font-loader-eet --enable-cpu-c --enable-scale-sample --enable-scale-sm Jan 30 04:37:31 ooth --enable-convert-8-rgb-332 --enable-convert-8-rgb-666 --enable-convert-8-rgb-232 --enable-convert-8-rgb-222 --enable-convert-8-rgb-221 --enable-convert Jan 30 04:37:34 -8-rgb-121 --enable-convert-8-rgb-111 --enable-convert-16-rgb-565 --enable-convert-16-rgb-555 --enable-convert-16-rgb-444 --enable-convert-16-rgb-rot-0 --en Jan 30 04:37:37 able-convert-16-rgb-rot-270 --enable-convert-16-rgb-rot-90 --enable-convert-24-rgb-888 --enable-convert-24-bgr-888 --enable-convert-32-rgb-8888 --enable-con Jan 30 04:37:40 vert-32-rgbx-8888 --enable-convert-32-bgr-8888 --enable-convert-32-bgrx-8888 --enable-convert-32-rgb-rot-0 --enable-convert-32-rgb-rot-270 --enable-convert- Jan 30 04:37:43 32-rgb-rot-90 --enable-pthreads --enable-gl-sdl --enable-gl-flavor-gles --enable-gles-variety-sgx Jan 30 04:37:46 holy Jan 30 04:37:47 oops Jan 30 04:37:49 didnt think it was that big Jan 30 04:38:00 PuffTheMagic: you don't need all that mess! :) Jan 30 04:38:13 PuffTheMagic: really it can't check for some of that? :( o_O Jan 30 04:38:26 PuffTheMagic: or just like being explicit? :) Jan 30 04:38:35 xcomp: its an ebuild Jan 30 04:38:41 im not doing this manually Jan 30 04:38:43 dtzWill: PuffTheMagic likes to set all the options just the way he wants them - he's a gentoo user Jan 30 04:38:50 PuffTheMagic: the ebuilder needs an education Jan 30 04:39:04 some of those --enables are not necessarily the most optimal Jan 30 04:39:25 rwhitby: :) Jan 30 04:39:25 xcomp: i didnt write it Jan 30 04:39:29 i can modify it Jan 30 04:39:34 PuffTheMagic: I know ... Jan 30 04:39:40 all i did was tack on the gl-sdl stuff Jan 30 04:40:12 PuffTheMagic: you need --prefix $prefix --disable-static --enable-buffer --enable-software-sdl --enable-fb --enable-gl-sdl --enable-gl-flavor-gles --enable-gles-variety-sgx --disable-doc Jan 30 04:40:19 that's all Jan 30 04:40:46 no pthreads? Jan 30 04:41:05 if pthread is not enabled, it's because it's currently not the most optimal Jan 30 04:41:09 you want it off Jan 30 04:42:13 don't --enable something unless you actually require it for your specific purpose Jan 30 04:43:33 if in doubt check out the Makefile in cross-compile Jan 30 04:43:57 xcomp: thats usually what i do, but then again, in gentoo i use use flags to set those options and i didnt really pay attention to all the crazyness that was in there Jan 30 04:44:34 PuffTheMagic: yeah -- gentoo's ebuild maintainers are obviously zealous without knowledge Jan 30 04:44:57 yeah i realize, i've been using it for 10 years Jan 30 04:45:03 -O9 -funroll-loops -fskip-instructions Jan 30 04:45:35 btw, don't bother building evas with anything higher than -O2, it won't help :) Jan 30 04:45:45 xcomp: u dont have to tell me that Jan 30 04:45:48 im not a ricer Jan 30 04:46:32 yeah -- but your ebuild maintainer obviously is Jan 30 04:46:44 it was all working fine till i enabled sdl-gl Jan 30 04:47:22 i suggest setting your configure flags for every efl lib manually and ignore the ebuild parameters altogether Jan 30 04:47:57 xcomp: using exactly the flags you toldme to got it detecting EGL Jan 30 04:48:02 especially since, for example, you should pass --disable-doc since doc files are unnecessary Jan 30 04:48:55 xcomp: you are arguing witht he wrong person here Jan 30 04:49:07 i dont go messing with ebuilds till i have an issue Jan 30 04:49:49 PuffTheMagic: oh ... not arguing, just suggesting -- I was rather jolted by those evas configure options :) Jan 30 04:49:58 i can see that Jan 30 04:50:12 oh wow dtzWill Jan 30 04:50:18 ok so its still not finding gl.h Jan 30 04:50:24 it shouldn't Jan 30 04:50:24 I'm getting 100% in Pokemon Ruby now, with sound off Jan 30 04:50:25 and it dont like my libGLESv2.so Jan 30 04:50:30 gl.h is not used Jan 30 04:50:31 that i copied from the pre Jan 30 04:50:36 I can get 100% with sound on as well, but frameskip is heavy Jan 30 04:50:37 xcomp: well its looking for it Jan 30 04:51:42 xcomp: oh re: -O3 i see i do have that set, i was testing something out with fceux i think and forgot i left it there Jan 30 04:52:06 i usually alwas usein -O2 for everything but i think i was trying to prove to someone -O3 was worse Jan 30 04:52:08 you might need the LDFLAGS as well with -rpath Jan 30 04:52:51 i have LDFLAGS setup Jan 30 04:54:09 rwhitby: ubuntu one client appears to be c and python, maybe he was talking about erlang on the server side? Jan 30 04:56:04 destinal: maybe, haven't looked into it - listened to it on the last flight Jan 30 05:01:15 * rwhitby bbl - final leg on the journey home Jan 30 05:04:40 xcomp: sdl-gl is finally building now Jan 30 05:07:26 PuffTheMagic: excellent Jan 30 05:07:27 i was missing some .so's Jan 30 05:07:33 and I finally got my mom off the phone Jan 30 05:07:35 was nothing cflag related Jan 30 05:07:51 xcomp: now can u show me the line you were using to init evas Jan 30 05:07:52 i was going to ask if you remembered libEGL.so Jan 30 05:08:00 i was missing a few Jan 30 05:08:48 look at the expedite code Jan 30 05:09:01 grrr Jan 30 05:09:02 ;) Jan 30 05:09:40 i haven't implemented ecore support yet Jan 30 05:10:05 what in ecore needs to be connected Jan 30 05:10:19 ecore-evas Jan 30 05:10:35 yeah i just dont know what that does specifically Jan 30 05:11:15 oh Jan 30 05:11:31 PuffTheMagic: ecore-evas spares you having to interact directly with the underlying display system Jan 30 05:11:32 the ecore_evas_sdl_gl_new() function i need Jan 30 05:11:33 :D Jan 30 05:11:39 yeah, that function Jan 30 05:12:29 guess i cant do anything yet Jan 30 05:12:52 PuffTheMagic: it glues ecore and evas together Jan 30 05:13:14 ecore abstracts mainloops, events, display system Jan 30 05:13:31 (but just basic window system stuff like raw emnpty windows) Jan 30 05:15:24 egaudet: ping Jan 30 05:15:28 raster: i was going to ask if your maemo had sdl support and if perhaps you could use it to test sdl_gles Jan 30 05:15:50 its a piece of shit and has bricked itself\ Jan 30 05:16:06 uh oh ... lol Jan 30 05:16:11 (yes it bricked itself all on its own) Jan 30 05:16:18 the same one you were testing with yesterday? Jan 30 05:16:24 no Jan 30 05:16:29 that was another bit of hw Jan 30 05:16:30 :) Jan 30 05:16:34 oh :) Jan 30 05:16:35 PuffTheMagic, pong Jan 30 05:16:37 as yes... unreleased hw Jan 30 05:16:57 the4 thing still has a bit debug board extension growing out of its side Jan 30 05:16:58 naturally you always get to have all the fun Jan 30 05:17:17 but yummy.... amoled... mmmmmmm gurgle Jan 30 05:17:49 i'm guessing you haven't gotten around to building sdl for your new toy yet Jan 30 05:19:04 raster: custom chip? or major manufacturer? Jan 30 05:20:00 xcomp: not bothering to. we're not using sdl at all. its all x11 :) Jan 30 05:20:08 major manufacturer Jan 30 05:21:19 raster: well I meant sdl using x11, but yeah you don't have any reason to at this point Jan 30 05:21:49 i'm guessing it's not an omap Jan 30 05:22:05 although I don't know what new omaps are there that I don't know about yet, honestly Jan 30 05:22:28 xcomp: zero reason :) Jan 30 05:22:34 and its not omap.. correct Jan 30 05:22:38 its better Jan 30 05:22:58 (better than your omap3430 in the pre - for sure) Jan 30 05:23:11 yeah -- I saw the bench :) Jan 30 05:23:22 totally smoked the pre Jan 30 05:24:10 it has the advantage of coming out later - of course it can have a better soc Jan 30 05:24:37 until TI announces a 1.2GHz 45nm OMAP Jan 30 05:24:45 announces =- sure Jan 30 05:24:53 but when will a product SHIP with it Jan 30 05:25:02 announce is pointless Jan 30 05:25:05 Is pre-plus underclocked as well? Jan 30 05:25:06 that takes years Jan 30 05:25:12 i'm an product engineer, I know :) Jan 30 05:25:13 yup Jan 30 05:25:39 as such alrerady have dev-prototype devices with nvida tegra cortex-a9 jobs in them Jan 30 05:25:48 qualification/characterization/entitlement/ramp generally takes up to 3 years for a new platform Jan 30 05:25:57 (ie little tiny boxes with batteries in them - not final casings or boards of course) Jan 30 05:26:01 and new device on an established platform still takes 2 years Jan 30 05:26:22 i'm working on something that has a final casing and board its mostly fixing up niggling small hw issues and 99% is sofwtare Jan 30 05:26:22 ooh, cortex-a9 Jan 30 05:26:38 ok not 99% - 90% of the work left is software Jan 30 05:26:50 we've got new 45nm embedded flash stuff going now but it's going to be 2013 at least before that stuff hits market Jan 30 05:27:01 destinal: cortex-a9 is old new by now... its been shipping in socs for a while now Jan 30 05:27:05 not in products tho Jan 30 05:27:21 sure, but I don't have any dev boards in front of me :) Jan 30 05:27:30 xcomp: yeah. and thats just the soc's themselves - another 2 years for first products Jan 30 05:27:39 destinal: i dont have the tegra here either Jan 30 05:27:42 others play with that Jan 30 05:27:55 raster: yeah, another couple of years before the product hits catalog Jan 30 05:28:01 i'm more abotu the immediate goal of a product coming out with e on it... :) Jan 30 05:28:11 raster: super job security for people like me :) Jan 30 05:28:17 it's got a good chance Jan 30 05:28:50 as i'm floating around on this 22nd floor office in korea and have had to extend my stay to 3 weeks now to cover all the shit to do here before i go back to sydney to continue my evilness Jan 30 05:28:51 :) Jan 30 05:29:02 xcomp: hhahahahaha indeed Jan 30 05:29:24 raster: especially when your product doesn't go into consumer electronics and has average lifespan of 10 years :D Jan 30 05:29:44 raster: a mobile consumer device with e? (evas btw?) Jan 30 05:32:57 xcomp: which expedite profile was 320x480 Jan 30 05:33:02 all my apps and patches aren't showing up in preware ??????? Jan 30 05:33:18 PuffTheMagic: hvga Jan 30 05:33:59 xcomp: but consumer electronics si so much mroe fun Jan 30 05:34:09 puffthemagic : didn't that happen to you? Jan 30 05:34:15 raster: definitely ... don't get me wrong Jan 30 05:34:19 xcomp: it moves fast. next year i will have a bett er gpu, more cpu cores and horse power, more ram and flawsh... and more Jan 30 05:34:38 raster: I walk by the DSP/Da Vinci guys cubes every day and I get real jealous Jan 30 05:34:47 i cabn play the wait and see game of "well i9 cant make it do it now without the software ebing a horrible mess, but.. i can just wait 1 year and better hw will solve the problem" Jan 30 05:34:50 they always got gadgets to play with Jan 30 05:34:53 destinal: yes. Jan 30 05:35:05 i don't get gadgets Jan 30 05:35:08 destinal: with e. and evas. and edje. and elementary.. and... Jan 30 05:35:20 cool Jan 30 05:35:20 i get to do boring stuff Jan 30 05:35:37 raster: sounds like a cool platform Jan 30 05:35:39 read currents and measure voltages Jan 30 05:35:41 gadgets are cool. mmmmmmm TOY! Jan 30 05:35:41 bleh Jan 30 05:36:18 xcomp: yeah i dont get anything with gl-sdl Jan 30 05:36:22 works with sdl Jan 30 05:37:18 if i zoom out then in it looks garbled and dont move Jan 30 05:37:29 sounds like where you were Jan 30 05:37:39 yeah I just got garbage from the get go Jan 30 05:37:48 Wasn't someone talking about stuff not showing up installed in preware but it was still there??????????? Jan 30 05:37:50 PuffTheMagic: it works until u zoom in+out? Jan 30 05:38:04 (and what is "zoom in and out"?) Jan 30 05:38:11 rwhitby: no for me it was just a black screen Jan 30 05:38:13 err Jan 30 05:38:14 raster: he means flick into card mode/full screen Jan 30 05:38:15 raster: ^^ Jan 30 05:38:28 PuffTheMagic: aaah Jan 30 05:38:34 xcomp: ok Jan 30 05:38:39 cant help u there Jan 30 05:38:47 have u got any working gl-sdl demo apps? Jan 30 05:38:53 ie are u sure it works reliably at all? Jan 30 05:39:00 i spend my days finding bugs in gl drivers Jan 30 05:39:09 it works in x Jan 30 05:39:12 that's what I know :) Jan 30 05:39:22 lucky for me i get to fix them before products come out and stand on my soap-box and point the finger of blame @ the gl drive4r libs Jan 30 05:39:31 i was going to try to convert my old ematrix demo to the pre Jan 30 05:39:47 raster: there's no redirection Jan 30 05:39:55 (man some of the shit i've seen as attempts to fix the bugs... have been worse than the bugs originally - they just create even more bugs that are harder to solve) Jan 30 05:39:56 it's fullscreen or thumbnail Jan 30 05:40:15 tmzt: oooh.. no compositing at all Jan 30 05:40:25 tmzt: actually it's not a thumbnail ... it still runs in card mode unless the app actually stops Jan 30 05:40:27 that sucks Jan 30 05:40:42 xcomp: even with gl? Jan 30 05:40:51 yeah Jan 30 05:41:08 the app has to detect the card mode event and "pause" Jan 30 05:41:18 runs in card mode? Jan 30 05:41:28 like with the html scenes? Jan 30 05:41:30 yeah, sdl apps don't automatically stop in card mode Jan 30 05:41:44 the app actually has to do the pausing actively Jan 30 05:41:55 i'm pretty sure about that Jan 30 05:42:35 that's why you can run multiple dooms Jan 30 05:45:22 right Jan 30 05:46:25 hehe Jan 30 05:46:41 xcomp: sure, but if it decuides to keep rendering, it can, and cards will update - i've seen that work, but... does it work with gl? Jan 30 05:46:45 I wonder if this could be overloaded to get a single X server with mutliple windows as cards Jan 30 05:46:56 ie is sdl-gl known to work at all on the pre? Jan 30 05:47:21 then just start the one X server when any visible window (mapped, see compiz thread) is visible Jan 30 05:47:22 tmzt: it could be Jan 30 05:47:34 nothing i have seen in the palm luna ui cant be done in x Jan 30 05:47:55 (with sufficient extensions, wm, compositor etc. etc. work) Jan 30 05:48:12 sure, but there's no open luna Jan 30 05:49:07 I'm now wondering what to do about flash ads on websites, I think after what you said yesterday it's not coming to any device with a ten hour battery life that depends on video being decoded by dsp or hw Jan 30 05:49:36 seems to me a very valuable framework like mojo could allow for widgets and apps to work essentially the same way Jan 30 05:49:46 as selef contained enhanced html5 scenegraphs Jan 30 05:52:56 raster: all gl apps on the pre right now have to use gl at least the way I understand it Jan 30 05:53:05 because sdl is required for the luna integration Jan 30 05:53:34 so that includes all the games out there - nfs, call of duty etc. Jan 30 05:54:29 there's just something we're missing right now, and we'll eventually find it Jan 30 05:55:05 tmzt: i can tell u that flash is already on devices. n900. its on manny phones from samsung already out there. its there. its not being used primarily as a "ooh we need youtubeeezzzz roxxoorrz!" Jan 30 05:55:15 it happens to be able to do it Jan 30 05:55:33 and its used for a mountain of other things - even developi\ng entire uis for apps Jan 30 05:56:48 we'll get flash ... in a couple of weeks Jan 30 05:56:49 maybe less Jan 30 05:56:57 the big 1.4 Jan 30 05:57:27 xcomp: ok. so gle-s works. sdl-gl works too in a simple demo app? Jan 30 05:57:41 raster: that I haven't tried, and I should Jan 30 05:57:56 i think zsoc was working on that Jan 30 05:58:08 and he was able to get a demo going Jan 30 05:58:09 not sure Jan 30 05:58:20 now that you mention it ... let me poke around crosscompile Jan 30 05:59:08 ok Jan 30 05:59:13 well if a simple demo works Jan 30 05:59:30 then u can begin tyo trust the sdl+gl combo somewhat before u get to more complex things like evas Jan 30 06:00:44 there is a glues-demos in cross compile Jan 30 06:05:28 yeah, i'm running the cylinder demo right now Jan 30 06:05:30 definitely works Jan 30 06:05:42 now to look at the code and seek clues Jan 30 06:06:06 raster: to answer the other question, gl works in card mode but it's really slow Jan 30 06:06:24 regular sdl seems to be pretty slow in card mode too, though Jan 30 06:09:57 aaaaah Jan 30 06:10:13 last i saw of the pre luna code was some awful software handling of pixel data for cards Jan 30 06:10:18 irt was pretty bad Jan 30 06:10:27 like multiple copies where it could do 1 or 0 Jan 30 06:10:43 and yeah - all gooing thru ther cpu Jan 30 06:10:47 they need to rewrite luna with evas Jan 30 06:10:52 hehehehe Jan 30 06:10:57 then they could port to x! Jan 30 06:10:57 :) Jan 30 06:11:05 brilliant Jan 30 06:11:28 tho... i'm on the "screw that... just work on a product, out of the box... coming with e+x11" Jan 30 06:11:29 :) Jan 30 06:11:33 save the pain Jan 30 06:14:06 no porting Jan 30 06:14:18 native-out-of-the-box goodness Jan 30 06:14:28 what i see as mroe viable is adding as "webapp" launcher Jan 30 06:14:46 ie uses eflwebkit + adds the js extensions palm added (hookking to backend services) Jan 30 06:14:54 ie tis a brtowser just with no controls Jan 30 06:15:01 juts a window filled with html Jan 30 06:15:05 easy enough to do Jan 30 06:15:10 and now u re-created webos's app env Jan 30 06:15:24 u can have a different desktop ui thing than palm - that doesnt matter Jan 30 06:15:31 tho the palm ui for webos is well done Jan 30 06:16:08 yep Jan 30 06:16:28 you know technically Jan 30 06:16:41 we could have an alternative efl ui for palm pre Jan 30 06:16:50 just kill luna Jan 30 06:17:02 use eflwebkit for the browser Jan 30 06:17:14 heh Jan 30 06:17:22 everything fb Jan 30 06:18:31 can be domne Jan 30 06:18:36 tho i'd suggest using x Jan 30 06:18:49 yeah, have windowing Jan 30 06:18:54 so u get the ability to put on all the existing world of x apps Jan 30 06:19:09 be they efl, gtk, qt, sdl, raw opengl/es+x11 etc. Jan 30 06:19:33 i understand why some peolpe think that u need a whole new display system to cover uis on devices Jan 30 06:19:41 but really - the device uis are doable within x11 framewiork Jan 30 06:19:47 u have lots of stuff u dont use/need Jan 30 06:19:56 but thats just some extra flash benig used Jan 30 06:19:58 oone day u may need it Jan 30 06:20:55 for example - theres already an android runtime for ubuntu Jan 30 06:20:59 ie android windows in x windows Jan 30 06:21:14 and the rest of android stuck on a host full linux system Jan 30 06:21:26 u can go on with the advantages of going x11 Jan 30 06:21:30 like ease of development Jan 30 06:21:42 needing no simulators/emulators Jan 30 06:23:29 does android use X11? I didn't know Jan 30 06:23:58 no - it doesnt Jan 30 06:24:15 but theres a port of the runtime to use x11 as ther display surface Jan 30 06:24:34 android did its own "ooh lets make our own display system too" thing Jan 30 06:33:47 raster: does the absence of OES_texture_npot and texture_rectangle pose any issues? Jan 30 06:35:12 well if no npot - it will fall abck to rounding texture up to powers of 2 Jan 30 06:35:15 so it will work Jan 30 06:35:20 it wont use textrue rectangles Jan 30 06:36:33 right now I'm using the exact same init flags as the glues demo without success Jan 30 06:40:09 xcomp: well not sure Jan 30 06:40:13 it should work fairly easiyl Jan 30 06:40:23 unless u forgot something about the waysdl-gl works Jan 30 06:40:24 as i said Jan 30 06:40:31 if u have a gl-sdl examle that works Jan 30 06:40:34 look at that for ideas Jan 30 06:40:46 yeah... that's what i'm doing right now Jan 30 06:42:41 PuffTheMagic, fix inotify Jan 30 06:44:02 init definitely seems fine Jan 30 06:44:16 reports depth 32, opengl es 2.0 for version string Jan 30 06:48:17 xcomp: it may be egl context or surface issues Jan 30 06:48:29 i dont knwo how much sdl wraps and does for u Jan 30 06:48:32 raster: yeah...that's what i'm looking into now Jan 30 06:48:43 i think i'm comfortable with the sdl wrapper not being the problem at this point Jan 30 07:08:20 YES Jan 30 07:08:23 I FIXED IT Jan 30 07:08:54 key events don't seem to be working Jan 30 07:08:57 but it's displaying Jan 30 07:09:09 stupid output resize was killing the gl context :) Jan 30 07:11:31 wow image blend unscaled is slow Jan 30 07:11:35 6.75 fps Jan 30 07:11:57 that ain't right Jan 30 07:12:52 raster: yep that code you gave me was to make a test of a luna like system then I found out I would need a backend for webkit and some way to render to a pixbuf Jan 30 07:13:18 the good news is my final test does actually work now with some help from phh in #htc-linux Jan 30 07:13:27 final blitting test Jan 30 07:14:39 where's the android runtime for ubuntu? Jan 30 07:14:50 could it be ported back to arm? Jan 30 07:15:28 raster: just found the chromium tree of the qualcomm msm video driver, seems the Chrome tablets/smartbooks will use Xorg Jan 30 07:16:05 this could make a normal X based linux port trivial if they live up to the specs for the open bootloader protocol Jan 30 07:17:17 https://www.codeaurora.org/gitweb/quic/xwin/?p=xf86-video-msm.git;a=log;h=refs/heads/chromium Jan 30 07:23:36 evening all Jan 30 07:34:29 tmzt: chrome tablets? Jan 30 08:06:36 Is it just me or do apps in the "Palm-Beta" feed not work? Jan 30 08:06:55 Or did I miss something.. Jan 30 08:20:55 raster: so it works, but the bench is not good -- 6.31 fps for image blend Jan 30 08:28:33 I wonder if it would be possible to get zsnes working on the pre.. Jan 30 08:29:04 nope Jan 30 08:29:16 didn't notice all the x86 assembly :( Jan 30 08:29:41 xcomp: oooh what was wrong - why didnt it work? Jan 30 08:29:56 resize was busting the context Jan 30 08:30:04 had to add a check for SDL_RESIZABLE Jan 30 08:30:32 i wonder if that also has something to do with the poor bench (although I don't think it should) Jan 30 08:30:42 i know it starts of with w/h of 1x1 Jan 30 08:30:46 aaaaah Jan 30 08:30:56 it shouldnt Jan 30 08:31:00 but u never know drivers Jan 30 08:31:06 can ui do a full Jan 30 08:31:18 expedite -e gl_sdl -a > file.csv Jan 30 08:31:19 ? Jan 30 08:31:27 (rubn all tests and upload the csv) Jan 30 08:31:29 ? Jan 30 08:31:40 yeah give me a few minutes and I'll have it up Jan 30 08:32:53 cool Jan 30 08:33:08 hmm, I should fix the engine info output to use stderr **** BEGIN LOGGING AT Sat Jan 30 08:39:17 2010 Jan 30 08:39:19 xcomp: yes, but its probably still alpha and unoptimised Jan 30 08:39:52 raster: perhaps certain gl stuff is being accelerated and some other things aren't Jan 30 08:44:28 raster: ooh, something segv'ed Jan 30 08:45:42 image data argb Jan 30 08:46:55 unfortunately gdb is useless Jan 30 09:03:07 xcomp: on the sgx it will either be all or nothing as the same acceleration core exists regardless of windowing system front end Jan 30 09:03:15 its plit in the driver for vendors to poer to their display infra Jan 30 09:03:31 as for image data argb... interesgin Jan 30 09:03:36 gl segged or software Jan 30 09:03:40 run afull software test first Jan 30 09:03:43 then try gl Jan 30 09:03:46 hrm Jan 30 09:03:50 For those that are interested re Facebook and timezone issues: http://geeklog.lucid.net.au/article.php?story=20091211084306281 Jan 30 09:04:01 raster: gl segged Jan 30 09:04:31 software is fine Jan 30 09:06:17 lingfish: that is some useful info Jan 30 09:06:17 xcomp: well upload a performance dump iof software Jan 30 09:06:19 lets see Jan 30 09:06:32 yeah software should be done soon Jan 30 09:06:38 xcomp: thanks... took a LONG while to research and nut it all out Jan 30 09:06:46 But works a great, just did it via an ssh to my Pre :) Jan 30 09:06:49 lingfish: now we need a script Jan 30 09:06:50 :) Jan 30 09:06:59 God I wish Facebook would get a f'ing CLUE on storing epoch timestamps! Jan 30 09:07:08 facebook api is a freaking mess Jan 30 09:07:20 Morons store in epoch format, but not UTC?!? Noobs. Jan 30 09:07:43 it's so bad ... for a while there they were breaking api every other week Jan 30 09:07:46 Well, as stated in the blog... could just be cron'd I guess... but there are risks. Jan 30 09:08:14 Oh I agree... but man, something as simple as an epoch they even mangled... and everyone hates on Palm for it Jan 30 09:10:59 yeah ... it's facebook's fault clearly Jan 30 09:11:16 but still, it would be nice for palm to attempt to work around it Jan 30 09:11:22 Indeed Jan 30 09:11:31 and hopefully my article is noticed/helps in that regard. Jan 30 09:12:07 raster: i'm choosing to not muck with the enum order for Ecore_Evas_Engine_Type ... or would that be OK? Jan 30 09:14:05 yeah ... it's probably a bad idea Jan 30 09:16:33 xcomp: dont muck unless u have to Jan 30 09:16:35 dont think u need to Jan 30 09:16:40 yeah don't need to Jan 30 09:16:53 beyond a perverse pleasure gained in having your enums listed alphabetically or something Jan 30 09:28:06 raster: http://xcomputerman.com/files/palmpre_expedite_software_sdl_bench.csv Jan 30 09:32:40 heh, nice broken link Jan 30 09:32:56 ick Jan 30 09:33:10 raster: http://xcomputerman.com/files/efl_webos/palmpre_expedite_software_sdl_bench.csv Jan 30 10:00:16 xcomp: totally kill your results.. ehre... in softtware Jan 30 10:00:36 yeah I'm sure :) Jan 30 10:00:58 doing a run in softwre x11 Jan 30 10:01:01 running 32bpp btw Jan 30 10:01:05 dont know what pre runs Jan 30 10:01:14 it's 32bpp as well Jan 30 10:01:20 ok Jan 30 10:01:22 so fair comparison Jan 30 10:01:47 i'm doing my tests in an hvga window in a 480x800 scren Jan 30 10:02:24 so i probably would get a bit faster if i could run the screen at a lower res to save some refresh bandwdth Jan 30 10:03:22 40.68 , Widgets File Icons Jan 30 10:03:22 100.35 , Widgets File Icons 2 Jan 30 10:03:26 for starters Jan 30 10:03:28 :) Jan 30 10:04:04 what are you comparing it with, and x86? Jan 30 10:04:09 no Jan 30 10:04:15 another cortex-a8 soc Jan 30 10:04:19 ah Jan 30 10:04:26 clocked a little higher (800mhz) Jan 30 10:04:39 but runing x11 Jan 30 10:04:55 (i love to destroy myths that x is slow... which it isnt...) Jan 30 10:04:55 what is the gpu running at on that one? Jan 30 10:04:56 SGX based? Jan 30 10:05:01 tyhats software rendering Jan 30 10:05:04 ah Jan 30 10:05:06 i'll do a gpu run after Jan 30 10:05:06 yes Jan 30 10:05:09 sgx based Jan 30 10:06:22 (it takes a while to run the tests) Jan 30 10:08:18 come on expedite.. do your thing! Jan 30 10:08:54 xcomp: u clocked @ 500mhz still? Jan 30 10:12:15 raster: i think sdl bumps the cpu up to 550, but i'm not sure Jan 30 10:12:42 cat check /sys/devices/system/cpu/cpu0/cpufreq... ? Jan 30 10:12:56 let me try Jan 30 10:13:34 it bumps it up to 600 Jan 30 10:13:34 almsot finished sofware test Jan 30 10:13:44 aaah ok' Jan 30 10:13:45 so it's definitely running at 600MHz Jan 30 10:13:53 good - we know your basseline Jan 30 10:15:40 hmmm..... Jan 30 10:16:03 emotion_object_init() fails, what's the most likely reason? (emotion_test works) Jan 30 10:24:17 xcomp: wrong module name? Jan 30 10:24:41 "xine" Jan 30 10:25:29 hmmm Jan 30 10:25:31 that shoudl work Jan 30 10:25:36 if u have a xne module built? Jan 30 10:26:09 yeah Jan 30 10:26:28 not sure Jan 30 10:26:32 eina_init not called? Jan 30 10:26:35 some other efl init? Jan 30 10:27:02 ecore_evas Jan 30 10:27:19 i added eina_init(), no dice Jan 30 10:30:14 xcomp: ok Jan 30 10:30:15 rresults Jan 30 10:30:24 software-x11 hammers your pre Jan 30 10:30:31 rect solid frew is slower Jan 30 10:30:31 5% Jan 30 10:30:34 everything else faster Jan 30 10:30:44 some tests 169% faster Jan 30 10:30:59 one is even 302% faster Jan 30 10:31:11 oooh 981% faster fo another one Jan 30 10:31:43 gl vs software is a mixed bag Jan 30 10:31:58 raster: sometime later I'll build with sdl-primitive and see if there is an improvement in performance Jan 30 10:31:58 some tests 200-900% faster Jan 30 10:32:10 one is 81% slower even Jan 30 10:32:38 overall gl is slower than software in expedite Jan 30 10:32:44 but then expedite isnt a realistic usage test Jan 30 10:32:51 its an isolate the operations test Jan 30 10:40:05 yeah, gl can be mixed Jan 30 10:42:29 yeah Jan 30 10:42:34 good u have a choice Jan 30 10:42:41 choose the ones that does th ebest for u Jan 30 10:42:45 need to run Jan 30 10:42:47 ciao! Jan 30 10:48:13 sup? Jan 30 10:52:43 not much Jan 30 11:01:35 Does anybody know how I can access the version number of my app in Ares? I'm looking for a similar command to Mojo.Controller.appInfo.version - any hints? Jan 30 13:07:14 iSmite: i just watched the palm keynote thingy from the 7th, interesting. Jan 30 16:01:33 What command line tool can I use to remove a patch? Jan 30 16:04:42 I have a patch I can't remove via Preware and I just wanted to force it since I checked and it isn't actually applied (not sure why) Jan 30 16:07:28 riskable: the suggestion is to remove all patches with epr Jan 30 16:07:47 I just have one patch to remove Jan 30 16:08:05 ...and I want to know how it is done anyway Jan 30 16:08:06 riskable: otherwise ipkg -o /media/cryptofs/apps remove patchname will remove without running the pre removal script Jan 30 16:08:14 Ahh, excellent Jan 30 16:08:35 it sounds like you don't want to remove the patch though, just tell preware it's not there? Jan 30 16:09:55 jettero: well that's all that really does if the patch never installed properly Jan 30 16:10:09 That seems to have done the trick Jan 30 16:10:53 Thanks Jan 30 16:11:07 Wish that was in some documentation somewhere Jan 30 16:12:18 riskable: there's good reason it isn't. Jan 30 16:13:08 riskable: you have passed to competence test by logging into irc, and finding the patch package name :p Jan 30 16:13:17 LOL Jan 30 16:13:39 Hmm... Preware patches won't install saying lsdiff is missing Jan 30 16:15:08 maybe lsdiff is missing Jan 30 16:15:45 Yeah, well, there's no "lsdiff" listed in "ipkg-opt list". I installed diffutils thinking maybe it's in there Jan 30 16:16:17 I'm a little surprised preware didn't install it for you Jan 30 16:16:26 something is fishy Jan 30 16:16:29 No "lsdiff" in "List of Everything" in preware Jan 30 16:17:07 What repository is it in? Jan 30 16:17:24 BTW: I'm surprised Preware didn't install it for me too Jan 30 16:17:34 riskable: you webos-internals feed is off or corrupt Jan 30 16:17:41 s/you/your/ Jan 30 16:17:41 zsocwerkin meant: riskable: your webos-internals feed is off or corrupt Jan 30 16:17:54 Hmm Jan 30 16:18:21 go to 'manage feeds', check that it is switched on, and there is a valid url below it Jan 30 16:18:52 Yeah, it is... Looking at the feed in Preware shows the webos-internals feed URL as just "feed" Jan 30 16:19:01 Bizarre Jan 30 16:20:26 The optware feed is messed up too... I'm just going to reinstall preware Jan 30 16:23:24 riskable: install via the boot-strap _please_ Jan 30 16:23:42 That's exactly what I did Jan 30 16:23:56 kk. that should fix it Jan 30 16:35:21 I still can't install any patches... They all require org.webos-internals.lsdiff as a dependency Jan 30 16:35:38 I installed patchutils via ipkg-opt (which includes lsdiff) but that didn't work Jan 30 16:36:08 riskable: webos-internals feed working? Jan 30 16:36:42 last I checked it was Jan 30 16:36:46 It appears to be Jan 30 16:37:04 riskable: does lsdiff appear in list of everything? Jan 30 16:37:18 No it doesn't Jan 30 16:38:19 I just told Preware to update feeds manually and now it shows up :) Jan 30 16:38:30 Must've just been a network error on the webos-internals feed or something Jan 30 16:39:26 riskable: er, more likely a persistent setting in preware thinking it was already updated, from before you reinstalled ipkgservice and fixed the feed :) Jan 30 16:39:37 Ahh, probably Jan 30 16:39:44 I do have it set to only update once daily now that I think about it Jan 30 16:41:47 :) Jan 30 16:44:46 Thanks for your help... I'll be back later I think Jan 30 16:52:29 whats new Jan 30 17:24:34 foo Jan 30 17:24:43 bar Jan 30 17:36:05 wtf is ministat? Jan 30 17:51:06 PuffTheMagic: counts minutes used and texts sent Jan 30 17:51:29 from phone's own database Jan 30 17:53:06 o Jan 30 17:56:03 hrm, my wife's pixi isn't playing the "toss the card into heaven" sound... but other sounds are working Jan 30 17:56:07 any idea how to deal with that? Jan 30 17:56:51 system sounds up? Jan 30 17:57:16 dunno, how do you check that? Jan 30 17:57:50 go to Sounds and Ringtones Jan 30 17:57:53 oh, snap, found it. thanks Jan 30 17:58:04 it was your tip that obviated it, thanks Jan 30 17:58:37 you know I'm not sure I used that word right, even having looked it up Jan 30 17:58:53 heh common mistake cause it's indepedent of ringer volume and updates seem to practically mute it Jan 30 17:59:06 yeah, seems media volume, ringer volume and system volume are all separate Jan 30 17:59:12 I wish I could find one place to change them all Jan 30 18:00:51 music volume is kinda weird cause it varies depending on if headphones are plugged in or not Jan 30 18:04:41 PuffTheMagic, aufs Jan 30 18:04:57 what about it Jan 30 18:05:05 where do I get it Jan 30 18:05:18 I want to try some of the unionfs stuff out Jan 30 18:05:23 http://aufs.sourceforge.net/ Jan 30 18:05:26 i tried funionfs but Jan 30 18:05:28 NO Jan 30 18:05:31 NO Jan 30 18:05:32 NO Jan 30 18:05:37 only use aufs Jan 30 18:06:16 that site says aufs2 2.6.27 and later Jan 30 18:06:21 I can't find aufs1 anywhere Jan 30 18:06:21 aufs .> unionfs Jan 30 18:06:57 murderfs > * Jan 30 18:12:58 egaudet: http://aufs.cvs.sourceforge.net/viewvc/aufs/ Jan 30 18:13:16 egaudet: use that or you might have to use unionfs Jan 30 18:13:24 but if u have the choice use aufs Jan 30 18:15:26 egaudet: should be able to build that as a module out of tree Jan 30 18:15:40 but that will require you have kernel headers installed to link against Jan 30 18:15:50 and a palm kernel built i believe Jan 30 18:17:24 egaudet: it might just be easier to use unionfs at this point Jan 30 18:18:25 egaudet: but we could even mix and match themes with this :D Jan 30 18:19:13 egaudet: i can get a unionfs module in preware for you ;) Jan 30 18:19:25 it will take a little more work to get it into sb2 setup Jan 30 18:20:59 omg omg omg Jan 30 18:21:10 egaudet: unionfs-fuse Jan 30 18:21:11 :D Jan 30 18:21:13 even better Jan 30 18:22:08 egaudet: http://podgorny.cz/moin/UnionFsFuse Jan 30 18:26:23 PuffTheMagic, yea that's what I was trying to build Jan 30 18:26:31 nice Jan 30 18:26:45 but I dunno much about fuse Jan 30 18:26:54 its easy Jan 30 18:27:01 u dont need to know nothing about fuse Jan 30 18:28:54 egaudet: the palm kernel already has fuse in it Jan 30 18:28:59 so u dont need to build that Jan 30 18:28:59 /var/unionfs: /usr/lib/libfuse.so.2: version `FUSE_2.8' not found (required by /var/unionfs) Jan 30 18:29:42 which version of unionfs-fuse did u build? Jan 30 18:29:58 0.23 Jan 30 18:30:29 and u linked to a too new fuse Jan 30 18:31:11 I linked to the fuse on the pre Jan 30 18:31:55 or I thought I did woops Jan 30 18:32:01 pre version is 2.7 Jan 30 18:32:03 you are right I screwed up my env and got 2.8 in there Jan 30 18:32:36 PuffTheMagic, Funionfs vs. unionfs-fuse any idea what the difference Jan 30 18:33:07 no Jan 30 18:33:14 i never heard of funionfs Jan 30 18:33:31 but its not in portage Jan 30 18:33:35 so it must suck Jan 30 18:34:30 alright now to find out how to use unionfs Jan 30 18:35:19 its pretty easy Jan 30 18:35:22 basically like a tmpfs Jan 30 18:35:27 except its transparent Jan 30 18:35:38 there are some mount options too Jan 30 18:35:43 but its basically pretty easy Jan 30 18:45:12 so we can now do something like Jan 30 18:45:19 I think aufs / aufs2 is the way forward Jan 30 18:45:31 unionfs -o nonempty /media/internal/.usr /usr Jan 30 18:45:46 unionfs is deprecated, right? Jan 30 18:45:49 all patching and everything for the future should use this Jan 30 18:45:52 it's unionfs-fuse Jan 30 18:46:02 and it's working Jan 30 18:46:25 ah Jan 30 18:46:53 doesn't that put the whole fs in userland? wouldn't that slow things down? Jan 30 18:47:26 destinal: it dont slow down Jan 30 18:47:30 hmm Jan 30 18:47:33 good point Jan 30 18:47:36 its all "code on the userland" Jan 30 18:47:40 its all on the hd Jan 30 18:47:49 same bottle necks Jan 30 18:47:57 for some reason I figured that kernel level accesses could be optimized to be faster Jan 30 18:47:58 no difference than if it were bundled in the kernel Jan 30 18:48:01 or a "kernel module" Jan 30 18:48:03 well it still has to do a lot of system calls so it's not as efficient as kernel Jan 30 18:48:21 destinal: fuse extends kernel level access to the VFS by 1 level of abstraction Jan 30 18:48:25 whether we'll notice the impact or not is something we need to test Jan 30 18:48:30 its not gonna significantly impact anything Jan 30 18:48:43 PuffTheMagic: so the syscalls can slow a bit but flash is already the bottleneck Jan 30 18:49:19 destinal: i think there are also speedups to be had from fuse Jan 30 18:50:05 destinal: http://podgorny.cz/moin/UnionFsFuse Jan 30 18:50:22 for some reasons why fuse might be better than in kernel Jan 30 18:50:45 right, context switches Jan 30 18:50:54 destinal: i like how the root fs can be mounted second Jan 30 18:50:57 that's what I was thinking of but couldn't say Jan 30 18:51:05 that means we could mount this over /media internal Jan 30 18:51:13 and it wouldnt break in usb mode Jan 30 18:51:45 (not that i am saying we need that for themes) Jan 30 18:51:59 ok so it works fine in my test dirs, now to test on all of /usr Jan 30 18:52:17 egaudet: why not / ? :) Jan 30 18:52:41 destinal: i wouldnt Jan 30 18:53:32 destinal: egaudet is going at this like there is a torch under his ass Jan 30 18:53:35 are all our patches in / ? Jan 30 18:53:37 err /usr? Jan 30 18:53:46 not all Jan 30 18:53:59 I read something somewhere about some unionfs freezing on / Jan 30 18:54:15 we dont want /var Jan 30 18:54:16 that might have been funionfs though and not unionfs-fuse but I assume they might run into similar issues Jan 30 18:54:18 or /dev Jan 30 18:54:21 or /sys Jan 30 18:54:23 or proc Jan 30 18:54:24 right Jan 30 18:54:27 or tmp Jan 30 18:54:34 so I think hand-picking what we want like /usr and /etc Jan 30 18:54:52 egaudet: u know Jan 30 18:55:04 hmm what bout /media/cryptofs/apps Jan 30 18:55:04 unionfs-fuse might be nice Jan 30 18:55:14 but i see us using unionfs for a more stuff Jan 30 18:55:28 i don't get it? Jan 30 18:55:42 so maybe we should get the kmod working Jan 30 18:55:52 I don't see any problems with unionfs-fuse Jan 30 18:56:20 egaudet: im thinking of using unionfs instead of rpath ;D Jan 30 18:59:37 hmm it's not working right on /usr Jan 30 18:59:58 how are u doing it Jan 30 19:00:20 unionfs -o nonempty /var/usr_ufs=rw /usr Jan 30 19:01:28 so /var/usr_ufs is your test dir that is getting mounted rw over /usr? Jan 30 19:02:05 yea Jan 30 19:02:17 does /usr have to be ro first? Jan 30 19:02:40 what do you mean Jan 30 19:02:55 it works on /var/blah1 /var/blah2 Jan 30 19:02:55 is root mounted ro? Jan 30 19:03:03 I tried mounted rw and ro Jan 30 19:03:13 WOOHOO! Jan 30 19:03:15 what about just doing /usr/palm Jan 30 19:03:27 * dBsooner won a PC.net Contest! Jan 30 19:03:39 what u win? Jan 30 19:03:50 Spare battery and charger Jan 30 19:03:57 nope Jan 30 19:06:42 I think we need some aufs Jan 30 19:06:47 * iSmite is listening to Insight by Depeche Mode from Ultra Jan 30 19:06:52 shit sorry wrong room Jan 30 19:07:04 * iSmite fails Jan 30 19:10:56 egaudet: :) Jan 30 19:12:44 ok so I can bind mount /usr/palm to /var/palm Jan 30 19:13:07 and then unionfs /var/usr_ufs/palm=rw:/var/palm=ro /usr/palm and it works Jan 30 19:14:12 egaudet: cvs -z3 -d:pserver:anonymous@aufs.cvs.sourceforge.net:/cvsroot/aufs co -P aufs Jan 30 19:18:34 egaudet: so you're thinking overlays for patching, easy off switch etc, but doesn't that mean you have to think to turn it off before upgrade for it to be useful? Jan 30 19:19:07 that's a really good point, you'll be blocking OTA changes basically Jan 30 19:19:31 I mean we could potentially have a service manage all this and detect updating Jan 30 19:19:58 use inotify like Puff will like :) Jan 30 19:20:25 inotify on what Jan 30 19:20:50 good question but I know the update process has specific things it does. we could also just patch the update app to call a service Jan 30 19:20:55 but we're getting into more complexity Jan 30 19:21:03 egaudet: what is causing people to have their webos-internals feed source changed to 'feed'? it seems to be happening a lot Jan 30 19:21:15 no clue Jan 30 19:21:28 hmph. Jan 30 19:21:30 but we need to have Preware/ipkgsrvice to a feed-health check on each boot Jan 30 19:22:01 egaudet: we also need a webos-internals feed fix utility in the precentral feed :p Jan 30 19:22:23 or have preware recognize the corruption and reset it Jan 30 19:22:29 that's a better idea Jan 30 19:22:42 er, have ipkgservice reset it :p Jan 30 19:23:03 feed-health check right Jan 30 19:23:11 umm Jan 30 19:23:50 i think we can figure out when an OTA is gonna happen Jan 30 19:24:46 has anyone used dbus-util to monitor service calls while the clicked on update? Jan 30 19:24:58 USB mode doesn't break unionfs Jan 30 19:25:42 the OTA obviously has to mount the update partition right Jan 30 19:25:48 so we get a udev event on that Jan 30 19:25:59 should be good? Jan 30 19:26:16 grrr Jan 30 19:26:17 empty_zero_page Jan 30 19:26:22 unknown symbol Jan 30 19:26:25 in aufs Jan 30 19:26:30 when i tried to load it Jan 30 19:26:51 also if there's anything in the future we need to do that USB mode breaks, we can detect usb mode via udev event like Palm already does Jan 30 19:27:24 i can fix that Jan 30 19:27:25 :D Jan 30 19:27:47 egaudet: wanna try out aufs? Jan 30 19:28:07 well what is it going to add that unionfs doesn't have :P Jan 30 19:28:15 well u tell me Jan 30 19:28:22 you have now used this more than me Jan 30 19:28:39 i have a kmod Jan 30 19:28:46 andthe tools Jan 30 19:28:52 well unionfs seems to work good as long as I bind mount /usr dir first somewhere else, and use that Jan 30 19:29:04 ok Jan 30 19:29:04 butt give me aufs so I have it Jan 30 19:29:12 because I don't have any kernel stuff set up Jan 30 19:30:40 egaudet: could it be that fuse unionfs can't overlay a filesystem that has files open? Jan 30 19:31:54 or maybe has checks, and the bindmount fools thems Jan 30 19:32:17 it's possible Jan 30 19:32:19 http://omploader.org/vM2ViMg/aufs1-2.6.24-palm-1.3.5.tar.bz2 Jan 30 19:32:32 but at the same time it's using 2 different ideas Jan 30 19:32:44 first way that's failing is trying to unionfs mount one dir onto the nonempty /usr Jan 30 19:32:52 and then /usr only has the cntents on that one dir Jan 30 19:33:09 egaudet: u can just insmod aufs.ko Jan 30 19:33:11 the way i'm doing it now is unionfs to merge 2 dirs and then mount on /usr Jan 30 19:33:22 so it could still have the same "issue" but the merge is merging everything first Jan 30 19:33:47 Rod mentioned that USB mode kind of breaks aufs Jan 30 19:35:20 patch USB mode startup Jan 30 19:35:37 usbmode wont break unionfs-fuse though :D Jan 30 19:36:36 yep :) Jan 30 19:37:27 so we can probably just throw away my on the fly launcher icon replacement stuff Jan 30 19:37:54 why? Jan 30 19:38:38 root@palm-webos-device:/# umount /usr Jan 30 19:38:38 umount: cannot umount /usr: Device or resource busy Jan 30 19:38:52 PuffTheMagic, well it's not really needed if we do the unionfs stuff right? Jan 30 19:39:02 no i like it Jan 30 19:39:08 I mean sure the icons change on the fly without luna restart, but the rest of the theme will still require a luna restart Jan 30 19:39:12 makes for light weight themes Jan 30 19:39:17 for peopel that just want to change icons Jan 30 19:39:34 OK, well the last one I have to find is the app-launcher icon Jan 30 19:39:42 it doesn't show up in listLaunchPoints Jan 30 19:41:51 by the time we're done with this we're going to have the same system we currently have in place... Jan 30 19:42:03 except with unionfs Jan 30 19:42:22 which means "EPR" and any other "back to stock" fix is "umount -l " Jan 30 19:42:24 yay Jan 30 19:44:42 and I still want the "load theme css with every other css" patch Jan 30 20:07:16 so we can't just unionfs the whole /usr either for security purposes Jan 30 20:10:18 egaudet: I'm sure you've not been sleeping, maybe missing work, and even ignoring your family and friends over your understandable worry that the VK may be causing memory/CPU issues on my pre Jan 30 20:10:51 are you here to put my mind at ease ? Jan 30 20:11:02 egaudet: you can rest easy. Jan 30 20:11:44 I believe that I have traced my issues to installing the VZ pmnetconfigmanager .ipk Jan 30 20:12:23 hey do ipk's in the app catalog have control scripts? (pre/post/etc?) Jan 30 20:12:23 I restored the files from that .ipk from the doctor rootfs and my Pre has behaved quite well since then Jan 30 20:13:52 dtzWill: I would imagine most don't need them? maybe pandora? Jan 30 20:14:12 zsoc: agreed, I'm just curious :). Jan 30 20:14:31 I'm quite sure they do not have scripts Jan 30 20:14:53 egaudet: thank you. it would be silly for them to, at least in the way that we do them in preware Jan 30 20:15:01 My theory is that the config files in the VZ version have iptables rules for forwarding, but I don't have the rest of the forwarding config in place, so it may be that some networking and or kernel buffer was getting abused by the half-complete forwarding config Jan 30 20:15:03 right Jan 30 20:16:08 PuffTheMagic, destinal, what/how should we limit unionfs if it were to be used for security purposes Jan 30 20:16:26 security purposes? Jan 30 20:16:28 i dont understand Jan 30 20:16:29 how so Jan 30 20:16:36 Obviously no binary containing directory should be unionfs'd Jan 30 20:16:52 what happened to /usr/local/lib ? we don't like -rpath? Jan 30 20:17:06 dtzWill, completely different dicussion Jan 30 20:17:12 this is about theme api Jan 30 20:17:16 egaudet: sorry i'll be quiet :) Jan 30 20:17:23 we are still going to use -rpath AFAIK Jan 30 20:17:27 egaudet: ah, understood. thanks Jan 30 20:18:23 So we can't just go and unionfs /usr to some /media/internal/ area and allow a user to overwrite some important binary with malware Jan 30 20:19:33 so we either need a ro layer that can only be controlled by root scripts or we need a pre-defined list of OK dirs to unionfs directly to /media/internal right? Jan 30 20:19:37 egaudet: i have an idea Jan 30 20:20:43 union an empty tmpfs directory over the target path then make symlinks onto it Jan 30 20:22:17 a sym link on a per file basis Jan 30 20:22:33 that's a lot of management and we need a giant list of pre-defined OK file paths Jan 30 20:22:49 no it could be folders Jan 30 20:23:31 symlink what ever you were gonna mount before Jan 30 20:23:36 OK so basically instaed of unionfs a bunch of separate pre-approved folders, unionfs one tmp dir that sym links pre-approved dirs Jan 30 20:24:20 no Jan 30 20:24:24 i mean like this Jan 30 20:27:58 mkdir =p /var/theme/usr/palm; mount -t tmpfs none /var/theme/usr/palm; unionfs /var/theme/usr/palm=rw /usr/palm Jan 30 20:27:59 then Jan 30 20:28:57 ln -s /meda/internal/theme/ the service would setup the unionfs at startup and then manage the symlinks when themes switch Jan 30 20:31:27 would that work? Jan 30 20:32:14 that's fine, the tmpfs is not needed could just be some random dir in /var anyway but my thing is we don't want all of /usr/palm to be overwritten easily like that Jan 30 20:32:14 PuffTheMagic: what's the latest kernel you were able to get to work on pre? Jan 30 20:32:34 2.6.24.7 Jan 30 20:32:41 well I suppose /usr/palm might be safe Jan 30 20:33:21 egaudet: well if u are worried Jan 30 20:33:23 but we also need /usr/lib/luna/luna-applauncher/icon.png Jan 30 20:33:28 then i think it HAS t be on a file basis Jan 30 20:33:33 and we could have a blacklist Jan 30 20:33:37 of files/extensions Jan 30 20:33:46 yeah I think we have to do it on a file basis, at least file type Jan 30 20:33:52 its not hard to loop thought a dir Jan 30 20:34:17 because with /usr/palm like the above, users now can just install applications over USB and that opens up to a world of maliciousness Jan 30 20:34:31 egaudet: is this gonna be done before rwhitby wakes up Jan 30 20:34:44 lol if you can freeze time Jan 30 20:34:49 could we get a whole project done while he is away on a 1 week trip :D Jan 30 20:34:59 in the last 2 days of a 1 week trip ;) Jan 30 20:35:33 hehe Jan 30 20:35:46 but this is far from close to being anything right now :( Jan 30 20:38:48 all I can do is change icons, woopty Jan 30 20:40:53 PuffTheMagic: btw ... gl-sdl is fixed (although it's not performing like we expected) Jan 30 20:41:10 software-sdl is significantly faster in many operations Jan 30 20:41:59 so rebuide evas? Jan 30 20:42:35 yes, update and rebuild Jan 30 20:42:48 i've also added ecore support Jan 30 20:42:58 i'll update elementary sometime later Jan 30 20:43:43 i suspect that gl will still work perfectly fine for most UI operations Jan 30 20:45:42 the previous bug was due to the initial resize call -- apparently libsdl on the pre doesn't know that it should just ignore a resize request instead of obliterating the GL context Jan 30 20:45:55 egaudet: q:D Jan 30 20:46:01 oops Jan 30 20:46:02 xcomp: :D Jan 30 20:46:16 xcomp: what are the slow operations? Jan 30 20:46:17 xcomp: lol Jan 30 20:46:44 PuffTheMagic: image blend Jan 30 20:46:46 destinal: we are gonna have gl terminus in a few min Jan 30 20:47:20 image blends were the operations that had a pretty bad frame rate (~6fps Jan 30 20:47:23 xcomp: thanks for do all this Jan 30 20:47:24 PuffTheMagic: sure, but without a performance boost, it's not as exciting as it should be Jan 30 20:47:41 destinal: he said some operations were slow Jan 30 20:47:47 destinal: terminus dont use image blend Jan 30 20:47:55 I don't know why yet, right now the hypothesis is that the gl libs aren't as complete as they should be Jan 30 20:48:34 xcomp: I also experienced unexpected slow downs when I used alpha blending in GL Jan 30 20:48:54 right, maybe the pre GL ports avoid that stuff for now Jan 30 20:49:02 interesting Jan 30 20:49:05 on raster's mystery future device the benchmarks are smoking -- but that device blows the pre away in performance in every category Jan 30 20:49:16 xcomp: not using much gl i wrote it off as not something to expect to be done quickly, but if you're claiming it's particularly slow that explains my experience--seemed to me that blending 2 textures should be something a gpu can do effectively (That's all i was doing, heh) Jan 30 20:49:24 xcomp: where can I buy a mystery device? :P Jan 30 20:49:32 xcomp: do you have a link? :) Jan 30 20:49:35 destinal: haha same thought xD Jan 30 20:50:05 dtzWill: ahh, so that seems to confirm it then. maybe someone can ask palm why alpha blending in gl is not being accelerated Jan 30 20:50:37 caused that guy they stole from AMD to do drivers hasnt gottent to it yet Jan 30 20:50:39 xcomp: I wouldn't be surprised if they were working on it, especially when the powervr guys start talking to them which just happened Jan 30 20:50:40 destinal: unreleased hardware :) -- he says he still has a large debug board sticking out of it lol Jan 30 20:51:11 (regarding running their demos on the pre) Jan 30 20:51:20 dtzWill: even more curious was the fact that expedite crashed on the test where it edits image alpha data Jan 30 20:51:52 xcomp: edits the alpha data? are you using gl1 or 2? Jan 30 20:51:58 GL2 Jan 30 20:52:26 one of the tests manipulates image ARGB data Jan 30 20:52:34 xcomp: sure, that should be just fine o_O Jan 30 20:52:34 it stalls for about a second then segvs Jan 30 20:53:02 xcomp: something to note is that gles doesn't support things like indicating the row length/the more copmlicated packing formats, something that caused me issues Jan 30 20:53:03 unfortunately gdb on the pre is useless so I couldn't backtrace Jan 30 20:53:25 xcomp: but iirc you said that this code should already by gles compat, in which case that wouldn't be the issue (just something i ran into) Jan 30 20:53:34 expedite gl is nice :D Jan 30 20:53:43 time for terminus Jan 30 20:53:46 xcomp: you can't get backtrace from gdb on the pre? :( Jan 30 20:53:55 xcomp: guess i've lucked out so far Jan 30 20:54:02 dtzWill: yeah, this is the same code working on raster's device (although that device uses x11) Jan 30 20:54:30 xcomp: bummer Jan 30 20:54:33 dtzWill: yeah I'm thinking I'm going to have to compile gdb specially for use Jan 30 20:54:52 raster's device is also powervr-sgx Jan 30 20:54:59 xcomp: do you know what's wrong with the existin gdb? Jan 30 20:55:06 s/existin/existing/ Jan 30 20:55:06 dtzWill meant: xcomp: do you know what's wrong with the existing gdb? Jan 30 20:55:33 I don't know ... when I backtrace I get nothing Jan 30 20:55:53 just the name of the function that crashed like memcpy() and an address Jan 30 20:55:55 I've been able to backtrace userland stuff Jan 30 20:55:58 nothing in the stack Jan 30 20:56:05 xcomp: could you have stack corruption? Jan 30 20:57:00 if it's memcpy, that seems even more likely Jan 30 21:03:10 possibly Jan 30 21:03:18 but i've seen it every time I've run gdb Jan 30 21:03:30 terminus runs like shit Jan 30 21:03:37 if u can even call it that Jan 30 21:05:15 and way slow Jan 30 21:06:13 xcomp: when i run it with terminus i get a screen corruptness on the top and right a little more Jan 30 21:10:46 destinal: so all of terminus's flakeness exists in the gl version Jan 30 21:11:01 so its gonna a take a group effort clean up terminus Jan 30 21:11:37 fuse and union-fs added to WIDK Jan 30 21:11:41 unionfs-fuse * Jan 30 21:12:48 woo. mass awesome porting of broken engines Jan 30 21:12:50 xD Jan 30 21:13:03 where is oil? he would appreciate this :p Jan 30 21:13:21 What are we gaining from terminus vs. terminal Jan 30 21:14:28 dont know yet cause its not done... seems faster Jan 30 21:14:35 could support custom fonts easily Jan 30 21:14:37 options? Jan 30 21:14:45 yeah prefs galore Jan 30 21:14:53 screen rotations soon hopefully Jan 30 21:14:54 idk Jan 30 21:15:27 i just think i would be easier to work with sdl to make a terminal than the what we are currently doing in the plugin Jan 30 21:15:41 plus sdl type can get into official catalog im sure Jan 30 21:17:06 terminus properly uses the current users login so if palm decides to have sdl games run as a non-root user terminus would log into that users account Jan 30 21:17:15 then someone could always sudo or su Jan 30 21:17:29 if they novaterm'd in first and edited /etc/group Jan 30 21:18:01 egaudet: a term in the app catalog would be nice wouldnt it ? Jan 30 21:18:17 PuffTheMagic: you didn't ask, but that would be _amazing_ :) Jan 30 21:18:27 PuffTheMagic: you think they'd let you do that? isn't "unsafe" or something silly? Jan 30 21:18:38 yea well I am assuming the PDK will allow plugins Jan 30 21:19:42 destinal: did you have plans on adding new features to termplugin or are u just trying to fix it and then work on terminus? Jan 30 21:20:44 PuffTheMagic: I plan on developing termplugin further but am also considering rewriting it with an "adapter" model. Jan 30 21:21:19 :( Jan 30 21:21:26 basically to stay out of Luna's process space as much as possible there would be a thin layer in the NPAPI plugin which would connect I/O to an outside process (like classic does) Jan 30 21:21:40 destinal: what if we could embed terminus in a plugin Jan 30 21:21:45 insteaad of a stand alone binary Jan 30 21:22:08 well a standalone binary is actually what I'm aiming for Jan 30 21:22:14 but with a plugin to hook it in Jan 30 21:22:27 destinal: why? Jan 30 21:22:39 ns plugin? Jan 30 21:22:47 so make terminus core a .so Jan 30 21:22:51 PuffTheMagic: because if you segfault while you're part of luna the whole UI comes crashing down Jan 30 21:22:52 and then a wrapper gui Jan 30 21:23:17 also it's easier to debug a seperate process Jan 30 21:23:51 easier to write them in one test harness and plug them into another too Jan 30 21:24:35 im not sure i follow how what you are planning would not bring down luna with a segfault Jan 30 21:24:38 I want to make it easy for anyone to port something to a plugin, there's still a lot of stuff SDL isn't the right answer for Jan 30 21:25:28 PuffTheMagic: imagine a thin plugin that calls an outside binary and maps in shared memory for video buffer and sends I/O events back and forth Jan 30 21:25:36 that's what classic does, so I know the concept is viable Jan 30 21:26:55 if the plugin segfaults, yeah, bye bye LunaSysMgr , but if the process does and the plugin is robust enough, it can close gracefully Jan 30 21:27:31 thats sorta orthoganal to a terminal project Jan 30 21:27:51 I basically want the plugin and terminal to seperate Jan 30 21:27:57 and work on both Jan 30 21:29:10 and I'd like to figure out some way to use standard graphics toolkits in the binary / external process Jan 30 21:29:46 and then make porting easy to get random apps into plugins and wrap them in mojo UI's Jan 30 21:29:59 s/plugins/THE plugin?/ Jan 30 21:30:00 destinal meant: and then make porting easy to get random apps into THE plugin? and wrap them in mojo UI's Jan 30 21:30:18 basically I'm tired of terminal being the only plugin Jan 30 21:30:27 or plugin-based app rather Jan 30 21:32:05 IMO palm should have developed this kind of thing long ago, and maybe they did for motionapps, but the fact that there's still nothing publically available for it sucks. Now maybe this will be in the official PDK Jan 30 21:32:08 but I have no idea really Jan 30 21:33:37 I expect though that it would be far easier to port something like dfbterm to work in this model than it would terminus Jan 30 21:37:51 but that's mainly because I don't think palm has done anything to support putting SDL / OGL stuff into the same card as a mojo app Jan 30 21:38:11 if they do get THAT, then terminus would be a good possibility Jan 30 21:41:08 not sure how to expose javascript callable functions in the plugin into the external binary in a standard way. maybe plugin plugins. Jan 30 21:41:11 :P Jan 30 21:44:42 i dont really care too much about having mojo mixed with my native app Jan 30 21:48:04 PuffTheMagic: then why bother with plugins? Jan 30 21:50:41 I mean, we already have SDL and type: game for the stuff we don't want any mojo in Jan 30 21:57:07 http://www.engadget.com/2010/01/30/visualboyadvance-turns-your-palm-pre-into-game-boy-emulator/ Jan 30 21:57:20 destinal: why can't plugins use gl? Jan 30 21:57:32 click to full screen or whatever Jan 30 21:58:09 tmzt: eh, you could with two different cards, but meh Jan 30 21:59:05 go to game mode when you click Jan 30 21:59:15 like I was discussing with raster for X apps Jan 30 21:59:46 well all you'd need is that external binary to be linked with SDL / opengl and it could still communicate with the plugin Jan 30 21:59:59 it just seems awkward Jan 30 22:00:58 I guess it would make game launchers easy Jan 30 22:07:04 I just want X apps to be cards Jan 30 22:23:24 whatever happened to the guy who ported the vnc client ? he said he was going to work on kdrive (X server) next Jan 30 22:29:29 added steps to get into linux recovery mode rather than just run doctor Jan 30 22:29:29 to http://www.webos-internals.org/wiki/Last_Resort_Emergency_BootLoader_Recovery Jan 30 22:32:43 (i was able to recover a mistake I made in an upstart script that stopped the system from booting by booting into the installer image, connecting with novaterm, and mounting the disk) Jan 30 22:37:14 neat. I wonder how difficult it would be to make a recovery ramdisk image Jan 30 22:38:39 jacques: what more would you be looking for besides what's on the installer image? some kind of automated recovery? Jan 30 22:39:31 destinal: maybe some tools not included in the install image. since I don't know what's in the install image I can't say Jan 30 22:39:45 I mean, when you novaterm in there's a pretty reasonable environment with busybox etc Jan 30 22:40:48 I guess a lot of what's on standard recovery images wouldn't apply here - tools to repair filesystems, etc Jan 30 22:41:16 well, it could actually. Jan 30 22:41:29 destinal: how do you get the install image to not do the install? Jan 30 22:41:38 it doesn't do anything but boot Jan 30 22:41:49 it's intended to be driven by the doctor sending it commands Jan 30 22:42:10 oh cool. showing my ignorance of that whole process Jan 30 22:42:23 jacques: yeah it's a really good design Jan 30 22:42:48 a human can manually do all kinds of things in place of the doctor just by novacom open tty:// or better, novaterm Jan 30 22:42:48 destinal: how about sshd ? I wuld like to boot to a ramdisk and then ssh in and be able to xfer files over scp etc Jan 30 22:43:05 hmm, I'd probably normally use novacom put and novacom get Jan 30 22:43:58 but I suppose you could add usbnet and wifi support and dropbear Jan 30 22:44:33 I like the novacom stuff, it's very handy Jan 30 22:45:14 I didn't even know novacom put and get existed Jan 30 22:45:34 jacques: palm really did a good job with their low level support toolset Jan 30 22:45:42 IMO Jan 30 22:47:02 just adds to the "can't brick a pre short of hardware damage" thing, you can always omap3 boot to bootie boot to linux over USB and novaterm and you've got a shell Jan 30 22:47:30 without having anything on the flash Jan 30 22:47:50 yeah that's a great feature Jan 30 22:48:19 I wonder if we could use bootie to start u-boot Jan 30 22:48:28 bootie has a chainboot commands Jan 30 22:48:54 the issue would be making it do it without someone sending the command over usb Jan 30 22:49:23 ah yeah Jan 30 22:49:27 bootie is actually a LOT like uboot Jan 30 22:50:16 u-boot allows dual-licensing? Jan 30 22:50:31 IIRC it does... Jan 30 22:51:01 jacques: ah, if it's BSD licensed than bootie may be based off of uboot Jan 30 22:51:57 I thought it was GPL Jan 30 22:52:38 http://en.wikipedia.org/wiki/Das_U-Boot Jan 30 22:53:16 ah. well then hopefully bootie's just inspired by it :) Jan 30 22:54:18 maybe we should remove this line from the wiki: "It seems to have a lot of u-boot code in it as well" **** BEGIN LOGGING AT Sat Jan 30 22:55:46 2010 Jan 30 23:08:36 test Jan 30 23:11:06 * oil looks for destanial-FAKE Jan 30 23:11:19 hi Jan 30 23:12:21 hi Jan 30 23:12:43 hey oil. did you guys ever consider a list mode for channels and querys rather than cards? Jan 30 23:13:02 yes, its still on the drawing board Jan 30 23:13:06 I will try implementing it myself if you guys will commit it Jan 30 23:13:14 test Jan 30 23:13:32 ok, I reversed the equivalent of -R Jan 30 23:13:40 welcome to ircd-seven, everyone Jan 30 23:13:41 http://git.webos-internals.org/trac/wIRC/ticket/19 Jan 30 23:14:21 i have a problem installing trackmywork with preware. when i tap on the install button another window opens. it looks like the app catalog in green and i can tap on a "download for free" button. then it seems to install but the message installing does not disappear. Jan 30 23:14:46 i get an icon in the launcher but this only links to the app catalog windows Jan 30 23:14:56 is this thing on? Jan 30 23:14:58 Unfortunately "destinal" should be destinal-work, but I'm not there now Jan 30 23:15:00 what can i do? Jan 30 23:15:02 and he won't listen to me Jan 30 23:15:05 :) Jan 30 23:15:53 ~~~~ Jan 30 23:15:54 ARGH!!! STOP IT jacques!!! Jan 30 23:16:07 cool @ oil Jan 30 23:16:21 at any rate, we're no longer requiring identify to speak, formerly known as +R but now +q $~a Jan 30 23:16:26 that was the basic idea anyways Jan 30 23:16:43 the javascript spam, hopefully, should be dead now Jan 30 23:18:25 when I get a chance I'm going to get weechat connecting to ircd-seven via SSL Jan 30 23:18:41 (and x-chat) Jan 30 23:19:02 apparently you can identify by SSL certificate now Jan 30 23:19:07 I haven't checked into how, yet Jan 30 23:19:10 nice Jan 30 23:24:40 jacques: alas, apparently I misread that. only the server is authenticated and identify is still the normal way. that would have been cool though :) Jan 30 23:25:16 maybe someday Jan 30 23:26:43 cool, apparently you can lock a channel that has to be secure to prevent anyone without SSL from joining Jan 30 23:31:03 jacques: ah, so SASL is the new secure authentication.. Jan 30 23:47:06 anything new Jan 30 23:47:08 ? Jan 30 23:48:47 PuffTheMagic: ircd-seven is new Jan 30 23:49:28 hopefully the death of spambots and not needing +R (or +q $~a (now)) Jan 30 23:50:01 (the death of the evil javascript cross protocol posts anyway) Jan 30 23:51:42 * ird made chilli Jan 30 23:51:47 chili* Jan 30 23:52:18 chiillii Jan 30 23:55:05 hi everyone Jan 30 23:56:00 *hm* i have to deactivate palm-beta in preware, then everything is fine O_o Jan 31 00:00:28 ..waitin on the flash update Jan 31 00:00:33 whats good jacques Jan 31 00:00:43 u devin those games? Jan 31 00:00:52 that VM stuff u get that to work? Jan 31 00:02:35 writing webos apps only includes html and javascript? Jan 31 00:03:00 ffs Jan 31 00:03:28 is peggle out for the pre i thought it was but i cant see it anywhere Jan 31 00:03:31 PuffTheMagic, got some more luna C service RE Jan 31 00:03:35 to do Jan 31 00:03:46 (process:18837): LunaService-WARNING **: LSMessageReply: Jan 31 00:03:46 You are attempting to send a reply to a reply message. Jan 31 00:03:46 I'm going to allow this for now to more easily reproduce some bugs Jan 31 00:03:46 we encountered with services using LSCustomWaitForMessage Jan 31 00:03:46 receiving a reply-to-a-reply, but soon this will return an error. Jan 31 00:04:02 LSCustomWaitForMessage Sounds really useful Jan 31 00:07:13 PuffTheMagic: go back to wIRC Jan 31 00:09:46 hmm.. so not only was there an engadget article about vba, but it was following by a ton of positive Palm comments. interesting. Jan 31 00:11:42 linkage Jan 31 00:12:46 morning Jan 31 00:12:54 it is. Jan 31 00:12:56 unfortunately josh didn't write it (arguable the only real journalist they have imo)... but it's there Jan 31 00:13:26 http://www.engadget.com/2010/01/30/visualboyadvance-turns-your-palm-pre-into-game-boy-emulator/ Jan 31 00:13:39 rwhitby did you have a chance to look at the book proposal? Jan 31 00:14:04 so what's the fix for 1.3.5 failed migration? EMH still? Jan 31 00:14:05 zsoc: yeah I also thought all the positive comments were interesting ... Jan 31 00:14:40 it has inaccuracy.. but at least it cites the precentral article, which i feel was written nicely Jan 31 00:15:03 destinal-home: EMH is _the_ fix :> no reason you should need anything else Jan 31 00:15:05 not sure about the part about pixi .... Jan 31 00:15:16 rwhitby did you have a chance to look at the book proposal? Jan 31 00:15:29 jacques: yeah that's... pretty bad. as Josh would put it "like a sixth grader rushing to write a report about a book he never read" ;) Jan 31 00:15:43 * zsoc makes certain comments in hopes that certain people will find them while googling Jan 31 00:17:49 Rick_work: yep, sent back the comments Jan 31 00:18:00 Rick_work: did it on the last flight leg ;-) Jan 31 00:18:37 rwhitby: catch up on logs during trip or doing it all at once now? :> Jan 31 00:19:20 rwhitby can you copy me? Jan 31 00:20:58 jacques: yes, any line in the wiki which says that bootie is derived from u-boot code should be deleted because it is completely not true. Jan 31 00:21:24 hmm this stupid app launcher icon is going to be annoying Jan 31 00:21:40 Rick_work: sent Jan 31 00:21:46 thanks Jan 31 00:23:48 egaudet: so the blocker for aufs or anything like that when I looked at it previously is how to hook it in so that it works for usb mode and OTA update. getting an overlay FS is the easy bit :-) getting it out of the way when needed is the hard bit. Jan 31 00:24:58 rwhitby which is why I like the idea of simply defining a standard naming for override css files and doing a one-time patch that calls them from the css code in framework. Jan 31 00:25:00 rwhitby, if we can't find anything to hook into the OTA we could inotify on /usr/lib/ipkg/info Jan 31 00:25:13 AFAIK that dir is ONLY touched on doctor and OTA Jan 31 00:26:13 Rick_work, yeah there are only a few things that we NEED to overlay Jan 31 00:26:25 app css's and the framwork css's. Jan 31 00:26:32 nope Jan 31 00:26:39 so I had previously not gone done the overlay path cause I couldn't get a reliable safe way of removing it temporarily, especially when webOS files or services were running from it (imagine a page-in required from the text section of a binary). Jan 31 00:26:44 the sysmgr icon(s) like wave image Jan 31 00:26:59 and I think I'm going to change from multiple .css files to one theme.css Jan 31 00:27:04 but if you can crack the reliable temporary unmounting at the right times, then everything else is straight-forward Jan 31 00:27:29 egaudet oh yeah, and the and the app icons. Jan 31 00:27:32 Rick_work: we're considering this for patches as well as themes Jan 31 00:27:49 eric "reliable" gaudet, that's what I call 'im. Jan 31 00:27:52 (and any other modification at all that homebrew needs to make to palm files) Jan 31 00:27:53 Rick_work, I can overwrite every icon on the fly Jan 31 00:27:58 egaudet there's a problem with the one theme css idea. -- it doesn't load AFTER the application css loads. Jan 31 00:28:06 except the app launcher icon Jan 31 00:28:10 so the application css still wins. Jan 31 00:28:20 Rick_work, I can make it load after the app css Jan 31 00:28:34 * destinal-home feels like an idiot -- looking for EMH (never had to deal with this first time around but helping someone with a 135 migration problem) Jan 31 00:28:34 just load theme.css after loading any css in the frameworks' loadStylesheet Jan 31 00:28:53 rwhitby, no need to unmount during OTA Jan 31 00:29:21 egaudet: ? Jan 31 00:29:40 egaudet: if you don't unmount during OTA, then later when you do unmount you leave the Pre in a broken sate Jan 31 00:29:41 dBsooner: you wont find it if the -internals feed config is corrupt Jan 31 00:29:42 state Jan 31 00:29:45 inotify the /usr/lib/ipkg/info/ Jan 31 00:29:49 rwhitby: Rick_work: where is this EMH thing? Jan 31 00:29:54 not in preware apparently Jan 31 00:29:57 and when we detect OTA after it is done we "fix" stuff Jan 31 00:30:02 destinal-home: preware, webos-internals feed Jan 31 00:30:08 zsoc: huh? Jan 31 00:30:09 hmmm, not showing up here Jan 31 00:30:25 our overlayed directory will now contain the GOOD OTA files,w e can umount when OTA is done and then fix the real fiels Jan 31 00:30:26 files* Jan 31 00:30:27 destinal-home: what's the URL in your manage feeds page? Jan 31 00:30:44 but I'm also not suggesting overlaying the entire /usr Jan 31 00:30:49 egaudet: that sounds fraught with danger Jan 31 00:30:50 we can only overlay non-binary areas IMO Jan 31 00:31:35 well, you can't unmount if a file in the area is in use Jan 31 00:31:36 bootie page line about u-boot removed Jan 31 00:31:37 rwhitby, OK so let's get a udev event on the update partition mounting Jan 31 00:32:04 destinal-home: is your feed broken too? Jan 31 00:32:06 rwhitby: feed URL is correct -- odd Jan 31 00:32:06 how does the OTA handle "file in use" Jan 31 00:32:09 IS EVERYONE'S FEED BROKEN? Jan 31 00:32:21 something strange Jan 31 00:32:26 jacques: our understanding is that bootie was written to have a similar command set, but the code is completely new and proprietary, not a derivation from anything else. Jan 31 00:32:34 EMH is showing up fine over here Jan 31 00:32:54 dBsooner: you as well verify the feed url in your 'manage feeds' page. Jan 31 00:33:10 my feed URL is fine Jan 31 00:33:23 destinal-home: you try list of everything 'emer' ? Jan 31 00:33:38 zsoc: tried that as well as all in feed Jan 31 00:33:47 destinal-home: ipkg -o /media/cryptofs/apps list | grep emergency Jan 31 00:34:23 yeah, it's there, just not showing up in preware which is bizarre Jan 31 00:34:31 not bizarre Jan 31 00:34:34 if you didn't reinstall ipkgservice Jan 31 00:34:36 you are using /var Jan 31 00:34:43 I did reinstall it Jan 31 00:34:48 zsoc: why you keep pingin me? Jan 31 00:34:48 lol Jan 31 00:34:51 egaudet: so I don't want to pour cold water on anything, I'm just saying that cracking the update and usb mode problem is necessary before spending any significant time on overlay stuff (which will be wasted time if it can't be cracked) Jan 31 00:34:52 then it becomes bizarre :P Jan 31 00:35:01 * dBsooner doesn't like loading preware.. because destinal won't fix it. Jan 31 00:35:06 dBsooner: i asked you a question? Jan 31 00:35:13 dBsooner: preware's fine Jan 31 00:35:16 -18:32:29- (+zsoc) dBsooner: you as well verify the feed url in your 'manage feeds' page. Jan 31 00:35:30 rwhitby, well I'm not wasting any time on overlay stuff, although I did put unionfs-fuse and fuse in the WIDK Jan 31 00:35:34 and they -work- Jan 31 00:35:46 egaudet: yeah, that's not a waste Jan 31 00:35:50 but I'm focusing on everything we can do with themeing outside of any type of overwrite/overlay Jan 31 00:35:52 rwhitby: you get my pm's? Jan 31 00:36:05 dBsooner: oh. wow. never mind. i'm sorry :< i'm off my rocker, new meds Jan 31 00:36:15 right now I have C service in a decent position to overwrite icons Jan 31 00:36:19 dBsooner: dunno, depends on whether they got lost in the big freenode changes Jan 31 00:36:19 zsoc: i think you are tab failing Jan 31 00:36:20 ;) Jan 31 00:36:30 egaudet: is it in a repo? Jan 31 00:36:31 rwhitby: i'll flood them to you again Jan 31 00:37:57 not yet Jan 31 00:38:11 I just got it to a good functional place Jan 31 00:38:48 destinal-home: you restart java services after you reinstalled ipkgservice? Jan 31 00:38:54 ie: ran the postinst :p Jan 31 00:39:22 zsoc: bootstrap script does that. if anyone is not using bootstrap script or a real installer, then they fail :-) Jan 31 00:41:07 rwhitby: i try to find peoples fail :( Jan 31 00:41:23 We need a preware for preware installation. Now _THAT'S_ a bootstrap. Jan 31 00:43:11 which we could write if we had a novacom installer Jan 31 00:45:05 Rick_work: tell me again, why don't we? i mean legitimately. Jan 31 00:45:30 because we only have linux drivers that aren't owned by palm? Jan 31 00:46:12 Ok, then we claim our program is SO amazing it has to be 201mb because of all the awesome in it Jan 31 00:46:31 and people will be liek "wow look at how big it is, must be full of awesome" Jan 31 00:46:47 OK I'm throwing away the symlink idea Jan 31 00:47:02 * zsoc tips his hat to raster Jan 31 00:47:56 burrp Jan 31 00:48:01 zsoc: moinin Jan 31 00:48:37 raster: gl_sdl runs, but has surprising performance issues with texture blending, any expertise? Jan 31 00:49:57 which blending? Jan 31 00:50:18 i hear it crashes with argb data - more worrying as that smells of a driver bug Jan 31 00:51:13 * rwhitby bbl Jan 31 00:53:43 http://www.rasterman.com/files/pre-vs-other.html Jan 31 00:54:08 er.. alpha blending in general i believe Jan 31 00:54:26 theres a comparison for you of the pre softwre sdl vs another cortex-a8 soc i have using x11 instead Jan 31 00:54:36 http://www.rasterman.com/files/other-vs-gl.html Jan 31 00:54:51 thats a comparion of this same soc - software vs gl Jan 31 00:54:53 i assume the Pre is the column on the left? :p Jan 31 00:55:04 (its sgx based - 800mhz cortex-a8, sgx 540) Jan 31 00:55:07 yes Jan 31 00:55:10 hmm.. Jan 31 00:55:11 as the top of the column says Jan 31 00:55:22 that gives u a relative idea of things Jan 31 00:55:26 it renders funny in lynx :P Jan 31 00:55:33 gl can be spotty in tersm of beating software - but this is not sa real-life test Jan 31 00:55:45 its meant to isolate operations so they can be tested and improved Jan 31 00:55:53 oooh Jan 31 00:56:04 u'd want a gui browser as its a nice color-coded table Jan 31 00:56:16 yes i figure that out now hehe Jan 31 00:56:22 well that is fascinating data. Jan 31 00:56:28 zsoc: what about wIRC Jan 31 00:56:29 and surprising to me Jan 31 00:56:39 PuffTheMagic: go do it.. morely. Jan 31 00:57:04 i want to get termiunus workin a little better Jan 31 00:57:23 may be fighting a losing battle :< Jan 31 00:57:38 zsoc: en on desktop u'll see its similar Jan 31 00:57:42 raster: in my inexperience, i would agree it could be a driver issue. Jan 31 00:57:43 why do you think that? Jan 31 00:58:46 raster: that beast 'machine' you have there tho is sick :> Jan 31 00:59:18 PuffTheMagic: evas seems to be less effective on webos than we originally thought? Jan 31 01:00:29 zsoc: no software is fast as hell Jan 31 01:00:52 PuffTheMagic: ah, i will digress Jan 31 01:01:10 zsoc: what are you basing that on? Jan 31 01:02:18 PuffTheMagic: my very limited experience with terminus and expedite, and the gl_sdl engine turning out not to be a cure all Jan 31 01:03:12 http://www.rasterman.com/files/x86-soft-vs-gl.html Jan 31 01:03:15 zsoc: who said cure all? terminus runs faster than termplugin/terminal, it just has bugs that make things disply not perfectly yet Jan 31 01:03:36 thats my laptop (2.8ghz core2 vs radeon hd4650) Jan 31 01:03:51 so eeven there software can beat gl with some things Jan 31 01:04:20 i know things i can improve in the gl engine, but only a bit here and there Jan 31 01:04:53 PuffTheMagic: enterminus was never "finished"a s a terminal emulator - its the same bugs as on a desktop Jan 31 01:05:00 fix those bugs therer and they'lll be fixed on yoour pre Jan 31 01:05:01 :) Jan 31 01:05:02 your Jan 31 01:05:11 raster: yes i realize Jan 31 01:05:23 i dont think zsoc realizes enterminus wasnt finnished Jan 31 01:05:49 zsoc: rememebr - software engine will do the same calculations with the cpu on every system - its idtenntical Jan 31 01:06:02 the slowness webos can create is in getting those calculated pixels to the screen Jan 31 01:06:09 the faster that path of course, the better Jan 31 01:06:45 as such in 32bpp evas's path is as fast as it gets on both fb and x11 - they should have only 1 copy (unless u run a compositor) Jan 31 01:07:26 raster: there is a elementary demo right? Jan 31 01:07:34 * zsoc has an epiphany Jan 31 01:07:34 PuffTheMagic: comes with elementary Jan 31 01:07:38 elementary_test Jan 31 01:07:54 rememebr - its finger based. uc an scroll that list with click+drag :) Jan 31 01:08:01 zsoc: we ned to run elementary_test, that will give us an idea how well it will work Jan 31 01:08:09 it? Jan 31 01:08:19 real life widgets and usage Jan 31 01:08:27 zsoc: efl stuff on the pre Jan 31 01:08:35 zsoc: you were just saying you dont hink its worth it Jan 31 01:08:38 it should work just fine Jan 31 01:08:44 at hvga - u have very few pixels to render Jan 31 01:08:46 raster: i think it does Jan 31 01:08:48 PuffTheMagic: no no, was misunderstood Jan 31 01:08:55 i'm workiing on my "beast" soc with 480x800 Jan 31 01:08:57 PuffTheMagic: expedite is not a good test? Jan 31 01:09:04 so 2.5x as much to pump out Jan 31 01:09:21 fewer cache hits than you get too Jan 31 01:09:25 480x800 needs a name, it's getting popular as a mobile device res Jan 31 01:09:30 (as caches are the same size) Jan 31 01:09:37 sure - got 800v600 mhz Jan 31 01:09:44 but... thats a LOT more pixel Jan 31 01:09:50 sand on that... its rather decent Jan 31 01:10:21 zsoc: most of the stuff in expedite we wont be using is a normal app Jan 31 01:10:24 raster: you have an extremely unique way of expressing things. Jan 31 01:10:28 zsoc: the stuff in elementary is though Jan 31 01:10:30 PuffTheMagic: i see now, i spoke too soon again :p Jan 31 01:10:35 zsoc: expedite is a good start to see if it sucks really badly and where (what operations - text? images? texture mapping/rotattion?) Jan 31 01:10:45 and then u can address the issues Jan 31 01:11:17 expedite isnt to benchmark how well your average apps will work Jan 31 01:11:35 its about isolating rendering operations and testing them as much as possible in isolation Jan 31 01:11:50 (also test some other non-engine specific paths that interact with engijne logic) Jan 31 01:12:01 in the end expedite is 80% useful for people writing engines Jan 31 01:12:13 and writing new objects for evas and new engine features etc. etc. Jan 31 01:12:35 ;its not amazignyl useful for average develoeprs to know how well their app will scroll Jan 31 01:13:16 expedite really "makes sense" once u know how the rendering works internally Jan 31 01:13:21 for example Jan 31 01:13:30 i know that the gl engine re-renders the whole buffer Jan 31 01:13:53 in gl u cant sanely to smooth vsynced and fast partial-swaps (ie swap only the region that was changed) Jan 31 01:14:16 u effectively have to re-render all of it then swap the whole bufffer (another copy from back to front buffer) Jan 31 01:14:39 (unless u have real fb swaps with fb address panning - then u still re-render whole buffer,but swap is "free") Jan 31 01:15:04 azakus: just tweeted you, heh Jan 31 01:15:22 so tests that really push evas's core that minimises updates to only render what changed, work really well on software bug worse on gl, unless gl is so incredibly fast it makes up for being "dumb" by just brute force Jan 31 01:15:38 makes sense Jan 31 01:15:51 so gl is pretty broken? ;) Jan 31 01:15:56 so take thhis as a starting point at any rate Jan 31 01:16:12 u'll also notice - the computationalyl simple things - software can beat gl in hands-down Jan 31 01:16:15 or equal it Jan 31 01:16:19 even on a desktop Jan 31 01:16:23 image crossfade for example Jan 31 01:16:40 very simple unscaled alpha blend with fixed alpha per pixel Jan 31 01:16:51 gl and software are neck and neck Jan 31 01:17:05 look at rotation or smooth scaling and gl really pulls head Jan 31 01:17:33 image blend solid unscaled - another example. its a simple blit from image to buffer - softwaree and gl are neck and neck on desktop Jan 31 01:17:49 anyway Jan 31 01:18:00 so always take results with a grain of salt Jan 31 01:18:12 gl was build for 3d, not 2d. d needs are different Jan 31 01:18:25 evas's design is much more geared around a gl-style rendering model, but with optimisations Jan 31 01:18:42 as such there are gl-es extensions for partial buffer swaps - nokia added them for the n900 Jan 31 01:18:55 but... they dont do their swaps "vsynced" Jan 31 01:19:03 so ull get tearing/artifacts Jan 31 01:19:51 hmm Jan 31 01:20:04 i wonder why there are no 'hybrid' engines floating about Jan 31 01:20:54 because its generally very painful to access data from both cpu and gpu Jan 31 01:21:03 as data lives on the gpu often with a very slow read interface Jan 31 01:21:12 i see Jan 31 01:21:12 on embedded thats not the case (currently) Jan 31 01:21:29 but that has been the case on desktops for years Jan 31 01:21:34 isn't what most of compiz is doing 2d? why the hardware rendering? Jan 31 01:21:51 its doing it all in gl - yes Jan 31 01:22:01 but its not rendering the finer details Jan 31 01:22:13 text, buttons, checkboxes, etc. etc. Jan 31 01:22:22 ah Jan 31 01:22:28 when u want your rendering pixel-perfect, thgins begin to get troublesome Jan 31 01:23:55 software also has one big plus - its STABLE Jan 31 01:24:09 it will work everywhere and anywhere without hw support or drivers Jan 31 01:58:51 how do you create a symbolic link in C? Jan 31 02:03:22 .. you mean without the ln utility? Jan 31 02:06:23 egaudet: man 2 symlink Jan 31 02:06:26 *gasp* Jan 31 02:06:28 that was hard Jan 31 02:06:40 :) Jan 31 02:07:07 LOL Jan 31 02:07:11 apple-free engadget Jan 31 02:12:01 For the helloplugin, do I have to cross-compile it? The instructions don't mention anything about it Jan 31 02:13:39 riskable: hi. you can compile it natively on an x86 box and copy it to your emulator (since it's x86 also) or crosscompile it ARM Jan 31 02:13:55 For reference, I tried using the webos-internals PDK scratchbox2 env but I can't seem to get libx11 configured properly in it Jan 31 02:14:17 riskable: you shouldn't need libx11. there may be some incorrect references to it in the plugin however Jan 31 02:14:36 why do people insist on cross compiling libx11 Jan 31 02:14:38 I don't think anyone has done anything with it for a very long time Jan 31 02:14:39 Ahh, I suspected as much Jan 31 02:14:53 it's like a super freudian action where people want X but wont say it lol Jan 31 02:15:25 ...because there's nothing in the default sb2 env for x11 (and you'd think that by now someone would've taken the time to port the necessary bits to compile NPAPI plugins by now) Jan 31 02:15:49 I'm the original author of it (basically I took one of mozilla's samples and ported it to palm's way of doing NPAPI) but termplugin may be a better example even though it's much much more complex Jan 31 02:16:58 destinal-home: term is in my quick launch, btw, don't know if i ever told you that xD Jan 31 02:17:19 zsoc: :) Jan 31 02:18:35 riskable: Since both plugins (and since basically nobody has done any others) I have totally rethought things -- I want to do a thin plugin which communicates with an external native binary Jan 31 02:19:03 anyone know what are some popular (and in depth) themes ? Jan 31 02:19:45 egaudet: left4dead is probably the most ridiculous Jan 31 02:19:55 riskable: classic actually did this with their plugin, and they refer to it as an "adapter" Jan 31 02:19:59 egaudet: changes most everything there is to change afaik Jan 31 02:21:07 Interesting Jan 31 02:21:57 riskable: I was talking about this idea earlier, but basically right now if your plugin crashes or leaks memory or threads, it's LunaSysMgr (the UI for the whole system) that's impacted Jan 31 02:22:21 since the plugin becomes part of the LunaSysMgr process Jan 31 02:22:32 Well, the plugin I wrote just executes a couple commands on the shell Jan 31 02:22:55 yeah, if your plugin is really tiny then you don't need the adapter model for sure Jan 31 02:22:57 Real simple stuff... Just things you can't do via Mojo SDK Jan 31 02:23:44 you don't need any of the window /paint stuff either Jan 31 02:24:26 just make your function and make it callable by javascript Jan 31 02:25:50 That's the plan Jan 31 02:25:58 Unfortunately I don't know C/C++ very well so this could take a while Jan 31 02:26:12 I need a plugin *more* basic than helloplugin :) Jan 31 02:26:21 hey, i need some help, i installed the virtual keyboard and now my hardware keybord stoped working Jan 31 02:26:42 anyone heard of the before Jan 31 02:26:53 R_E: the chances of those things being tied together are ridiculously unlikely. Jan 31 02:27:11 so i have a broken pre? Jan 31 02:27:12 (can chances be unlikely? than doesn't sound right...) Jan 31 02:27:33 riskable: so why use a plugin rather than a service? Jan 31 02:27:50 R_E: I am not the authority on the subject. Jan 31 02:28:26 Well, I already wrote a service to do what I want but it seems silly to keep it running there all the time just for the purpose of executing, "/etc/init.d/" restart Jan 31 02:28:42 does anyone know who i might ask for more info ? Jan 31 02:28:54 riskable: why not just use the existing upstart service? Jan 31 02:29:10 R_E: egaudet can verify the likely hood of it Jan 31 02:29:13 Also, Palm is likely to never allow apps with their own services but they might allow their own plugins (hence the forthcoming PDK) Jan 31 02:29:19 likely hood, now _there's_ an odd idiom. Jan 31 02:29:22 riskable: ah, I see what you mean Jan 31 02:29:34 thanks Jan 31 02:29:54 riskable: It's not at all certain that palm means NPAPI by "plugin". We're all assuming so, but it could be something else.. Jan 31 02:29:55 riskable: so upstart via NPAPI? interesting. Jan 31 02:30:14 plugin will likely mean sandbox :p Jan 31 02:30:27 riskable: since native games will also be part of the "Plugin development kit" Jan 31 02:30:44 which are SDL binaries rather than anything you'd normally think of as a plugin Jan 31 02:31:16 destinal-home: they get away with the misnomer because _anything_ in "web"OS is a 'plugin' xD Jan 31 02:31:35 sure, you plug a program into a web browser and everyone says "plugin" Jan 31 02:31:49 i never liked the terminology honestly Jan 31 02:32:02 I'm just trying to keep all my options open Jan 31 02:32:13 riskable: we'll all for abstract thinking :) Jan 31 02:35:14 I *really* want my app in the app store... This waiting and mystery from Palm is almost as frustrating as the waiting and mystery of my app! Jan 31 02:36:47 The sky's the limit with what we can do with native apps... I just hope Palm doesn't make the sandbox too restrictive Jan 31 02:37:03 I want to write apps that change the way things work on the whole platform to make the experience better Jan 31 02:37:35 ...but I can't do that if my apps can't touch stuff outside of /media/internal Jan 31 02:41:26 zsoc: yo Jan 31 02:41:33 xcomp: yo Jan 31 02:43:13 hm? Jan 31 02:43:36 riskable: yep, i like the way you think indeed Jan 31 02:44:12 zsoc: looking for someting to do? Jan 31 02:45:28 PuffTheMagic: I'm playing defend your honor. i'm on a new med and it's completely messing with my brain, so a flash game is probably all i can handle right now Jan 31 02:45:38 PuffTheMagic: alternatively, i may have been turned into a super genius, so i'll give something a shot Jan 31 02:46:06 you should turn that focus into getting all the efl stuff into preware :D Jan 31 02:47:36 zsoc: Start taking ADD meds? hehe Jan 31 02:48:20 I remember the first time I started taking Adderall... Whew! Felt very strange in my head Jan 31 02:48:31 destinal the crusher of dreams is here Jan 31 02:49:22 PuffTheMagic: lol Jan 31 02:49:46 PuffTheMagic: ... er.. "all the efl stuff"? Jan 31 02:50:19 zsoc: yeah all the efl libs need to be packaged and avail in preware so we can make sweet apps witht hem :D Jan 31 02:50:30 esp if u want elementary wIRC Jan 31 02:50:33 wIRC 2.0 Jan 31 02:51:49 PuffTheMagic: crap my head esploded Jan 31 02:52:26 lool Jan 31 02:52:27 PuffTheMagic: does terminus use evas? Jan 31 02:52:39 yeah it does Jan 31 02:52:53 ... so evas itself can be compiled against ... er... statically? Jan 31 02:54:56 oh man, every time i even think about doing something, i stop wanting to Jan 31 02:55:26 flash games are so good Jan 31 02:55:33 can't wait for flash on webos Jan 31 02:55:38 lock myself in the bathroom Jan 31 02:55:57 okok ... i guess i'll start with... eina? Jan 31 02:56:03 or ecore? Jan 31 02:57:56 zsoc: static linking couldbe done but its not easy Jan 31 02:59:51 zsoc: so i dont know where your true interest lay but if you want to help hack on terminus you are welcome to and once i start on moving wIRC to elementary if you want to help there you too you can **** ENDING LOGGING AT Sun Jan 31 02:59:57 2010