**** BEGIN LOGGING AT Sun Dec 10 02:59:58 2006 Dec 10 07:00:50 morning all Dec 10 08:09:25 morning all Dec 10 08:24:38 cyrilRomain : morning Dec 10 09:10:17 any python masters around ? Dec 10 10:15:56 Gerrath_: pushing directfb now Dec 10 10:17:22 zecke : morning. need some help with icecc.bbclass as my pythonese are not that good Dec 10 10:17:43 zecke : i've able to locate where the problem is Dec 10 10:17:55 okay :) Dec 10 10:18:05 will be here for some more minutes Dec 10 10:19:18 zecke : first it fails when looking if cross libc is installed -> os.stat(os.path.join(ice_dir, target_sys, 'lib', 'ld-linux.so.2')) Dec 10 10:20:13 hmm Dec 10 10:20:21 zecke : if i check for libc.so it finds it and proceeds Dec 10 10:21:40 do we have a dynamic linker (ld-linux.so.2) installed? Dec 10 10:21:41 zecke : then when it tries to generate the icecc env tar.gz it fails to add libc and friends, as it uses native objdump and friends Dec 10 10:22:20 zecke : not in the cross dir for powerpc Dec 10 10:23:08 zecke : it has ld-2.3.90.so Dec 10 10:23:30 oh this whole test is stupid :) Dec 10 10:23:47 RP: BitBake's EventHandlers need more information :) Dec 10 10:25:27 Gerrath_: you will need to fix directfb a bit :) Dec 10 10:26:01 zecke : basicaly what happens is that the package generated misses the libs and links to ccplus. as a result all jobs to remote machines fail and get compiled local Dec 10 10:26:33 Ifaistos: sounds plausible Dec 10 10:26:47 worst, it doesn't link remotely so ld-linux.so.2 is not needed at all Dec 10 10:26:57 zecke : i opened the generated .tar.gz and show it :) Dec 10 10:28:06 zecke : it's just checks for ld-linux.so.2 to verify the libc is there, but it does not need it anywhere Dec 10 10:29:18 Ifaistos: For some reasons I wanted gcc-cross to be compiled locally, and same for glibc Dec 10 10:29:28 so I wanted to check if the 'right' toolchain has been built already Dec 10 10:30:19 zecke : but isn't that handled by the blacklists ? Dec 10 10:30:40 Ifaistos: we have two blacklists in this case Dec 10 10:31:04 03Shane 07org.oe.dev * rec6d6f07... 10/ (9 files in 3 dirs): Dec 10 10:31:04 packages/directfb: Add rc2 of directfb from #1618 by Dec 10 10:31:04 Drop the old unfetchable versions of directFB and replace them with Dec 10 10:31:04 the shiny rc2 of directFB. Dec 10 10:31:04 This patch is from #1618 by Shane Volpe and adds tslib support to Dec 10 10:31:04 directFB. Dec 10 10:31:06 Ifaistos: and I do not think this is needed at all, we should create two different ICECC_VERSIONS depending on how much was built Dec 10 10:31:33 zecke : Actually we need 3 :) Dec 10 10:31:47 'n' Dec 10 10:31:47 zecke : One for native Dec 10 10:32:02 zecke : One (possibly) for cross-kernel Dec 10 10:32:19 brb Dec 10 10:32:19 zecke : One for all cross compiling Dec 10 10:33:54 * zecke had to umount a disk Dec 10 10:35:42 zecke : This is what icecc needs in the enviroment archive -> http://pastebin.ca/274032 Dec 10 10:37:57 zecke : and this is what goes into the icecc.bbclass build enviroment archive -> http://pastebin.ca/274037 Dec 10 10:37:59 Ifaistos: I wonder how expensive regenerating the tar.gz would be Dec 10 10:41:38 zecke : I believe we need 2 or best case 3 tar.gz files if we try to speed up all compiling processes. Dec 10 10:42:09 Ifaistos: one strategy would be to recreate the tar.gz and use the md5sum as name Dec 10 10:42:25 if the toolchain changes -> md5sum is different -> new toolchain will be uploaded Dec 10 10:42:37 if md5sum stays the same -> no toolchain will be uploaded Dec 10 10:43:10 Ifaistos: so I wonder how expensive creating the tar would be :} Dec 10 10:43:47 zecke : would the icecc --build-native work for the cross-compilers, as OE has all enviroment variables pointing to them ? Dec 10 10:44:55 zecke : i believe that the way its now done (by parsing the name of file, could be used but by parcing which CC is used Dec 10 10:44:55 Ifaistos: maybe, it calls the create-env script Dec 10 10:47:26 Ifaistos: I will need to encode some videos now but I will try to think about it Dec 10 10:47:59 zecke : i don't think there is a create-env script in recent versions of icecc.... it handles it directly (not sure though) Dec 10 10:51:03 zecke : i tried to modify the lib addition routine to something like this -> http://pastebin.ca/274047 Dec 10 10:51:53 zecke : but it does not seem to work.... and as i said my pythonese are not that good Dec 10 10:52:32 Ifaistos: well :) Dec 10 10:52:49 Ifaistos: when we pack the toolchain we need to package Dec 10 10:53:04 the compiler frontend gcc, g++, gcj Dec 10 10:53:12 the cc1 an cc1plus Dec 10 10:53:20 and the assembler as Dec 10 10:53:35 now as can be dynamically linked and the compiler suite as well Dec 10 10:53:53 this is why ld-linux.so.2 libc.so and libfd will be put in the tar by icecc Dec 10 10:55:42 zecke: I agree event handlers need more info :) Dec 10 10:57:46 Ifaistos: I will try to write a python version of create-env don't how this fits into my studying schedule Dec 10 10:58:39 zecke : with a little modification i think the existing bbclass would work Dec 10 10:58:49 zecke : it's half way there Dec 10 10:59:23 Ifaistos: yes, creating the 'right' tar.gz is the tricky part :) Dec 10 10:59:37 Ifaistos: all we need to know is which gcc to use Dec 10 10:59:50 this will tell us the site file, and where the cc1 and cc1plus will be Dec 10 11:29:06 hi zecke Dec 10 11:29:54 Hello Dec 10 11:49:26 re Dec 10 12:46:50 Hi all, I need help with an oz3541 akita build. I get cross compile badness on gpe-plucker. Here is the error: http://pastebin.ca/274122 Dec 10 12:58:42 DoTheDog: did you try to clean it Dec 10 13:02:04 Yeah, I still get the cross compile badness. It looks like it is looking for /usr/local/gpe which doesn't exist. Dec 10 13:02:54 I actually had the same problem with gpe-mixer, but I added a patch from the previous version and it looks like it fixed it. The patch on this doesn't seem to help. Dec 10 13:15:55 morning Dec 10 13:16:29 Morning chouimat, you know how to fix cross compile badness? Dec 10 13:16:50 DoTheDog: do not include -I/usr/include ? Dec 10 13:17:09 DoTheDog: normally the buildsystem deservers some fixes Dec 10 13:21:27 The patch does this: -CFLAGS += -I/usr/include/gpe Dec 10 13:21:28 +PACKAGE_CFLAGS += -I/usr/include/gpe Dec 10 13:21:50 This seemed to work on gpe-mixer. Dec 10 13:24:29 so apply knowledge to solve it? Dec 10 13:24:47 damn too advance for me :-S Dec 10 13:25:01 lol linux >.< Dec 10 13:25:57 Zecke: hail Dec 10 13:26:09 Zecke: are you comming to froscon ? Dec 10 13:26:13 gerwinin: how are you? could you send me your current mail address? Dec 10 13:26:27 Zecke: okay gerwin@vanderkruis.net Dec 10 13:26:45 Zecke: Did koen tell you about event :) Dec 10 13:26:46 gerwinin: I will be at FOSDEM for sure and at the chaos congress Dec 10 13:26:48 That is about the extent of my knowledge ;) Dec 10 13:27:01 Zecke: for me fosdem for sure chaos maybe Dec 10 13:27:39 DoTheDog: Well, you try to update gpe packages? Dec 10 13:28:39 zecke: You mean mtn? Dec 10 13:29:46 DoTheDog: no, the question is: Does this happen with the 'stock' gpe? Dec 10 13:29:52 DoTheDog: or did you change anything? Dec 10 13:30:22 zecke, this is 'stock' gpe. I just run bitbake gpe-image Dec 10 13:31:13 DoTheDog: which configuration? Dec 10 13:31:46 zecke: oz3541, akita, 2.6 kernel Dec 10 13:33:46 DoTheDog: hmm autobuilds of oz3541 currently fail on coreutils-native Dec 10 13:34:16 DoTheDog: solution is easy: You need a text editor and the capabilities to use it Dec 10 13:34:22 zecke: how do I tell if I have gotten that far. Dec 10 13:34:34 zecke: You mean edit the .bb file? Dec 10 13:34:34 1.) find out who includes something from /usr/include /space/include /usr/local/include Dec 10 13:34:47 2.) use quilt to create a new patch Dec 10 13:34:54 3.) edit the bad make file Dec 10 13:34:59 4.) test compile Dec 10 13:35:07 5.) go to 3. ot 6.) Dec 10 13:35:10 6.) quilt refresh Dec 10 13:35:15 7.) add this patch to the bb file Dec 10 13:35:18 8.) send patch Dec 10 13:35:25 Zecke: I am updating the mythtv bitbake file now Dec 10 13:35:51 Because 0.18 does not compile with gcc4 Dec 10 13:36:09 zecke: thanks I'll give that a try. Dec 10 13:38:20 Zecke: Will oe have a booth during fosdem ? Dec 10 13:40:11 zecke: Can you take a look at bug 1667? Dec 10 13:40:15 !oebug 1667 Dec 10 13:40:16 * * Bug 1667, Status: NEW, Created: 2006-12-09 18:17 Dec 10 13:40:17 * * hma(AT)syd.odn.ne.jp: gpe-mixer-0.42-r0: CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include/gpe Dec 10 13:40:18 * * http://bugs.openembedded.org/show_bug.cgi?id=1667 Dec 10 13:40:40 * Laibsch points to topic as a hint for gerwinin Dec 10 13:41:24 gerwinin: yes Dec 10 13:42:35 DoTheDog: Maybe bug 1667 is interesting for you as well. Dec 10 13:46:26 Zecke: Shall I bring some demo machine with mini-itx ? Dec 10 13:46:39 gerwinin: if it is possible, sure Dec 10 13:46:49 Zecke: I have : set top box , asterisk machine , and I am busy with wifi radio Dec 10 13:47:30 gerwinin: we want to expose something, we have no plan what we will expose though Dec 10 13:47:47 Zecke: I think I can help here a bit as well Dec 10 13:48:50 pb_: Can you take some time and comment on bug 1353? Dec 10 13:48:58 Zecke: I am planning to attent some more events as well , but do no want to stand there in an oe booth without having this thing discussed with yiu guys Dec 10 13:49:39 Laibsch: I'll take a look Dec 10 13:49:47 thanks a lot. Dec 10 14:18:18 zecke: I just commented out the +PACKAGE_CFLAGS += -I/usr/include/gpe line in the makefile and it seems to have worked. BTW when you say send patch, what do you mean? Dec 10 14:20:26 Laibsch, I fixed the bug for myself in gpe-mixer by adding the line: SRC_URI += "file://fix_makefile.patch;patch=1" to the end of the bb file. It seemed to work. It is just picking up the 0.32 patch file. Dec 10 14:21:56 DoTheDog: Can you put a comment to that effect in bug 1667 so that Takeshi benefits as well until a real fix is committed? Dec 10 14:22:15 DoTheDog: append patches to bugs/bugtracker, find the maintainer and ask him Dec 10 14:26:16 Laibsch: sure will do Dec 10 14:26:39 zecke: I'll attach it to the bug. Dec 10 14:29:14 later Dec 10 14:29:32 tmbinc_: hi! Dec 10 14:33:35 zecke: Thanks! Dec 10 14:48:51 * gerwinin just repaired his scooter Dec 10 15:05:53 !oebug 1353 Dec 10 15:05:54 * * Bug 1353, Status: NEW, Created: 2006-08-23 08:09 Dec 10 15:05:55 * * hma(AT)syd.odn.ne.jp: gpe-session-scripts shouldn't depend on gpe-bluetooth Dec 10 15:05:56 * * http://bugs.openembedded.org/show_bug.cgi?id=1353 Dec 10 15:08:19 <_gerwinin> leave Dec 10 15:38:55 re Dec 10 15:44:06 yo Dec 10 15:45:57 mickeyl: g'day Dec 10 15:46:25 hey pb_ Dec 10 15:48:22 hey mickeyl and pb_ Dec 10 15:48:32 and koen too Dec 10 15:48:43 hey mickeyl Dec 10 15:49:02 mickeyl: the frankfurt skyline by night looks nice from the autobahn Dec 10 15:49:27 hello all Dec 10 15:49:35 hey likewise Dec 10 15:52:34 hey koen, mickeyl Dec 10 15:52:56 Is it just coincidence the two of you reappear here together? Dec 10 15:53:10 coincidence Dec 10 15:53:44 koen: that's one thing I adore -- it's why i chose the frankfurt skyline for my website logo Dec 10 15:53:47 although I did pass frankfurt this weekend Dec 10 15:53:57 they are the same person, just different personalities Dec 10 15:54:15 heh Dec 10 15:54:21 i'm jeckyl, koen is mr.hide Dec 10 15:54:25 :D Dec 10 15:54:30 ade|desk: I know for a fact that is not true. Dec 10 15:54:33 :) Dec 10 15:54:40 prove it ! Dec 10 15:55:06 one side thinks he is from .nl the other .de Dec 10 15:55:27 ade|desk: I met them. Dec 10 15:55:32 that's what I call split personality Dec 10 15:55:35 paid actor Dec 10 15:55:36 Both. At the same time. Dec 10 15:55:58 ade|desk: I did not say I have proof to convince you. I just said I know for a fact that ... Dec 10 15:56:17 Maybe me, koen and mickeyl are all the same? Dec 10 15:56:24 Me is the dumb guys part ;-) Dec 10 15:56:27 heh Dec 10 15:56:29 :D Dec 10 15:56:46 all of you guys booked the Astrid yet? Dec 10 15:56:48 Did you guys book one room at FOSDEM then? Dec 10 15:56:50 if not, go ahead Dec 10 15:56:53 heh Dec 10 15:56:54 ah ... multi personalities ... now that is a brain fsck Dec 10 15:56:55 i did Dec 10 15:56:56 koen: re bug 1667, so this is something that hrw needs to push into oz354x? Dec 10 15:57:02 I booked. Dec 10 15:57:18 !oebug 1667 Dec 10 15:57:20 * * Bug 1667, Status: NEW, Created: 2006-12-09 18:17 Dec 10 15:57:21 * * hma(AT)syd.odn.ne.jp: gpe-mixer-0.42-r0: CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include/gpe Dec 10 15:57:21 * * http://bugs.openembedded.org/show_bug.cgi?id=1667 Dec 10 15:57:30 HopsNBarley: ping Dec 10 15:57:37 Laibsch: it's a series of patches Dec 10 15:57:54 HopsNBarley: "One silent ping only, yes'r." Dec 10 15:58:18 likewise: got your efika yet? Dec 10 15:58:45 yes, friday Dec 10 15:58:45 koen: how goes yours? xgl'ed up yet ? Dec 10 15:59:21 koen: I don't have an ATX brick here though. Dec 10 15:59:58 ade|desk: I've been away this weekend Dec 10 16:00:10 morning all Dec 10 16:00:26 ~lart rel for lack of chroot Dec 10 16:00:27 * ibot beats rel over the head with a microkernel for lack of chroot Dec 10 16:00:35 s/rel/rhel/ Dec 10 16:01:18 hey hrw Dec 10 16:01:23 and lack of chroot does ot allow me to check my 'bitbake bitbake chroot' Dec 10 16:01:55 Oh, and hrw is back to. What a merry party. Dec 10 16:02:02 * Laibsch waves to hrw. Dec 10 16:02:38 hi Laibsch Dec 10 16:02:53 its sunday afternoon, no good tv until top gear ;) Dec 10 16:03:00 Laibsch: .oz354x is supposed to have some breaks since Ipushed one fix for gcc Dec 10 16:03:07 koen: that was not really the question. The question is whether this is something that Marcin needs to take care of. Dec 10 16:03:11 ade|desk: I lack topgear here.... Dec 10 16:03:40 hrw: OK. Should people compiling stuff for themselves be advised to compile against .dev or against .oz354x? Dec 10 16:03:43 mickeyl: I booked single room in astrid about week ago Dec 10 16:04:04 good Dec 10 16:04:23 Laibsch: for oz 354+ - .oz354x for working, dev for experimenting Dec 10 16:04:31 wow the bbc don't transmit to poland yet... must be the last place in europe that they don't Dec 10 16:04:52 ade|desk: one of polish statons transmit topgear Dec 10 16:05:21 I just do not feel a need to pay 10EUR/month for few channels which I very rarely look Dec 10 16:08:18 hrw: Is the bogus local include patch from zecke in .oz354x as well? http://www.openembedded.org/the-q-a-team-strikes-back Dec 10 16:08:18 Laibsch: I would not call it bogus Dec 10 16:08:19 and it is in Dec 10 16:09:24 ~lart chroot Dec 10 16:09:25 * ibot drops a baby grand on chroot Dec 10 16:09:25 hrw: Wording taken from the website. I think it is "bogus local include"-patch, not bogus "local include"-patch ;-) Dec 10 16:09:34 ah. Dec 10 16:09:42 use '' next time ;d Dec 10 16:14:16 some see Pigi around in last months ? Dec 10 16:14:41 ~see pigi Dec 10 16:14:43 * ibot whispers to pigi "You didn't see anything...." Dec 10 16:14:50 ~seen pigi Dec 10 16:15:11 pigi was last seen on IRC in channel #asterisk, 194d 2h 5m 18s ago, saying: 'is there anyone that has successfully got asterisk working with eicon diva pci cards (passive cards ) ?'. Dec 10 16:15:27 pigi is MIA Dec 10 16:15:46 cu Dec 10 16:15:55 cu Dec 10 16:15:58 end of battery in not mine laptop Dec 10 16:16:02 :-/ Dec 10 16:21:08 koen: are you aware of http://www.diy-linux.org/ Dec 10 16:21:19 I am now :) Dec 10 16:22:37 koen: Greg is a (anal) perfectionist in building bootstraps from hand - I always keep it handy when comparing different build tool bootstraps (CLFS, OE, ...) Dec 10 16:23:41 koen: and there is PowerPC support in his bootstrap lately. Dec 10 16:25:11 likewise: the next commit is for you :) Dec 10 16:26:03 koen: you mean, I am about to commit something, or, as-in, koen is committing a present for me? Dec 10 16:26:15 a present Dec 10 16:26:38 03koen 07org.oe.dev * r99f274fa... 10/ (3 files in 3 dirs): glibc 2.5: fix some funkiness with mathinline.h Dec 10 16:26:40 that's must be a fix to my lighttpd .bb then... Dec 10 16:27:05 no, a patch to get glibc 2.5 to build for powerpc :) Dec 10 16:27:34 Ah. Well actually, I just started my first OE PPC build half an hour ago, so that will probably fail then :-) Dec 10 16:27:56 koen: In the meantime, I was reading up on diy-linux. Dec 10 16:29:52 hrw, hi !!! ... MIA ??? M... in action ??? Dec 10 16:30:43 koen: so an in-advance thank-you is in place already: thank you! :-) Dec 10 16:33:08 hey koen Dec 10 16:33:12 hey anushsh Dec 10 16:33:34 koen : wassup, how you doin Dec 10 16:54:31 morning kergoth Dec 10 16:55:43 hey pb_, been a while. how ya been? Dec 10 17:02:43 hey likewise! Dec 10 17:03:18 HopsNBarley: Hey there. Wanted to ask if your distro is uclibc by default? Dec 10 17:04:53 no, it's not. i haven't even tried that. Dec 10 17:04:59 kergoth: pretty good, thanks. you? Dec 10 17:05:20 HopsNBarley: koen just committed a present for us, making glibc 2.5 build for powerpc. Dec 10 17:06:12 I think HopsNBarley has a similar local fix for that Dec 10 17:09:28 my fix was not terribly elegant: i just used 2.3.2 as "initial", then 2.5 intermediate and final built fine. Dec 10 17:10:20 i did see koen's commits of the recent patch as well as 2.5 initial, and i of course applaud that! dividends of the efika? Dec 10 17:10:47 yep Dec 10 17:14:05 koen you get the 5200? Dec 10 17:14:21 as cpu? Dec 10 17:14:22 yes Dec 10 17:15:42 it's installing debian right now, when that works, I'll chroot to and OE built rootfs Dec 10 17:22:49 * koen Doh's Dec 10 17:23:07 D'oh! Dec 10 17:23:15 03koen 07org.oe.dev * r76f1fdf6... 10/ (1 conf/machine/efika.conf): efika: include ppc603e in extra archs Dec 10 17:27:56 Hello Dec 10 17:28:04 hey B_Lizzard Dec 10 17:28:18 I updated the bb file for emelfm, but I'm having two problems Dec 10 17:29:38 First, I need to know if I can export BUILD_STRIP, which is used in the do_compile script to something other than the default, cause the default doesn't work Dec 10 17:29:52 Can I just export it in the .bb file? Dec 10 17:30:33 Cause, in the do_compile script it's "BUILD_STRIP="strip", which needs to be "BUILD_STRIP="sh3-linux-strip" Dec 10 17:31:04 a build should not run strip by itself Dec 10 17:31:11 that breaks debug packages Dec 10 17:31:22 Also, when I do that manually, the package completes, but in the ipk file, "control" lists no dependancies. Dec 10 17:31:38 so 'export BUILDSTRIP=echo' would be one way to solve it Dec 10 17:31:42 Actually, changing it to that made it make the dbg packages Dec 10 17:32:16 But I'll just change it to echo, if that's the right way to go about it Dec 10 17:32:42 Would exporting in in the .bb file do the trick? Dec 10 17:35:14 Also, what about that dependancy thing? Dec 10 17:35:44 In the bb file, gtk+ is listed in DEPENDS Dec 10 17:53:32 How do I make sure that libcap/include/sys/capability.h is included when compiling libcap? It seems like this is not the case at the moment. Dec 10 17:54:09 I believe I need something like "-I$PATH". My question is what is the right path? Dec 10 17:54:24 And where do I set via bb file? Dec 10 18:35:50 koen: you optimist, mplayer isnt autotooled Dec 10 18:36:37 hi Dec 10 18:36:52 hi nicolasfr Dec 10 18:36:57 XorA|goner: it just smells like it Dec 10 18:37:43 hrw: that stench never washes off :-) Dec 10 18:38:08 I have two recipes for OE (libsdl-sharprom and scummvm-sharprom). What's the best way to submit them (I use to tar them and attach to the bugtracker, but never asked if it was the way to do it)? Dec 10 18:39:15 nicolasfr: I prefer stuff I can read directly :) Dec 10 18:39:15 nicolasfr: how they differfrom normal ones? Dec 10 18:39:29 hrw: let me guess Dec 10 18:39:41 hrw: gcc2 has no __attribute__((constructor)) :) Dec 10 18:40:05 and I feel bad if we support proprietary platforms... Dec 10 18:40:13 hrw: for scummvm it needs a few patches in the code to compile with gcc-2.95 Dec 10 18:40:38 nicolasfr: SRC_URI_sharpcrap += ""? Dec 10 18:40:39 and for libsdl I based on libsdl-qpe but removed OPIE references (would not work with) Dec 10 18:41:57 hrw: oh yes for scummvm that can work, but not for libsdl as current patches are incompatible Dec 10 18:42:43 HopsNBarley: Is that a u-boot kernel image the protium build fails on?: http://pastebin.ca/274467 Dec 10 18:43:16 zecke: ok, what's the best way to do it then (so you can read directly). I tried mtn --diff but it did not take my new recipes Dec 10 18:43:33 mtn add, mtn diff Dec 10 18:43:38 mtn list unknown Dec 10 18:43:42 and then adding them Dec 10 18:43:58 zecke: ok, didn't know all this commands ;) Dec 10 18:44:05 likewise, checking... Dec 10 18:44:08 and mark them as text/plain in bugtracker Dec 10 18:46:11 likewise, it's wierd a uImage isn't getting made. Dec 10 18:47:15 check your run.do_compile: in kernel_do_compile you should see "oe_runmake uImage ..." Dec 10 18:47:43 HopsNBarley: No, not there. Let me show you my run.do_compile Dec 10 18:48:15 does anyone encounter this bug recently ? http://rafb.net/paste/results/57AT3J54.html Dec 10 18:48:52 ar is indeed missing from staging/target-sys/bin and a wrong one it choosen. binutils-cross store them in cross/target-sys/bin but that path is missing from $PATH Dec 10 18:49:12 HopsNBarley: Uhm, I was in run.do_install, run.do_compile *does* have it. Dec 10 18:51:00 hrw: I have to make a different recipe for scummvm-sharprom because it uses a different do_compile() from the current recipe Dec 10 18:51:21 do_compile_sharpcap() { Dec 10 18:51:56 HopsNBarley: log.do_compile shows: NOTE: make -e MAKEFLAGS= uImage CC=ccache powerpc-linux-gcc LD=powerpc-linux-ld Dec 10 18:51:57 ~curse OEE shlib code Dec 10 18:51:58 CHK include/linux/version.h Dec 10 18:51:58 this patch solves the problem but I'm afraid this is not the right way to fix it http://www.rafb.net/paste/results/h4kd5R53.html Dec 10 18:52:09 May you be reincarnated as a Windows XP administrator, OEE shlib code ! Dec 10 18:52:12 hrw: oh didn't know it was possible, good Dec 10 18:52:12 fscking shit killing ALL dependencies Dec 10 18:52:22 FSCK FSKC FSCK! Dec 10 18:52:43 hrw: I had that same error a few days ago, and now it is suddenly away again after an update.... Dec 10 18:53:08 likewise: I build iin Oregon, US. test in Poznan, PL Dec 10 18:53:32 I cheated and put angstrom-foo-objdump in /usr/local/bin Dec 10 18:53:39 both machines firewalled so have to use ssh chains Dec 10 18:53:41 hi hrw ... what MIA is for : M.... in ction ? Dec 10 18:53:57 Missing Dec 10 18:54:09 gremlin[it]: yep. he do no give any kind of life signs Dec 10 18:54:23 :P Dec 10 18:55:01 koen: btw - THX for i86-angstrom-linux Dec 10 18:56:15 hrw: so your test cycle is kind of screwy? Dec 10 18:57:01 ? Dec 10 18:57:05 hrw: whats gone wrong with shlibs now? Dec 10 18:57:34 XorA|goner: as before - lack of dependencies in packages Dec 10 18:57:56 hrw: I begin the think we hitting a bug in binutils :-( Dec 10 18:58:08 XorA: i686 target on i686 machine - anngstrom-2007.1/progar Dec 10 18:58:53 hrw: I wonder if it might be an idea to send stderr of objdump command to bitbake log Dec 10 19:00:23 and it fsckup one of my recent works Dec 10 19:03:24 if I use SRC_URI_sharprom-compatible += "myfile" the original package is not unpacked anymore Dec 10 19:04:21 Depends: nano-locale* Dec 10 19:04:24 thats all Dec 10 19:04:42 hen use SRC_URI_sharpcarap_append = Dec 10 19:05:03 hrw: thanks, let's try Dec 10 19:05:20 XorA: when I tried debugging the problem (showing stderr) it showed something like "Could not find package providing shlib blabla.so" Dec 10 19:05:34 so OE generate RedHat5 systems for me... dependency hell track Dec 10 19:06:01 likewise: your looking in totally the wrong play Dec 10 19:06:17 likewise: modify package.bbclass to output the stderr of objdump Dec 10 19:06:25 likewise: instead of sending it to /dev/null Dec 10 19:06:31 HopsNBarley: linux-storcenter depends on u-boot (mkimage) Dec 10 19:06:41 XorA: that's exactly what I did. Dec 10 19:06:41 hrw: does not work, I wonder if the "-" in "sharprom-compatible" could be the problem? Dec 10 19:07:02 XorA: the objdump seems to work ok. Dec 10 19:07:04 nicolasfr: hm. no idea now Dec 10 19:07:41 likewise: then print the results of the matching code and find out why is messes up then Dec 10 19:07:41 likewise: but if you get Could not find package providing shlib blabla.so Dec 10 19:07:51 you are looking at the wrong package Dec 10 19:08:03 you need to look at package.bbclass run for blabla Dec 10 19:09:24 XorA: ok. I see, I probably went into shlibs.bbclasses also, when I noticed the objdump stuff was OK. Where are the IRC logs for this channel? Dec 10 19:09:32 XorA: cecking in progress Dec 10 19:09:34 XorA: I think I pastebinned everything Dec 10 19:09:46 * hrw does not like to use laptop keboards... Dec 10 19:10:02 or any keyboards other then ergo ones ;) Dec 10 19:10:03 likewise: http://ibot.rikers.org/%23oe/ Dec 10 19:10:19 hrw: I am wondering if file == *.so* and objdump barfs if we should error on that file Dec 10 19:10:40 hrw: that would at least alert to error quicker Dec 10 19:11:05 rebilding nano.. Dec 10 19:11:20 dpkg-deb n redhat system is nice combo Dec 10 19:11:26 s/ n / on / Dec 10 19:12:09 sorry, gotta go. would like to see into this nicolasfr: My guess would be SRC_URI_sharprom-compatible_append = "myfile" (without the +) Dec 10 19:13:11 How do I make sure that libcap/include/sys/capability.h is included when compiling libcap? It seems like this is not the case at the moment. Dec 10 19:13:20 I believe I need something like "-I$PATH". My question is what is the right path? Dec 10 19:13:26 And where do I set via bb file? Dec 10 19:13:32 Dynamic Section: NEEDED libncurses.so.5 NEEDED libgcc_s.so.1 NEEDED libc.so.6 Dec 10 19:13:40 Laibsch: I did use _append without the + Dec 10 19:13:45 thats 4 lines (wrong pasted) Dec 10 19:14:08 Laibsch: extra cflags you need Dec 10 19:14:23 hrw: looks like it working then Dec 10 19:14:34 XorA: its objdump by hand Dec 10 19:16:55 hm. output of objdump on arm and i86 looks same... Dec 10 19:16:59 time to add debugs Dec 10 19:17:47 koen: any idea about my (binutils?) related issue ? Dec 10 19:24:09 ~lart python indentation Dec 10 19:24:10 * ibot cuts off python indentation's head with a halberd that could have been a little bit sharper Dec 10 19:24:21 'NOTE: Couldn't find shared library provider for libgcc_s.so.1' in do_stage of binutils-2.17.50.0.5-r0. Should I consider this note as a warning or an error ? Dec 10 19:24:49 normally warning.. recentl it came to error Dec 10 19:24:59 hrw: thx Dec 10 19:25:15 Working on bugs 1628 and 1659 here (libcap build failures). libcap depends on bison and flex for do_configure. In openzaurus that is taken care of via the gcc-cross -> binutils-cross dependency. This is assumed provided for Sharp ROM AFAIK and thus this dependency is not fulfilled automatically -> http://rafb.net/paste/results/AP0d7l31.html Dec 10 19:25:45 then add them Dec 10 19:25:48 Any objections to specifying explicitly in libcap bb file for "DEPENDS = bison-native flex-native"? This should not break anything. Dec 10 19:25:48 ? Dec 10 19:26:14 hrw: You know, I am still on probation. When handling stuff out of CJKV I rather ask before committing ;-) Dec 10 19:26:22 sure Dec 10 19:26:42 will do, then Dec 10 19:27:00 Laibsch: comment it why iin comment log Dec 10 19:27:18 sure Dec 10 19:27:47 Does bitbake REQ python 2.4?? Dec 10 19:28:40 hrw: maybe, insanity.bbclass does require py2.4 Dec 10 19:29:14 any sane distribution has py2.4 as default nowadays Dec 10 19:29:15 zecke: lines 579-581 of package.bbclass des not work for me Dec 10 19:29:22 zecke: RHEL4 is not sane then Dec 10 19:29:45 hmm Dec 10 19:30:35 os.popen should work :) Dec 10 19:30:46 fd.readlines() as well Dec 10 19:30:49 and fd.close too Dec 10 19:30:52 selinux killing you? Dec 10 19:31:23 XorA: maybe.. but it looks disabled in /etc/ configs Dec 10 19:31:32 according to help from fedora developers Dec 10 19:31:43 hrw: lrg is a fedora guy Dec 10 19:31:58 anyway it looks like fs.readlines() does not get any data Dec 10 19:34:16 objdump by hand works. but not in package.bbclass.. I suspect PATH Dec 10 19:36:39 YES! Dec 10 19:36:42 PATH IS FSCKED Dec 10 19:37:02 koen cheat with cp OBJDUMP $PATH works Dec 10 19:37:35 bitbake:/data/build/oe/bitbake# ls /usr/local/bin/ Dec 10 19:37:36 arm-angstrom-linux-gnueabi-objdump i686-angstrom-linux-objdump powerpc-angstrom-linux-objdump Dec 10 19:37:38 so package.bbclass DOESNOT get cross/bin into PATH and fails silently Dec 10 19:38:42 like|bbiab - thanks for the bug report. Dec 10 19:38:44 03koen 07org.oe.dev * rf364e3e8... 10/ (3 files in 3 dirs): minimo: apply patch from mozilla bugzilla in an attempt to fix #1600 Dec 10 19:40:53 hm.. bitbake -b nano_1.3.9.bb -f -cpackage -e|grep PATH show proper setting Dec 10 19:41:07 OBJDUMP is also set correctly.. Dec 10 19:41:07 and without -b? Dec 10 19:41:29 koen: without -b I have to wait few minues for parsing Dec 10 19:41:56 and -e is not quie ok then iirc Dec 10 19:44:20 Ihave ugly fix Dec 10 19:45:13 hi Dec 10 19:46:11 [hrw@misc1 classes]$ mtn dif package.bbclass Dec 10 19:46:11 # Dec 10 19:46:11 # old_revision [76f1fdf6dff425bfdddcd67c4430bf4bc23ab36c] Dec 10 19:46:11 # Dec 10 19:46:12 # patch "classes/package.bbclass" Dec 10 19:46:14 # from [b4538ccc18abcaa057513563f37b2d25fdce5784] Dec 10 19:46:17 # to [969285b9af2e3bc59048d34cf396dd529e496c1e] Dec 10 19:46:19 # Dec 10 19:46:22 ============================================================ Dec 10 19:46:24 --- classes/package.bbclass b4538ccc18abcaa057513563f37b2d25fdce5784 Dec 10 19:46:27 +++ classes/package.bbclass 969285b9af2e3bc59048d34cf396dd529e496c1e Dec 10 19:46:29 @@ -576,6 +576,7 @@ python package_do_shlibs() { Dec 10 19:46:32 path = os.path.join(root, file) Dec 10 19:46:34 if os.access(path, os.X_OK) or lib_re.match(file): Dec 10 19:46:37 cmd = bb.data.getVar('OBJDUMP', d, 1) + " -p " + path + " 2>/dev/null" Dec 10 19:46:40 + cmd = "PATH=\"%s\" %s" % (bb.data.getVar('PATH', d, 1), cmd) Dec 10 19:46:43 fd = os.popen(cmd) Dec 10 19:46:45 lines = fd.readlines() Dec 10 19:46:48 fd.close() Dec 10 19:46:50 [hrw@misc1 classes]$ Dec 10 19:46:53 SHIT Dec 10 19:46:55 sorry.. mswin +me == bugs Dec 10 19:46:58 but thats FIX for dependency problem Dec 10 19:47:01 :) Dec 10 19:47:06 http://pastebin.ca/274564 Dec 10 19:47:18 please test it and commit if it will work for you Dec 10 19:48:06 cu Dec 10 19:48:17 battery out again Dec 10 19:51:48 is there a reason for ${CROSS_DIR}/${TARGET_SYS}/bin not being in the path ? (I still wonder if it is a binutils bug or a PATH one) Dec 10 19:52:24 cyrilRomain: read back a few lines Dec 10 19:52:38 cyrilRomain: you seem to be ignoring the discussion on the same bug you are puzzling Dec 10 19:53:25 XorA: not ignoring but not sure it was the same issue Dec 10 19:53:35 XorA: thanks Dec 10 20:04:10 I got some troubles while building glibc on my amd64,it segfault while generating locale.I saw it should be fixed by adding ENABLE_BINARY_LOCALE_GENERATION = 0 to my local.conf .but it didn t worked. Dec 10 20:06:56 magnet: problem is qemu-arm-* segfaults on amd64 Dec 10 20:07:15 magnet: if you change that option I think you need to -c clean glibc and redo Dec 10 20:08:29 I use native qemu,is that a problem ? Dec 10 20:08:34 magnet: on openSuSe that worked for me Dec 10 20:08:50 I m running gentoo. Dec 10 20:08:59 magnet: I assume they have some magic patch in the rpm (or its an i586 package) Dec 10 20:09:49 I tried cleanup / redo I get same issue. Dec 10 20:11:37 ENABLE_BINARY_LOCALE_GENERATION = "0" is certainly what I use Dec 10 20:13:36 erm gaim crashed. not my day :) Dec 10 20:15:37 20:43 < XorA> ENABLE_BINARY_LOCALE_GENERATION = "0" is certainly what I use Dec 10 20:16:53 do you put it at some special place in the config file. Dec 10 20:17:02 ? Dec 10 20:17:04 * koen wonders why scotland is in GMT -0:06 Dec 10 20:17:55 koen: because linux cant keep time very good :-) Dec 10 20:18:12 must go cook Dec 10 20:18:57 cya. Dec 10 20:20:03 QOTD: A student who changes the course of history is probably taking an exam. Dec 10 20:29:38 When I am in org.openembedded.dev dir I have to say "mtn --db=../oe.mtn pull monotone.openembedded.org org.openembedded.dev". Is there any way I can shorten this to "mtn pull" Or should I issue these commands from the OE top-level directory? Dec 10 20:30:06 Laibsch: sure Dec 10 20:30:19 Laibsch: you can set default server and probably even default glob Dec 10 20:31:28 zecke: How? http://www.venge.net/monotone/docs/Tutorial.html only says this should work. But it does not. I have already tried --set-default but it does not look like it worked. Dec 10 20:32:41 re Dec 10 20:34:13 Laibsch: good question Dec 10 20:37:56 03Laibsch 07org.oe.dev * r66ac81e5... 10/ (1 packages/libcap/libcap_1.10.bb): Dec 10 20:37:56 libcap: Make DEPENDS on flex-native and bison-native explicit. Closes 1659. Dec 10 20:37:56 This is being pulled in automatically for most DISTRO via gcc-cross Dec 10 20:37:56 and binutils-cross but not when these are assumed provided as for Dec 10 20:37:56 Sharp ROM for example. Dec 10 20:45:55 Laibsch: check _MTN/options Dec 10 20:46:07 koen: Thanks. Dec 10 20:52:02 is there some amd 64 users around ? Dec 10 20:59:08 me Dec 10 21:04:26 do you had troubles with glibc install ? Dec 10 21:09:08 What is the ${S} directory? Nothing on http://www.openembedded.org/directorystructure about it. Dec 10 21:09:37 Source dir Dec 10 21:09:40 check bitbake.conf :) Dec 10 21:13:39 Good night Dec 10 21:22:10 Is the line 'S = "${WORKDIR}/libcap-${PV}"' really needed for libcap_XY.bb? I guess not since this should be the default, right? Dec 10 21:24:35 Laibsch: looking at bitbake.conf this is indeed the default, so you can try without :) Dec 10 21:31:16 "morning" all Dec 10 21:31:25 hi rp Dec 10 21:31:32 'morning' RP Dec 10 21:31:39 hi. Dec 10 21:33:13 hi RP Dec 10 21:43:09 Can someone of the main devs give me the OK to push the patch for bug 1628? The only real change is removal of the definition of S which is redundant and addition of CFLAGS. The rest is just reordering. koen? hrw|gone? mickey|bbq? Dec 10 21:45:45 Anybody else? Comments? Dec 10 21:46:26 n8 Dec 10 21:58:14 tmbinc: re: this -lqte issue Dec 10 21:58:35 I can see that zaheer's build and mine have completely different Makefiles in build/tmp/work/dreambox-dvb-tools-v3-1.2-r0/dreambox-dvb-tools-v3-1.2 Dec 10 22:00:54 both Makefiles claim to have been generated by qmake (1.07a), so it seems to be getting the right one Dec 10 22:00:54 Laibsch: #1628 looks ok to me Dec 10 22:01:25 RP: Thanks for reporting back. Will push. Dec 10 22:03:37 tmbinc: http://noraisin.net/~jan/oe/ Dec 10 22:03:45 there are 2 makefiles in there, from zaheerm's build and mine Dec 10 22:03:53 do they tell you anything about where I might look? Dec 10 22:06:08 hmm.. for some reason qmake includes QT stuff Dec 10 22:08:28 could it be something from my env vars it's picking up, or installed pkgs? Dec 10 22:15:14 hm, anyone here knowing about qmake? Dec 10 22:15:37 we have the issue that qmake generates different makefiles on some host distros. Dec 10 22:17:28 check whether QMAKESPEC may be set in one of the env's Dec 10 22:17:33 if so, unset it before building Dec 10 22:18:11 it's not Dec 10 22:18:26 and I see an unset QMAKESPEC in the verbose debug output too Dec 10 22:20:41 add debug info Dec 10 22:20:52 change qmake.bbclass Dec 10 22:20:55 call qmake with -d -d -d Dec 10 22:20:57 (or something) Dec 10 22:21:01 then compare the output line by line Dec 10 22:21:41 03tmbinc 07org.oe.dreambox * r6ddf5a80... 10/ (6 files in 2 dirs): gdb: add gdb version suitable for nptl on mips Dec 10 22:21:53 03tmbinc 07org.oe.dreambox * r72aa69ee... 10/ (6 files in 3 dirs): python: add python-2.4.4 Dec 10 22:22:03 03tmbinc 07org.oe.dreambox * r6d842103... 10/ (1 conf/distro/opendreambox-1.4.conf conf/machine/dm7025.conf): opendreambox: use python 2.4.4, use gdb 6.5 (or 6.6.50.20061209 on dm7025) Dec 10 22:33:44 DEBUG 1: Project Parser: dmutils.pro:1 :TEMPLATE: :=: (subdirs) | DEBUG 1: Project Parser: dmutils.pro:1 :-e TEMPLATE: :=: (subdirs) Dec 10 22:33:50 first difference is -e in that line Dec 10 22:35:13 03Laibsch 07org.oe.dev * r361f5e74... 10/ (1 packages/libcap/libcap_1.10.bb): Dec 10 22:35:13 libcap: Append include directory to CFLAGS. Reenables compilation and fixes 1628. Dec 10 22:35:13 Seizing the opportunity to remove some cruft and beautify the bb. RP approved. Dec 10 22:37:50 tmbinc: what generates dmutils.pro ? Dec 10 22:39:36 mickey|bbq: what are you doing bbqing at 00:05??? Dec 10 22:39:38 ohhhhh Dec 10 22:39:41 echo -e "TEMPLATE=subdirs\nSUBDIRS=showiframe\n" >dmutils.pro Dec 10 22:39:59 XorA|gone: the same as you when you are 'gone' :D Dec 10 22:40:14 hope your weather is better than here Dec 10 22:40:21 sftmmab = suffering from too much meat at bbq Dec 10 22:40:32 mickeyl|sftmmab: :-D Dec 10 22:40:39 well... it's dark and 0 degree C Dec 10 22:40:50 gales and rain Dec 10 22:41:31 * RP listens to the wind causing the house to move gently :) Dec 10 22:41:44 ahahah Dec 10 22:41:51 tmbinc: it's because on ubuntu /bin/sh = ash Dec 10 22:41:56 and ash doesn't support echo -e Dec 10 22:41:58 ffs Dec 10 22:42:16 er, dash not ash Dec 10 22:42:22 ouch Dec 10 22:42:24 echo -e is a bash extension Dec 10 22:42:31 yeah and then it goes into that "heuristic" mode :) Dec 10 22:42:45 thanks for finding that out! Dec 10 22:42:57 now, how do I fix it? Dec 10 22:43:30 echo "TEMPLATE=subdirs" > dmutils.pro Dec 10 22:43:30 echo "SUBDIRS=${UTILS}" >> dmutils.pro Dec 10 22:43:31 ? :) Dec 10 22:43:52 what is -e doing anyway? Dec 10 22:44:13 'enable interpretation of backslashes' Dec 10 22:44:25 interesting Dec 10 22:44:30 we don't have any backslahses in there do weß Dec 10 22:44:34 so yank them Dec 10 22:44:41 tmbinc: well, I meant more permanently Dec 10 22:44:45 i know we have some for other .pro files Dec 10 22:44:48 as in, how to enforce oe's use of bash? Dec 10 22:45:02 it's my fault Dec 10 22:45:09 it could call /bin/echo explicitly Dec 10 22:45:09 the upstream source didn't had a makefile Dec 10 22:45:12 that should do it Dec 10 22:45:15 so i've generated it Dec 10 22:45:31 (i probably should not mention that i'm also responsible for the upstream source) Dec 10 22:45:42 hehe Dec 10 22:46:40 ok, testing using /bin/echo explicitly Dec 10 22:47:19 hm Dec 10 22:47:25 yep,that fixes it Dec 10 22:47:32 wouldn't removing the -e dependency a better thing? Dec 10 22:48:08 sure, but knowing you can do /bin/echo is a useful fact :) Dec 10 22:54:38 bah Dec 10 22:55:12 I finally get rid of that binary locale generation. Dec 10 22:55:44 I modified some glibc bb class to remove it. Dec 10 22:56:03 trough I think there is something wrong. Dec 10 22:56:31 those should not even have been enabled regarding comments. Dec 10 23:00:27 03xora 07org.oe.dev * rbb035df7... 10/ (6 files in 3 dirs): Dec 10 23:00:27 mplayer_0.0+1.0rc1.bb, mplayer_svn.bb : remove the -s from the install Dec 10 23:00:27 command in the Makefile. Should fix #1669 Dec 10 23:00:35 magnet: could you file that as a bug, please then, so that we can track it? Dec 10 23:03:03 sure. Dec 10 23:14:36 bug #1672 Dec 10 23:14:54 magnet: tnx Dec 10 23:15:46 tell me if you need more infos. Dec 10 23:16:51 magnet: tell me what you disabled in glibc.bbclass Dec 10 23:17:20 magnet: the variable is used in 1 place only: packages/glibc/glibc-package.bbclass:118: enabled = bb.data.getVar("ENABLE_BINARY_LOCALE_GENERATION", d, 1) Dec 10 23:17:54 magnet: I really need some sleep first. I may look this one over in the morning Dec 10 23:18:20 nite Dec 10 23:18:35 03Laibsch 07org.oe.dev * r3a8ccf83... 10/ (1 packages/libcap/libcap_1.10.bb): libcap: really remove that ${S} line now instead of commeting it out. Dec 10 23:20:04 nite Dec 11 01:30:53 03tmbinc 07org.oe.dreambox * r981c300b... 10/ (1 packages/dreambox/dreambox-dvb-tools-v3.bb): dreambox-dvb-tools-v3: remove bashism Dec 11 01:30:58 03tmbinc 07org.oe.dreambox * r48507009... 10/ (1 packages/python/python-2.4.4-manifest.inc): python-2.4.4: add manifest Dec 11 01:31:02 03tmbinc 07org.oe.dreambox * r58fdda5d... 10/ (3 files in 3 dirs): python-2.4.4: hack - add patch to display debug on gc fatal error **** ENDING LOGGING AT Mon Dec 11 02:59:57 2006