**** BEGIN LOGGING AT Fri Apr 03 02:59:56 2009 Apr 03 04:28:51 khem ? Apr 03 07:19:34 morning Apr 03 08:19:30 morning Apr 03 08:20:39 moin Apr 03 08:21:20 seen pwgen Apr 03 08:21:25 ~seen pwgen Apr 03 08:21:28 !seen pwgen Apr 03 08:21:28 pwgen was last seen on IRC in channel #oe, 3d 14h 39m 3s ago, saying: 'i have written a bug report regarding the SRC_URI bug id 5074'. Apr 03 08:21:31 pwgen (n=ew@90.185.120.221) was last seen quitting from #kexecboot 12 hours, 14 minutes ago stating (Remote closed the connection). Apr 03 08:21:41 wow.. Apr 03 08:21:59 do we have notes service anywhere? Apr 03 08:22:19 * Jay7 should post note to pwgen :) Apr 03 08:22:47 RP: ping? Apr 03 08:23:42 03Jeremy Lainé  07org.openembedded.dev * r8bb4025ed6 10openembedded.git/recipes/linux/ (linux-rp-2.6.26/defconfig-qemux86 linux-rp_2.6.26.bb): Apr 03 08:23:42 linux-rp-2.6.26: improve qemux86 defconfig, make preferred Apr 03 08:23:42 * add kernel modules for bridge support Apr 03 08:23:42 * make default preference 1 for qemux86 Apr 03 08:30:54 Jay7: I think nickserv has a notes-service, do /msg nickserv help Apr 03 08:31:36 jay7: no, I'm wrong Apr 03 08:38:49 is anyone here familiar with psplash? Apr 03 08:42:53 jeremy_laine: hi, some-of Apr 03 08:43:20 ant_work: ah, cool Apr 03 08:43:47 ant_work: my setup is as follows: psplash starts up on our framebuffer, so far so good Apr 03 08:44:15 ant_work: the last application that gets launched is a Qt/Embedded application on the framebuffer Apr 03 08:45:01 ant_work: my problems are: 1/ psplash pollutes my display by updating the progress bar for this last application Apr 03 08:45:31 ant_work: 2/ the progress bar doesn't even seem to reach exactly 100%, it stop just short of it Apr 03 08:46:24 ant_work: 3/ after a while, the console re-appears, overwriting my application's display Apr 03 08:47:26 hm..do you have some output on tty? Apr 03 08:47:42 where is your stderr? Apr 03 08:47:51 on fb? Apr 03 08:48:10 ant_work: ehm, at the moment I am testing on a qemux86 machine Apr 03 08:48:23 you could try to lower verbosity Apr 03 08:48:49 ant_work: the delay after which I get the console seems to be the 120 seconds which are at the end of the psplash init script Apr 03 08:49:03 btwI haven't seen any issue yet with psplash,...but I fought for bootlogo-fb-pollution :-) Apr 03 08:52:24 ant_work: ok, will keep looking Apr 03 08:52:59 ant_work: by the way, how do we contact upstream for psplash, it seems we have been carrying the logo-math patch forever Apr 03 08:53:36 btw: ira-utils, hostap-utils, ppp have QA 9 Apr 03 08:57:50 (klibc you already know about) Apr 03 09:02:48 ant_work: hostap-utils is pretty much obsolete :) Apr 03 09:03:25 jeremy_laine: I was checking the logs of last builds (few hours ago) Apr 03 09:04:43 jeremy_laine: well, there are a lot of bits from O-HAND Apr 03 09:05:06 same was for linux-rp Apr 03 09:05:28 ant_work: does RP have commit access for psplash? Apr 03 09:05:50 I suppose... Apr 03 09:06:39 I wouldn't mind getting in touch with whoever maintains psplash as I also wrote a patch for 1bpp and 2bpp support Apr 03 09:06:53 (useful on low-end LCD displays) Apr 03 09:07:39 jeremy_laine: good Apr 03 09:11:10 jeremy_laine: dhclient, cups too have QA= Apr 03 09:11:12 9 Apr 03 09:11:47 ant_work: who uses dhclient? it's an old version too .. Apr 03 09:12:12 I've built opie-image and x11-image Apr 03 09:21:07 gm all, hrw Apr 03 09:24:49 jeremy_laine: found another...obexpush Apr 03 09:25:39 and non dev contains .so, libgtkstylus, Apr 03 09:25:59 the rest seems just a bunch of 'invalid desktop-files' Apr 03 09:28:01 good morning Apr 03 09:28:59 morning Apr 03 09:31:41 non dev contains .so, libopieobex0 Apr 03 09:31:46 that's really all :-) Apr 03 09:32:16 jeremy_laine: wonderful cleaning! Apr 03 09:32:44 metadata is almost sane :) Apr 03 09:33:06 hrw: probably you branched one week too early.... Apr 03 09:39:55 ant_work: there always would be 'wait one week' Apr 03 09:40:10 he he Apr 03 09:57:50 Hi all Apr 03 09:58:46 I am tring to compile PACKAGE1 which depends on PACKAGE2... I upgraded PACKAGE2 , but when I try to recompile PACKAGE1 it still compiles with the older version of PACKAGE2 .. what should I clean or update or... ? Apr 03 10:37:55 florian: good morning Apr 03 10:38:21 hey pb_ Apr 03 10:47:31 hi Chris Apr 03 10:52:55 http://marcin.juszkiewicz.com.pl/2009/04/03/beagleboard-in-a-box/ Apr 03 11:37:40 hrw: I have the same slowest pc ever, it's my home server that runs this irssi+screen session Apr 03 11:38:11 is there a channel dedicated on making stable/2009 work ? Apr 03 11:38:42 keesj: ebox2300 based? Apr 03 11:38:45 BusError: no Apr 03 11:39:12 hey hrw ; having another shot at oe, since there seems to be a branch that is trying to work :> Apr 03 11:40:14 hw I sort of narrowed down a crashing problem (illegal instruction) related to libstdc++, on straight stable/2009 angstrom/glibc/armv4t -- rings a bell ? Apr 03 11:41:04 will run armv4t build Apr 03 11:41:09 om-gta01 will be ok? Apr 03 11:41:49 hmm yes should do. it's a s3c2410 ? Apr 03 11:42:01 yes Apr 03 11:42:08 no it' x86 based one http://www.linuxdevices.com/news/NS6828123924.html Apr 03 11:43:02 keesj: ebox2300 is name of mine which is vortex86sx based. Apr 03 11:43:07 keesj: your just share case Apr 03 11:44:02 ~curse gstreamer crap Apr 03 11:44:15 BusError: which recipe gives you that error? Apr 03 11:44:39 * hrw -> coffee Apr 03 11:44:59 well thats the problem, all builds, all installs, but if you try to run an app that uses the STL it crashes. I'm going to try to find you a real Apr 03 11:45:24 right now it did it with 2 apps I tried to compile. I'm trying to find a 'recipe' thst uses it.. Apr 03 12:38:08 BusError: so bug is on device not during build? Apr 03 12:42:27 hrw, yup. still trying to find a regular bb that uses the stl and doesn't require me to install a desktop env :> Apr 03 12:45:41 jeremy_laine: hi, have you tried to build a qt application out of OE tree ? Apr 03 12:47:15 BusError: iperf iirc Apr 03 12:48:01 ah great thanks I'm going to try that ome Apr 03 12:48:02 one Apr 03 12:50:19 mckoan: what do you mean? Apr 03 12:53:37 jeremy_laine: I'm trying to build a small sample qt app and I need to cross-compile it with the qt system I generated with OE Apr 03 12:54:20 if I launch qmake, I get this error : QMAKESPEC has not been set, so configuration cannot be deduced. Apr 03 12:54:28 Error processing project file: /home/koan/qt/appdemo/appdemo.pro Apr 03 12:54:56 * mckoan usually do not use Qt Apr 03 13:21:22 mckoan: why not use OE?? Apr 03 13:21:53 mckoan: if you absolutely cannot use OE, you will need to export every variable defined in qmake2.bbclass / qt4e.bbclass Apr 03 13:22:18 hrm Apr 03 13:22:20 jeremy_laine: what do you mean with "use OE"? Apr 03 13:22:21 openmoko dead? Apr 03 13:22:37 jeremy_laine: I'm using OE but not into its tree Apr 03 13:23:06 jeremy_laine: should I have to move my project into recipes ? Apr 03 13:24:23 mckoan: you'd indeed be better off writing proper bitbake recipes for your program Apr 03 13:25:22 jeremy_laine: thanks for the hint, BTW I arcgue nobody ever tried to build Qt apps out of OE tree Apr 03 13:26:31 jeremy_laine: do you have a sample .bb for build a Qt app like HelloWorld? Apr 03 13:27:53 mckoan: is it qmake based? Apr 03 13:28:43 let's say it is, are you using Qt/Embedded or Qt/X11? Apr 03 13:29:25 qt/x11 and I don't have any *make Apr 03 13:30:22 qmake should create Makefile for me Apr 03 13:37:48 jeremy_laine: maybe you could suggest me a recipe to use as example, Apr 03 13:39:37 mckoan: you will basically need "inherit qt4x11" in your recipe, not much else Apr 03 13:40:09 mckoan: qtnx-qte_0.9.bb is a starting point; though your recipe will be simpler Apr 03 13:40:18 arg sorry Apr 03 13:40:32 qtnx_0.9.bb for an X11 version Apr 03 13:42:03 good morning Apr 03 13:42:40 jeremy_laine: maybe also gemdropx_0.9.bb ? Apr 03 13:43:05 gemdropx is qt/e 2.x Apr 03 13:43:14 you do not want it Apr 03 13:43:28 uh ok thx Apr 03 13:47:54 the second problem is that OE works with downloaded packages and I have local sources, does exist a recipe working this way? Apr 03 13:56:06 has anyone of you ever experienced this with a newly created install " kernel BUG at fs/jffs2/file.c:251!" Apr 03 13:56:29 (I'm trying to add a new machine, cm-x300 to OE) Apr 03 13:57:15 after the error I get an segmentation fault, but the system boots, but after a reboot It won't come up anymore) Apr 03 13:57:30 1 st boot: http://rafb.net/p/VtAnMh91.html Apr 03 14:02:29 tsjsieb: You try to add a new machine and use an old kernel (2.6.26?). That way you'll have lots of problems sending your patches upstream :-/ Apr 03 14:03:36 tsjsieb: the first real problem is this: Node CRC 91aaa3dd != calculated CRC 84e0e8b2 for node at 0075173c Apr 03 14:03:42 I understand, but I have a working example of this kernel from the manifacturer for this specific machine, with a patch file Apr 03 14:03:50 tsjsieb: the rest seems to be follow-up problems Apr 03 14:04:17 tsjsieb: however, that can be anything: faulty hardware, wrong memory timing, error in the image that you "burned" into the device Apr 03 14:04:52 mckoan: if there are only a couple of files you could put them in "files" with your recipe, otherwise put a .tar.gz of the project in "files" Apr 03 14:05:58 tsjsieb: hehe, and maybe the "working example of a kernel" isn't really working Apr 03 14:06:08 schurig: thnx, I suspect it's something with my own created kernel / rootfs, because the example files from the manifacturer work correctly :) Apr 03 14:06:37 tsjsieb: you could compare your kernel to what is in the current kernel. If your problem is because of a programming error, chances are good that this has been fixed in the meantime. Apr 03 14:06:51 hello Apr 03 14:07:03 today i get my zaurus c1000 Apr 03 14:07:09 tsjsieb: okay, then I'd recommend to first use the provided-kernel, but your own filesystem. If that works, then the problem is in your self-made-kernel Apr 03 14:07:17 is it possible to boot from cf ? Apr 03 14:07:22 flameman: finally you have device with better kernel Apr 03 14:07:26 flameman: yes, it is Apr 03 14:07:33 tsjsieb: it could even be the case that provided kernel doesn't match the provided kernel source ... Apr 03 14:07:45 hrw, yes i am so happy :D you know how to boot from there ? Apr 03 14:07:48 flameman: you need ant_work or Jay7 - they are local zaurus hackers Apr 03 14:08:02 * Jay7 hides :) Apr 03 14:08:04 hrw oookkk :D very very happy Apr 03 14:08:19 flameman: you need kernel with pcmcia drivers in and proper root= in compiled-in cmdline Apr 03 14:08:38 kexecboot 0.5 should have pcmcia in kernel Apr 03 14:08:43 just try it Apr 03 14:09:10 jay7 so you mean ... using kexec in order to boot from cf ? Apr 03 14:09:37 oh, sh#t.. guy that I hope help me with ebay.com decline.. Apr 03 14:09:53 flameman: as universal way yes :) Apr 03 14:10:05 but you can use custom kernel too Apr 03 14:10:09 or u-boot Apr 03 14:10:18 all ways should work Apr 03 14:10:32 jay7 ... umm is there any image for the internal flash with a good boot loader ? Apr 03 14:10:45 kexecboot? :) Apr 03 14:10:51 yes, or u-boot Apr 03 14:11:07 http://www.angstrom-distribution.org/demo/kexecboot/ Apr 03 14:11:11 try kexecboot here Apr 03 14:11:14 ok Apr 03 14:11:26 I don't know any ready kernel with u-boot now.. Apr 03 14:11:31 may be ant know Apr 03 14:11:46 schurig: with the kernel provided, and my own rootfs, I only have the problem that the modules ar not found, wat's reasonable because the modules dir is named differently Apr 03 14:12:41 C1000 ---> akita ? am i right ? Apr 03 14:12:49 yes Apr 03 14:12:58 ok, let's go rock n roll :D Apr 03 14:12:59 remember this name :) Apr 03 14:13:08 do NAND backup before :) Apr 03 14:13:10 tsjsieb: then copy the modules manually into your image, with a hook :-) ROOTFS_POSTPROCESS_COMMAND = "name_of_shell_function" Apr 03 14:13:13 just to have :) Apr 03 14:14:31 schurig: thnx I will try that :) Apr 03 14:15:26 tsjsieb: also check power supply. I once had a device where reading was working, but writing did occassionally not work. In the end it was the power line towards the Flash on the PCB, which didn't give enougth current while eraseing. Killed hardware engeneer for that, got 24 years jail :-) Apr 03 14:17:36 schurig: lol :D I hope your not serious. But I suppose that's not the problem in my case, because the example files from manufacturer keep working every time Apr 03 14:18:52 tsjsieb: yeah, but but different kernel draw different currents at different times, e.g. because of different thread / workqueues / irq-routines running Apr 03 14:19:19 tsjsieb: also, it might be the case that the GPIOs aren't initialized the same as in the original sample, so maybe your board draws more current because of that Apr 03 14:20:14 *IF* the current to the chip is your problem (not likely, agreed), than it's not the sustained current, but the spike current (not sure how to say that in english) Apr 03 14:20:28 o.k thnx I will check :) Apr 03 14:39:02 jay7 is there a procedure to install the kexec into the internal flash ? Apr 03 14:40:36 http://www.angstrom-distribution.org/c1000-install-instructions Apr 03 14:40:45 just w/o initrd.bin Apr 03 15:32:52 hmm whats the bitbake target to rebuild the Package* files again ? Apr 03 15:33:32 package-index Apr 03 15:35:00 thx. hrw I have a new .bb with a kaemon that triggers commands on keypresses (buttons on boards).. is the right way to work is to give you a git tree/branch with the patch hangin on it ? Apr 03 15:35:03 khem, around? Do you recall just when/what that gdb/libthread fix was? I need to pull that in :) Apr 03 15:35:07 doing a git bisect atm Apr 03 15:35:47 BusError: use "git format-patch" maybe? Apr 03 15:36:29 hrw, sure, but I tought maybe a git pull from your end would be faster ? shall I email it to you ? Apr 03 15:37:42 BusError: ok, give me url then Apr 03 15:43:39 hrw, git://repo.or.cz/openembedded/mini2440.git head for-hrw Apr 03 15:43:56 it's based on the current stable/2009 Apr 03 15:44:24 patch for .dev will be Apr 03 15:45:01 mini2440? you work on adding that board? Apr 03 15:45:16 it already works. been for some time too :> Apr 03 15:46:13 I held uo OE some, because the -dev branch was to borken at the time. with a stable branch I'm having a go again Apr 03 15:48:02 I will build it for .dev and then push. Apr 03 15:48:11 you can then request it for stable Apr 03 15:48:21 OK thanks. Apr 03 15:48:42 it's just a clean, lean, simple tool really Apr 03 15:49:01 but rules are rules Apr 03 15:49:30 absolutely agree, I meant that it's a tool very nice on an embedded board :> Apr 03 15:49:54 any rule that prevents the chaos to happend is good in my book :> Apr 03 15:50:45 ~curse parsing time Apr 03 15:52:17 I'm still trying to get my c++ app to work. iperf works. my simple c++ works, my real app fails. even a print() after first line in main() fails. I;m trying to get gdbserver to work to see a bit more details Apr 03 15:58:36 03Michel Pollet  07org.openembedded.dev * rb3c9bf48ee 10openembedded.git/recipes/esekeyd/esekeyd_1.2.4.bb: (log message trimmed) Apr 03 15:58:36 esekeyd: ESE keyboard daemon recipe Apr 03 15:58:36 ESE Key Daemon is a multimedia keyboard driver for Linux. Apr 03 15:58:36 With the 2.6 kernel series it can also handle remote controls, Apr 03 15:58:36 as they are presented as keyboards. No kernel patch is required. Apr 03 15:58:37 It is a userspace program that pools /dev/input/event? Apr 03 15:58:41 interfaces for incoming keyboard key presses. Apr 03 15:58:43 03Marcin Juszkiewicz  07org.openembedded.dev * r0be1026692 10openembedded.git/recipes/openssl/ (4 files): openssl-native: removed FILESPATH setting which is not needed Apr 03 16:04:24 hrw, there was no MD5 in my patch... Apr 03 16:07:10 argh Apr 03 16:07:15 will add Apr 03 16:07:33 sorry I tried to make it "branch friendly" Apr 03 16:07:59 I knew you'd probably move it, and the md5 is prone to break Apr 03 16:09:17 03Marcin Juszkiewicz  07org.openembedded.dev * rf50da97984 10openembedded.git/conf/checksums.ini: checksums.ini: added entry for esekeyd, removed duplicate of gcc 4.3.2 Apr 03 16:09:34 BusError: next time commit both in one run - I used format-patch Apr 03 16:09:44 * hrw -> off for some time Apr 03 16:14:20 anyone uses gdbserver ? I installed it, and it quits immediately, with no error. thats on stable/2009 armv4t Apr 03 16:18:13 * Tartarus goes around kicking opkg again Apr 03 16:18:52 meta-toolchain-gpe installs task-sdk-base, which RDEPENDS (And DEPENDS) on task-sdk-bare Apr 03 16:19:05 It is not, however, trying to install task-sdk-bare Apr 03 16:19:30 Anyone know why? Apr 03 16:19:56 Also, is bb 1.8.12 still ok for .dev? -c rebuild isn't doing anything anymore.. Apr 03 16:21:48 * florian remembers someone did some strange changes to the toolchains and broke some of them Apr 03 16:23:05 Tartarus: rebuild in particular is broken but otherwise it should be fine AIUI Apr 03 16:26:47 broonie: isn't rebuild defined by base class? Apr 03 16:27:33 Tartarus: any more ideas about the curl-sdk issue? Apr 03 16:29:34 rebuild is defined in base.bbclass, and rebuild doesn't behave correctly in general nowadays Apr 03 16:30:06 Man I hate opkg Apr 03 16:30:18 Problem was that since libiw-dev didn't exist it didn't try and install task-sdk-bare Apr 03 16:30:35 Now to figure out why wireless-tools isn't packaging as expected Apr 03 16:39:04 OK, funky... Apr 03 16:39:14 * Tartarus tries wireless-tools for mx31ads not qemux86 Apr 03 16:41:23 on qemux86 wireless-tools is missing many, many files Apr 03 16:41:31 it's just libiw and ifrename Apr 03 16:52:10 yeah, this is all just wrong, wth.. Apr 03 16:52:24 * Tartarus will post this for review tho, it doesn't make sense that this is broken this badly.. Apr 03 17:17:56 Oh, feh. SDL 1.2.11 appears to have changed checksum :/ Apr 03 18:29:13 hey ... uclibc doesn't seem to want to compile for armv4t in Thumb mode Apr 03 18:29:26 apparently this was "fixed" but doesn't actually seem to be fixed :( **** ENDING LOGGING AT Fri Apr 03 18:33:43 2009 **** BEGIN LOGGING AT Fri Apr 03 18:36:14 2009 Apr 03 18:43:51 jo Apr 03 18:53:08 dhd: what error do you get Apr 03 18:54:05 morgen woglinde :) Apr 03 19:26:22 khem: it's an ICE, unable to find a register to spill, in _dl_dprintf() Apr 03 19:26:41 will it work to compile ldso as arm and the rest of uclibc as thumb? I'm trying that now Apr 03 19:27:01 this is gcc 4.2.4, oe stable/2009 branch Apr 03 19:46:47 http://community.buglabs.net/bballantine/posts/102-Three-Steps-for-OpenEmbedded-N-Bs - worth read for everyone ;D Apr 03 20:01:46 hmm... e2fsprogs is failing to build on uclibc Apr 03 20:02:26 dhd hm? Apr 03 20:02:32 what is the error? Apr 03 20:02:36 yeah - undefined references to __aeabi_read_tp Apr 03 20:02:41 in gen_uuid.c Apr 03 20:03:20 but I'm not sure it is actually libuuid specific, it's just that tst_uuid seems to be the first executable it builds Apr 03 20:03:54 (also I am building for armv4t in Thumb mode) Apr 03 20:04:11 hm oh Apr 03 20:04:12 is TLS not supported for Thumb? that would make sense I suppose Apr 03 20:04:32 is there a good way to force some packages to build as ARM and others as Thumb? Apr 03 20:04:56 there is no no TLS support yet in uclibc Apr 03 20:05:04 oh. Apr 03 20:06:29 nptl merge is still pending Apr 03 20:06:36 ah Apr 03 20:06:50 and ther will be no support for arm oabi Apr 03 20:06:54 curious though as to why this build would be failing, and if it's thumb related Apr 03 20:07:04 yeah Apr 03 20:07:14 I am just testing right now armv4 here Apr 03 20:07:15 my understanding is __aeabi_read_tp is a built-in that reads the TLS register Apr 03 20:07:31 and it's likely actually being called from some inline function in uclibc Apr 03 20:08:44 args Apr 03 20:09:03 why the heck its trying to build the old version of e2fsbprgs Apr 03 20:09:17 the one causing me problems here is 1.41.4-r1 Apr 03 20:09:23 in the stable/2009 branch Apr 03 20:09:29 building console-image for angstrom Apr 03 20:09:32 are you needing e2fsprogs or e2fsprogs-lib? Apr 03 20:09:38 no actually I have no need for them at all Apr 03 20:09:48 along with like 90% of the things that console-image is building actually :) Apr 03 20:11:43 oh I can just remove ext2 from MACHINE_FEATURES Apr 03 20:11:50 problem solved, I guess Apr 03 20:12:06 okay same error here Apr 03 20:16:32 hm I will check buildroot if they have a fix for this Apr 03 20:18:16 ah Apr 03 20:18:17 03Otavio Salvador  07org.openembedded.dev * rcdd5eb0d31 10openembedded.git/recipes/mozilla/ (firefox-l10n-pt-br_3.0.4.bb firefox-l10n.inc): Apr 03 20:18:17 firefox-l10n-pt-br: add Apr 03 20:18:17 Signed-off-by: Otavio Salvador Apr 03 20:18:19 03Otavio Salvador  07org.openembedded.dev * r35467d35ba 10openembedded.git/recipes/mozilla/firefox.inc: Apr 03 20:18:20 firefox.inc: disable application updating Apr 03 20:18:21 -disable-tls Apr 03 20:18:22 Signed-off-by: Otavio Salvador Apr 03 20:18:24 03Angus Ainslie  07org.openembedded.dev * rc2f22a74d9 10openembedded.git/conf/checksums.ini: Apr 03 20:18:27 Add elementtree checksum Apr 03 20:18:29 Signed-off-by: Otavio Salvador Apr 03 20:18:31 03Ihar Hrachyshka  07org.openembedded.dev * r89eeba8660 10openembedded.git/ (conf/checksums.ini recipes/ruby/ruby-dbus_0.2.1.bb): Apr 03 20:18:34 ruby-dbus: new recipe Apr 03 20:18:36 This is a package with D-Bus bindings for Ruby language. Apr 03 20:18:36 I will make a patch Apr 03 20:18:38 Signed-off-by: Ihar Hrachyshka Apr 03 20:18:40 Signed-off-by: Otavio Salvador Apr 03 20:18:42 03Otavio Salvador  07org.openembedded.dev * r0b9cb309ce 10openembedded.git/recipes/mozilla/firefox.inc: Apr 03 20:18:42 thanks! Apr 03 20:18:45 firefox.inc: properly support locale usage based in system locale Apr 03 20:18:55 Signed-off-by: Otavio Salvador Apr 03 20:18:57 03Otavio Salvador  07org.openembedded.dev * r7c1739c542 10openembedded.git/recipes/alsa/alsa-state.bb: Apr 03 20:19:00 alsa-state: register the script for the shutdown, storing the settings Apr 03 20:19:02 Signed-off-by: Otavio Salvador Apr 03 20:19:04 03Henry von Tresckow  07org.openembedded.dev * rc0bada0355 10openembedded.git/recipes/sg3-utils/ (files/Makefile.patch sg3-utils_1.24.bb): Apr 03 20:19:09 sg3-utils: Updated Makefile patch to deal with cross-compiling Apr 03 20:19:11 Signed-off-by: Henry von Tresckow Apr 03 20:19:13 Signed-off-by: Otavio Salvador Apr 03 20:19:15 Acked-by: Koen Kooi Apr 03 20:19:17 03Otavio Salvador  07org.openembedded.dev * r18019ca851 10openembedded.git/recipes/gnome/gnome-keyring_2.24.1.bb: Apr 03 20:19:22 gnome-keyring: depends on gconf to fix a FTBFS Apr 03 20:19:26 Signed-off-by: Otavio Salvador Apr 03 20:19:28 03Otavio Salvador  07org.openembedded.dev * r4159b51b90 10openembedded.git/conf/machine/ (geodegx.conf include/tune-geodegx.inc): Apr 03 20:19:31 machine/geodegx: optimize using gcc 4.3 -mgeode and a specific feed arch Apr 03 20:19:33 Signed-off-by: Otavio Salvador Apr 03 20:19:35 03Otavio Salvador  07org.openembedded.dev * rbd90cde539 10openembedded.git/conf/machine/ (geodelx.conf include/tune-geodelx.inc): Apr 03 20:19:42 machine/geodelx: optimize using gcc 4.3 -mgeode and a specific feed arch Apr 03 20:19:44 Signed-off-by: Otavio Salvador Apr 03 20:19:46 03Jader H. Silva  07org.openembedded.dev * r27b5b2934e 10openembedded.git/conf/distro/include/sane-srcrevs.inc: Apr 03 20:19:49 Add calibrateproto and libxcalibrate sane srcrevs Apr 03 20:19:55 Signed-off-by: Otavio Salvador Apr 03 20:19:59 (3 lines omitted) Apr 03 20:21:44 hmm well it does seem that building ldso as arm and the rest of uclibc as thumb works okay Apr 03 20:21:56 hm? Apr 03 20:22:06 I'm getting an ICE building uclibc as thumb Apr 03 20:22:13 I thought khem checked in his patches for ldso Apr 03 20:22:16 even with the patch to make it build without optimization Apr 03 20:22:23 yes, he did, for some reason they didn't fix it for me... Apr 03 20:22:33 hm hm Apr 03 20:22:36 please write him Apr 03 20:22:37 or rather they didn't fix all of the errors I guess Apr 03 20:22:55 sure - I will get a log from the failed build Apr 03 20:23:28 I also have a weird issue where even though /sbin/init points to /sbin/init.sysvinit, it still tries to use busybox init at boot Apr 03 20:23:48 which doesn't work, because busybox init doesn't understand the same inittab format Apr 03 20:24:55 is this because /linuxrc is pointing to busybox? Apr 03 20:24:59 (this is in minimal-image) Apr 03 20:36:30 he ant Apr 03 20:36:48 ant do you have armv6 toolchain at hand? Apr 03 20:36:58 hi Apr 03 20:37:06 hm.. armv5te Apr 03 20:37:14 would be okay too Apr 03 20:37:23 then I have Apr 03 20:38:40 ant can you test the following patch for ffmpeg? Apr 03 20:38:42 http://paste.debian.net/32331/ Apr 03 20:39:32 woglinde: should I try without patch at first? Apr 03 20:39:54 ???? Apr 03 20:39:55 nope Apr 03 20:39:58 ok Apr 03 20:40:19 hm even better you make bitbake -c configure ffmpeg Apr 03 20:40:26 ans then go to this dir Apr 03 20:40:40 compile with toolchain-gcc at hand to the assembler stage Apr 03 20:40:43 and look Apr 03 20:40:52 if there is the bx or the mov line Apr 03 20:42:24 hi gnutoo Apr 03 20:43:47 woglinde, hi there are things that I don't understand in the lwn post you have given to me...I understand the assembly but not all what is done with it...such as mov r3, #0xffff0fff Apr 03 20:44:15 gnutoo thats kernel magic for atomic cmp_swap on arm Apr 03 20:44:20 should I send a mail to the mailing-list for that? Apr 03 20:44:21 ok Apr 03 20:44:34 hm mom Apr 03 20:44:41 ah yes Apr 03 20:44:57 so I should keep the same numbers or is the number dependant from the number of instructions? Apr 03 20:45:49 gnu too this you mean? Apr 03 20:45:51 http://lwn.net/Articles/314235/ Apr 03 20:46:05 yes Apr 03 20:47:14 hm its purly for armv5 Apr 03 20:47:18 bx is used Apr 03 20:47:28 ah ok so it doesn't work for armv4... Apr 03 20:47:44 but this will be working with gcc-4.4 Apr 03 20:47:53 ok Apr 03 20:48:01 there is the linker bx patch included Apr 03 20:48:13 which patches bx to mov within binutils Apr 03 20:48:38 wow Apr 03 20:49:33 hm but you can use swp for pre armv6 Apr 03 20:49:40 yes I can Apr 03 20:49:42 makes no difference Apr 03 20:49:47 besides speed maybee Apr 03 20:49:51 dont know Apr 03 20:49:56 swp is atomic Apr 03 20:50:01 yes Apr 03 20:50:01 woglinde: still compiling deps... Apr 03 20:50:11 ant args yes Apr 03 20:50:45 by the way how much should I change the code beside changing the add ptr to add +1 and adapting it to armv4? Apr 03 20:50:46 some "dont want to call him idiot" builds schroedinger-gst plugins with schroedinger recipe Apr 03 20:51:06 so it pulls in whole gst and gnome-vfs and stuff Apr 03 20:51:13 I have a patch here to fix this Apr 03 20:51:28 gnutoo reminds me by the way Apr 03 20:51:58 that we use the not arm-optimizied libatomic-ops with pulseaudio Apr 03 20:51:59 woglinde: look on top build http://tinderbox.openembedded.net/builders/ant/ Apr 03 20:52:07 which uses always swp Apr 03 20:52:15 I wanted to fix this too Apr 03 20:52:41 ok Apr 03 20:52:47 ant hm Apr 03 20:53:06 ant hm this dont help much Apr 03 20:53:10 almost done (configure) Apr 03 20:53:14 ant can you do as I said Apr 03 20:53:30 I'm still at -c configure... Apr 03 20:53:35 ah sorry Apr 03 20:53:37 hehe Apr 03 20:53:47 didnt look this much at tinderbox yet Apr 03 20:53:50 -3 Apr 03 20:53:54 -2 Apr 03 20:53:55 so I thought it was finished Apr 03 20:54:05 -1 Apr 03 20:54:19 ok, done Apr 03 20:54:44 no schroedinger plugins here Apr 03 20:54:49 just schroedinger Apr 03 20:54:58 btw Apr 03 20:55:17 hehe yxes Apr 03 20:55:25 because you only did configure Apr 03 20:55:32 :/ Apr 03 20:55:37 dhd: is ICE only problem you see while compiling ldso ? Apr 03 20:55:43 look at the depends Apr 03 20:55:53 khem: yeah Apr 03 20:55:56 dhd: I would suggest to use gcc 4.3.x Apr 03 20:56:00 mom, now I should compile it by hand? Apr 03 20:56:05 if I add -marm to CFLAGS-ldso it works... Apr 03 20:56:06 hm they vanaished Apr 03 20:56:11 khem: ok will try that tomorrow, thanks! Apr 03 20:56:22 dhd: yes that will work. We have no problems in arm mode Apr 03 20:56:27 its thumb mode thats the challance Apr 03 20:56:51 woglinde: you mean -c compile now? Apr 03 20:56:58 ant__ hm its already assembler Apr 03 20:56:59 and what was the second problem you mentioned Apr 03 20:57:09 let me think Apr 03 20:57:11 ah the second problem is that the images are linking /libc -> /bin/busybox Apr 03 20:57:16 er not /libc, /linuxrc Apr 03 20:57:24 ldso is a beast its one huge file with huge functions and they all need to be in one Apr 03 20:57:32 ah Apr 03 20:57:38 and this makes the kernel use busybox init instead of sysvinit, which does not understand inittab Apr 03 20:57:41 -E schould be enough Apr 03 20:57:43 okay Apr 03 20:57:58 ant cd ffmpeg-0.5/libavcodec/arm/ Apr 03 20:58:00 khem: yeah sounds like a stress test for the compiler Apr 03 20:58:07 Thumb has fewer registers doesn't it? Apr 03 20:58:20 and type gcc -E dsputil_arm_s.S -o moo.S Apr 03 20:58:22 yep plan to make is simpler Apr 03 20:58:30 and look the last lindes Apr 03 20:58:33 aeh lines Apr 03 20:58:40 yes thumb is half registers Apr 03 20:58:43 args arm-blafasle-foo-gcc Apr 03 20:58:43 than arm Apr 03 20:58:53 okay I will try 4.3.x tomorrow... thanks Apr 03 20:59:17 go gcc-4.4 Apr 03 20:59:20 dhd: oh btw you need to use either binutils-2.18 or binutils-cvs Apr 03 20:59:22 go go go Apr 03 20:59:24 *g* Apr 03 20:59:31 binutils 2.19x wont work Apr 03 20:59:38 khem why? Apr 03 20:59:55 is it patched out in binutils-cvs? Apr 03 21:00:05 because it has bugs in trampoline code for interworking Apr 03 21:00:14 yes Apr 03 21:00:21 hm cant you provide a patch for 2.19? Apr 03 21:00:31 or is it to much? Apr 03 21:00:44 I could but I am lazy and it will be all work Christophe did lately Apr 03 21:00:55 pfffff Apr 03 21:00:56 lazy Apr 03 21:01:22 woglinde: gcc -E dsputil_arm_s.S -o moo.S Apr 03 21:01:23 dsputil_arm_s.S:22:20: error: config.h: No such file or directory Apr 03 21:01:26 all this was borked bigtime and now its gettig better Apr 03 21:01:34 ant aeh Apr 03 21:01:40 hopefully 2.20 is a good one Apr 03 21:01:48 ant search for the config.h Apr 03 21:02:04 and add -I../dir_to_config.h Apr 03 21:02:52 whee Apr 03 21:03:07 Time to test relocated SDKs of 4.1.2/4.2.4/4.3.3 with qtopia being compiled Apr 03 21:04:04 qtopia is dead Apr 03 21:04:06 dead dead Apr 03 21:04:13 that's fine Apr 03 21:04:24 this is just something that i've had break horribly my SDKs before :) Apr 03 21:04:31 if you need __aeabi_read_tp then you better use uclibc-nptl.bb Apr 03 21:04:46 khem hm Apr 03 21:04:50 so you ported tls Apr 03 21:04:55 but not nptl at all Apr 03 21:04:57 strange Apr 03 21:05:05 woglinde: tls is in there for nptl Apr 03 21:05:06 for now I have a fix here Apr 03 21:05:09 otherwise it wont work Apr 03 21:05:35 which disable tls for e2fsprogs for oabi Apr 03 21:05:40 hm Apr 03 21:05:41 s Apr 03 21:05:47 I have to test uclibc too Apr 03 21:05:48 args Apr 03 21:06:24 woglinde: can you try nptl ? Apr 03 21:06:34 pffff Apr 03 21:06:40 I am lazy Apr 03 21:06:40 I remember your device was old enough not to support it Apr 03 21:06:42 *g* Apr 03 21:08:05 go go gadget multicore Apr 03 21:10:24 ant? Apr 03 21:10:37 ant clean; ant all Apr 03 21:11:53 woglinde: re (child was unhappy...) Apr 03 21:12:00 bx lr Apr 03 21:12:01 Apr 03 21:12:01 .endfunc Apr 03 21:12:01 ant oh okay sorry Apr 03 21:12:06 ant cool Apr 03 21:12:08 thanks Apr 03 21:12:17 so I can push it Apr 03 21:12:23 mwester, hey, happen to see my wireless-tools patch go by on the ml? Apr 03 21:12:27 stole it from newlibc Apr 03 21:12:28 *g* Apr 03 21:12:42 Tartarus is it on patchwork Apr 03 21:12:44 ? Apr 03 21:12:50 woglinde, yes it's there too Apr 03 21:13:05 is there a request reviews thing there too ive missed? :) Apr 03 21:13:56 hm no Apr 03 21:14:03 I would say put it in Apr 03 21:14:11 heh Apr 03 21:14:44 woop, gcc 4.1.2 built it right Apr 03 21:14:46 whats the problem? Apr 03 21:14:46 time to try 4.2.4 Apr 03 21:15:35 woglinde, w/ the patch? I want distro people to look over and see if they're ok with me fixing ifrename package (didn't exist before, does now) Apr 03 21:16:00 I'm not sure if there should be an RRECOMMENDS added to the main package or anything Apr 03 21:16:12 Tartarus: Yep, saw that -- makes sense. Apr 03 21:16:32 woglinde, so what should I do exactly? 1)take the lwn code 2)make it use +1 instead of a pointer 3)replace the the bx with a mov (I hope there won't be issues with address range) ? Apr 03 21:16:33 i never thought about it, but why would ifrename be in wireless-tools? Apr 03 21:17:04 mwester, I too wondered why it was part of that source package when I saw it Apr 03 21:17:21 but my problem was no libiw means opkg blew up making SDKs for me Apr 03 21:17:47 So at least now it's in its own package, which was intended Apr 03 21:17:52 if >1 thing have ifrename Apr 03 21:18:01 I've never tried putting that lib into an sdk, so never ran into that. Apr 03 21:18:02 (which, um, you don't need anymore with udev, do you?_ Apr 03 21:18:12 er, why not? Apr 03 21:18:16 mwester, task-sdk-base has it :| Apr 03 21:18:26 kept task-sdk-bare from going in Apr 03 21:18:34 mwester, lemme pull up the rule.. Apr 03 21:18:38 * mwester goes to check to see if he uses -base Apr 03 21:18:56 # Rename onboard 8139 to lan Apr 03 21:18:56 KERNEL=="eth*", SYSFS{address}=="00:11:5b:da:32:dc", NAME="lan" Apr 03 21:19:48 mwester, -base is a definate non-base task, it's got all sorts of stuff in it Apr 03 21:19:53 -bare is the real bare minimum Apr 03 21:20:00 -base is "here's the random libs I wanted at the time" task Apr 03 21:20:02 I don't think I use either! Apr 03 21:20:13 You should use -bare, or tell me whats wrong with it :) Apr 03 21:20:31 I spent a little time making it right for *libc and being just *libc and c++ Apr 03 21:20:56 03woglinde  07org.openembedded.dev * r4c0ca67259 10openembedded.git/recipes/ffmpeg/ (ffmpeg-0.5/armv4.patch ffmpeg_0.5.bb): ffmpeg: fix armv4 arch, patch can be zapped if gcc-4.4 is out Apr 03 21:21:07 task-slugos-toolchain-target.bb:# Stuff contained in this SDK is largely taken from task-sdk-base.bb. Apr 03 21:21:07 03woglinde  07org.openembedded.dev * r5a90bc8ba1 10openembedded.git/recipes/e2fsprogs/e2fsprogs.inc: Apr 03 21:21:07 e2fsprogs: disable tls for now, if we are oabi and uclibc Apr 03 21:21:07 * make comment for uclibc nptl Apr 03 21:21:20 :) Apr 03 21:21:23 heh Apr 03 21:21:28 I shall make a note to fix that. Apr 03 21:21:38 If there's still time for you, please look at the task-sdk-bare Apr 03 21:21:42 thanks Apr 03 21:26:32 pah, this qtopia doesn't like gcc 4.3.3, time to find something a bit newer Apr 03 21:26:41 i'm not fixing c++ code errors if i can help it ;) Apr 03 21:26:50 haha Apr 03 21:27:18 it's beyond missing previously implicit headers, so i'm not even sure wtf it means Apr 03 21:28:14 Tartarus: what is error ? Apr 03 21:28:38 painting/qdrawhelper.cpp:5905: error: explicit template specialization cannot have a storage class Apr 03 21:29:43 bravo thats one step closer to ISO complaince for gcc Apr 03 21:31:10 Tartarus: whats does code look like on this line Apr 03 21:31:32 I am afraid you need to fix it in source code which I believe you dont want to Apr 03 21:32:00 googled and fixed since it was easy, eventually Apr 03 21:32:07 go from static inline .. to inline and it compiled Apr 03 21:32:29 *g* Apr 03 21:32:37 btw, going to post patches to make our SDKs relocatible shortly now Apr 03 21:32:45 or at least make the toolchain Apr 03 21:32:54 haven't thought about how to fixup environment-setup file yet Apr 03 21:37:56 kergoth, around? Apr 03 21:38:15 kergoth, what's that case where you don't need to modify FILESDIR anymore? Apr 03 21:38:18 somewhat, taking a break from banging my head against the wall trying to get autotools to behave Apr 03 21:38:39 And would this fit, from binutils-cross-sdk_cvs.bb: FILESDIR = "${@os.path.dirname(bb.data.getVar('FILE',d,1))}/binutils-cvs" Apr 03 21:39:01 FILESPATHPKG .= ":binutils-cvs" would do basically the same thing Apr 03 21:39:22 thanks Apr 03 21:39:34 (PKG is the var that has things like ${PN}, files, ${P}, ${BP}, ..) Apr 03 21:40:52 Anyone, what's the git mailbomb script? Apr 03 21:41:03 Gonna split this up into 3 patches I think Apr 03 21:41:36 i usually just git format-patch then git send-email, but i think theres scripts around to do it in one command, somewhere Apr 03 21:41:49 send-email, thanks :) Apr 03 21:43:07 i always forget the arguments.. i think its --smtp-server, etc if you want to use smtp. it defaults to sendmail Apr 03 21:43:28 yay for mstmp :) Apr 03 21:43:45 you can put those in .gitconfig Apr 03 21:43:54 [sendemail] Apr 03 21:43:59 section Apr 03 21:49:56 kergoth, is that correct for 1.5 or is that a 1.6 thing? Apr 03 21:55:41 heh, using git-native i've built rather than host git for this :) Apr 03 21:56:50 heh :) Apr 03 21:57:21 kergoth: is it just an impression or parsing time is longer then it was time ago? Apr 03 21:57:29 not sure Apr 03 21:58:06 thanks all, git send-email over gmail example here: http://morefedora.blogspot.com/2009/02/configuring-git-send-email-to-use-gmail.html Apr 03 21:58:18 kergoth: weeks ago I noticed another change: size Apr 03 21:58:30 before I could build all in ram Apr 03 21:58:48 all = console-img + opie-img +x11-img Apr 03 21:59:05 with rm_work was 2.5 gb approx Apr 03 21:59:08 bitbake itself hasn't changed much lately Apr 03 21:59:22 days ago I retried but was out-of-space :/ Apr 03 22:00:50 ah..kernel seems to have skipped rm_work... Apr 03 22:01:09 Bitbake hackers: please consider this patch, works for me: http://bugs.openembedded.net/show_bug.cgi?id=5002 Apr 03 22:01:57 btw I'd propose write-access for Roman I Khimov Apr 03 22:02:10 seems knowing what he's doing Apr 03 22:02:18 Ah git Apr 03 22:02:37 for some reason, git send-email eats the patch description that I see in git format-patch created files Apr 03 22:02:40 sh#t.. Apr 03 22:02:51 Tartarus: thats odd.. Apr 03 22:02:53 * Jay7 forget again about ssh-keys :) Apr 03 22:02:53 I'll test Roman's fix for race-conditions with multi-images tonite Apr 03 22:03:05 Jay7: shame on you Apr 03 22:03:31 ~lart Jay7 Apr 03 22:03:32 * ibot pries Jay7's back open with a screwdriver and flashes a new bootldr to Jay7 Apr 03 22:03:35 * Jay7 is just lazy ;) Apr 03 22:03:46 ha ha..new bootloader ;-) Apr 03 22:03:56 muahahah :) Apr 03 22:04:07 seems I have new kexecboot :) Apr 03 22:04:32 * Jay7 now can boot from misc. media :) Apr 03 22:04:34 anyone else seen a failure in do_svnrev for u-boot-mkimage-openmoko-native? Apr 03 22:04:40 * Tartarus tries something Apr 03 22:04:48 also, why the heck is that being built for every one of my task-boot builds Apr 03 22:04:55 hmm Apr 03 22:05:19 mkimage? Apr 03 22:05:33 kergoth: offhand: why do we still use that mkimage? we have mkimage-native in OE Apr 03 22:05:38 i have no idea Apr 03 22:11:29 hoi kergoth Apr 03 22:14:33 * Tartarus does it the old fashioned way Apr 03 22:18:59 kergoth: for some reason rm_work won't clean my linux-* dir... Apr 03 22:20:02 well, just linux-rp* Apr 03 22:20:10 linux-kexecboot is clean Apr 03 22:21:29 515M linux-rp-2.6.26-r11/ Apr 03 22:21:32 argh Apr 03 22:23:13 log.do_rm_work_all.31185│ 0│Apr 4 00:16 Apr 03 22:24:19 hm.. linux-kexecboot has log.do_rm_work.7519│ 127│Apr 3 03:56│ Apr 03 22:24:41 do_rm_work_all != do_rm_work Apr 03 22:24:46 ? Apr 03 22:40:46 I had (sorry, I am studing english) built X11-image for x86 arch. Minimalist console image working fine in QEMU. But with x11-image I see only blackscreen in QEMU. Apr 03 22:43:52 Whether it is necessary to check up it under another virtual machine? virtualbox?.. Apr 03 22:45:22 try kvm Apr 03 22:45:37 Did you ever tested x11-image? Apr 03 22:45:44 nope Apr 03 22:45:51 hm let me think Apr 03 22:46:06 I tested a openmoko image with qemu so ages ago Apr 03 22:46:10 so yes Apr 03 22:46:22 ok, I'll try kvm :) Apr 03 22:46:32 or, how about virtual box? :) Apr 03 22:46:47 fpga: blackscreen going from start? or after starting X11? Apr 03 22:47:06 no, just after kernel starting Apr 03 22:47:30 in console image I see udev long time starting and succesfull bott up to root console Apr 03 22:47:38 kergoth: even doing bitbake -c rm_work won't clean... Apr 03 22:47:40 http://rafb.net/p/Sdx39a88.html Apr 03 22:48:08 kergoth: sh: rm: command not found Apr 03 22:48:09 fpga: press Ctrl+Alt+3 and look for serial console Apr 03 22:48:27 may be all is there Apr 03 22:48:34 ok Apr 03 22:49:08 ant yes the sh: rm: command not found I saw here too Apr 03 22:49:46 weh, comes and disappears...silly Apr 03 22:50:46 problem is it bloats by 500mb my workdir... Apr 03 23:00:27 my parameter here: qemu-system-i386 -M pc -kernel bzImage -hda disk.raw -append "root=/dev/hda1 console=/dev/ttyS0" -net nic,macaddr=00:16:3e:00:00:01 -net tap Apr 03 23:00:27 But I cant see anything on Ctrl+Alt+3, see only text "serial0 console" Apr 03 23:01:42 console=ttyS0,115200n8 Apr 03 23:02:25 then all output should go to serial Apr 03 23:02:33 kernel + init Apr 03 23:03:01 ok, 10x :) Apr 03 23:07:53 still nothing, even with minimalist console image, where everything working and booting, I see kernel boot redirected to console and gone from screen, but serial console0 empty... OK, I'll try other emulator... Apr 03 23:08:39 hm.. seems strange.. Apr 03 23:14:33 * ant__ is sure: the phase 'NOTE: Resolving any missing task queue dependencies' takes now longer... Apr 03 23:23:07 * Jay7 -> sleep() Apr 04 00:01:58 i'm trying to build xawtv and one of its dependent packages is failing with... http://www.pastebin.ca/1381378 Apr 04 00:06:42 kergoth: here? Apr 04 00:09:28 denix, any chance you have pointers to the .la issue? I must admit I never recall how that stuff gets (ab)used Apr 04 00:09:45 But, I _think_ it's already OK, thanks to sysroot support Apr 04 00:09:54 ie if it says -L/usr/lib/libfoo, etc, etc, we're good Apr 04 00:10:03 it's only a problem if stuff from the build path leaks Apr 04 00:10:08 and we have QA checks on that Apr 04 00:10:36 Tartarus: the .la files in sdk have absolute paths Apr 04 00:11:01 Ah, I see it now, I think Apr 04 00:11:10 I bet the answer is "stop doing that" Apr 04 00:12:06 :) stop doing what? Apr 04 00:12:43 If I read that regex right, it's turning /usr/lib and such into the hard-coded path Apr 04 00:13:05 But /usr/lib is fine as our toolchain will say "Ah, they mean ....//usr/lib" Apr 04 00:13:35 hmm... Apr 04 00:13:39 iirc, at least Apr 04 00:15:49 I see my .la files in sdk containing not just /usr/lib, but rather a complete /usr/local/angstrom/arm/arm-angstrom-linux-gnueabi/usr/lib Apr 04 00:16:03 Yeah, that's what the regex is setting up Apr 04 00:16:10 Lemme check something real quick Apr 04 00:16:21 then I'm missing something - what regexp? Apr 04 00:17:23 Ah Apr 04 00:17:30 So in meta-toolchain.bb: Apr 04 00:17:31 will LIBTOOL_SYSROOT_PATH take care of this? Apr 04 00:17:35 # Fix or remove broken .la files Apr 04 00:17:50 The big honkin' regex around that Apr 04 00:18:15 The .la file starts out with /usr/lib and we turn it into /usr/local/... Apr 04 00:18:27 Yeah, LIBTOOL_SYSROOT_PATH _might_ be something we're missing Apr 04 00:18:40 Along with leaving the paths in the .la files alone Apr 04 00:20:15 Bah, all I see there is a post from RP saying that's be a good thing :) Apr 04 00:20:18 yeah, I've seen that big sed regexp - so, should it be removed? Apr 04 00:20:46 I tried to fix this regexp Apr 04 00:20:52 but it failed Apr 04 00:20:58 so, will it append LIBTOOL_SYSROOT_PATH with /usr/lib, when installed=no? Apr 04 00:21:01 revert my last commit to this Apr 04 00:21:08 you should be a little better Apr 04 00:21:16 denix, it might be a bit more complex than that Apr 04 00:22:42 However it used to work, I suspect it was not quite right WRT libtool; but if you come up with a technique/solution/amendment to your earlier patchset, I'll build a patched version of the FSO sdk/toolchain and throw Qt-Extended at it. :) Apr 04 00:23:43 qt-extended, eh? :) Apr 04 00:24:01 Point me at your build scripts for qt-extended and i'll see what I can do about stuff and/or things Apr 04 00:24:03 Yeah, I'm one of those suckers who bought the Openmoko devices ;-) Apr 04 00:24:41 http://moko.mwester.net/qtextended.html Apr 04 00:25:34 thanks Apr 04 00:25:46 will poke again later, i should probably take a shower like i planned 30min ago now :) Apr 04 00:26:12 http://www.qtextended.org/downloads/snapshots/ <---- Qt Extended sources :) Apr 04 00:26:32 was it discontinued recently? Apr 04 00:28:49 depends on who you ask that question of Apr 04 00:29:24 * mwester ended a sentence with a preposition; his English teacher will roll over in her grave. Apr 04 00:29:48 Nokia/Trolltech is almost done with it; one more release they said, IIRC. Apr 04 00:30:09 Then "it's open source", so the users can keep it going if they wish. Apr 04 00:30:55 In the meantime, it's a good application stack for the Openmoko hardware, and for today's purpose, it's a really good workout for a C++ toolchain/SDK! :) Apr 04 00:31:15 what can qtextended do for us compared against qt gpl? Apr 04 00:31:41 qt extended is the old qtopia stuff. Apr 04 00:32:06 They renamed it just to confuse people, I think. Apr 04 00:32:14 i'm not too familiar with all the different versions, and I got confused indeed. Apr 04 00:32:36 Let's ask differently: why would the community want to keep qt extended, now that qt is GPL? Apr 04 00:32:43 or LPGL? Apr 04 00:32:55 grr, LGPL Apr 04 00:33:16 qt is X11 Apr 04 00:33:20 qt extended is an application set -- phone dialer, SMS handling, etc. I think Qt is just the toolkit underneath. Apr 04 00:33:24 qtopia/qtextended is FB Apr 04 00:33:34 qt is always tied to x11??? Apr 04 00:33:40 Yes, qt extended/qtopia is direct to the FB Apr 04 00:34:01 IIRC qtopia had some connection to the old opie stuff. Apr 04 00:34:08 hmm, strange they got rid of that. So Qt Software is now targetting non-embedded devices only? Apr 04 00:34:32 opie was an alternative app stack Apr 04 00:34:57 I can't say; I didn't follow it too closely. If you ask me, they just didn't even want to bother now that Google has moved into that space. Apr 04 00:35:11 (but that's only my opinion) Apr 04 00:36:10 mwester: but if you look outside the phone market, what open-source embedded options are there? gtk+, qtextended. I think Google has Android which is a complete stack. Apr 04 00:36:56 * like2wise realizes he is not thinking straight anymore - needs sleep, 02:36 AM here... Apr 04 00:37:27 ok, looks like qt now can do FB directly... http://www.qtsoftware.com/products/platform/qt-for-embedded-linux Apr 04 00:38:30 denix: Argh, so Qt for Embedded Linux != successor of Qt Extended? Apr 04 00:38:56 * mwester thinks likewise is thinking more clearly than most people; he's put his finger on an important (and overlooked) point. Apr 04 00:39:11 What _are_ the options for small embedded devices? Apr 04 00:39:20 Nokia -> maemo. Apr 04 00:39:23 no, that's full qt - see "Qt is available for these platforms:" menu Apr 04 00:39:53 And what else? KDE? Gnome? Not likely - they are targeting desktop. Apr 04 00:40:36 So IMO, Nokia is focussing on the tablet devices with maemo, and Google is taking on phones with Android, and the two are setting up for a massive battle where the two technologies meet. Apr 04 00:40:36 used to be Windows, OSX and X11, now they've added WinCE, s60 and Linux/FB Apr 04 00:41:34 I guess the difference between old qtopia/qtextended and modern Qt is in the apps stack Apr 04 00:42:04 In any case, what's interesting to me is that WinCE, Symbion seem to be the losers in this scenario; with the only wild-card out there being (IMO again), Palm. Apr 04 00:42:08 http://www.qtsoftware.com/downloads/downloads and then click (L)GPL: all libs are full source code, SDK's all but for embedded linux Apr 04 00:42:14 what about Moblin? Apr 04 00:42:15 But Palm may be too small to matter, either. Apr 04 00:42:26 Dunno about Moblin -- what about it? Apr 04 00:42:55 Oh - and I do a disservice to corporations all over the world by failing to mention that RIM is still in the game... Apr 04 00:43:32 I need to play with Qt / Embedded Linux (FB). I think it's the best option around probably (although GTK might be doable) Apr 04 00:43:47 "In April 2009 Intel turned Moblin over to Linux Foundation." Apr 04 00:43:52 (from wikipedia) Apr 04 00:44:01 yesterday - big news, btw Apr 04 00:44:07 I guess that tells me that Intel is no longer interested. Apr 04 00:44:22 supposedly to enable competing platforms (i.e. non Atom) :) Apr 04 00:44:24 what exactly is moblin? counterpart to android? Apr 04 00:44:35 countersolution to android, rather Apr 04 00:44:53 GNOME Mobile based distro Apr 04 00:45:01 for netbook and MIDs Apr 04 00:45:03 ok Apr 04 00:45:56 thanks for the info, I hope I'll not get confused about this any longer. I'm not really involved with GUIs so far... Apr 04 00:46:10 I didn't know about Moblin -- as a Gnome user, that sounds great, but I'm afraid that I just don't see it being other than a hobbyist solution, lacking the kind of muscle to take on Google or Nokia. Apr 04 00:46:34 I'm off to bed, thanks for the sharing! Apr 04 00:46:38 cya Apr 04 00:46:39 * mwester is not an Android fan, but is ok with maemo. Apr 04 00:46:41 like2wise: gn Apr 04 00:46:50 gn Apr 04 00:47:01 tnx, you two too later. Apr 04 02:51:29 anyone plays with linux device driver before? Apr 04 02:51:41 like embedded programming? Apr 04 02:51:45 any development kit or books? Apr 04 02:54:22 help Apr 04 02:54:24 hello Apr 04 02:54:25 hello Apr 04 02:57:48 hi:) Fxxx you Motorola EZX phone! Apr 04 02:58:09 Hey moderators, kick gDD out Apr 04 02:58:17 he is saying rude words here Apr 04 02:59:22 should say rude to Moto,they never know what is OPEN. **** ENDING LOGGING AT Sat Apr 04 02:59:57 2009