**** BEGIN LOGGING AT Sun Apr 29 02:59:56 2007 Apr 29 04:56:33 If any one is awake: obs-ncore is not liking INCLUDEPATH (I think) Error is here -> http://paste2.org/p/2827 Apr 29 06:52:38 who can tell me wether "PREFERRED_PROVIDER_qemu-native = "qemu-native"" specified packages in distro/xxx.conf would be processed? Apr 29 07:41:35 hi all Apr 29 08:26:31 03ifaistos 07org.oe.dev * r1927e10d... 10/ (3 files in 3 dirs): Apr 29 08:26:31 packages/asterisk/asterisk_1.2.18.bb: Enable speex support for asterisk Apr 29 08:26:31 If using a cpu with no fpu make sure that speex compiles with fix-point support Apr 29 09:43:17 03ifaistos 07org.oe.dev * r9abdbb77... 10/ (5 files in 3 dirs): Apr 29 09:43:17 site/powerpc-common: Add a common site fike for powerpc arch Apr 29 09:43:17 Still needs work, especially uclibc part Apr 29 10:38:06 > This can be implemented by patching the upstream libc, Apr 29 10:38:06 sorry, but: over my dead body. Apr 29 10:38:11 * koen hugs harald Apr 29 10:38:26 * koen stabs people thinking that shipping a non-standard libc is ok Apr 29 11:46:32 koen: Can you add the Machine field? Would be very handy for testing all the individual machines, now. Apr 29 11:46:50 and tracking which machines have which bug in angstrom. Apr 29 11:49:29 Is there a field in bugzilla where you can select more than one entry (check boxes or multi-select table) to make it possible to specify that more than one machine is affected? Apr 29 12:18:03 Hi! Apr 29 12:19:41 Laibsch: I'm not sure it's sustainably useful. Real machine's problems are really kernel problems, and should be reported upstream. The rest is results of tangled machine-specific patches and overrides, and those should be solved as global problem. Apr 29 12:21:40 morning timtimred Apr 29 12:25:20 psokolovsky: Not sure about that. Apr 29 12:26:00 There can always be machine specific things. Like "Gui too big" or such things. Apr 29 12:26:27 I am no expert bug it seems to me that those are not always kernel bugs. Apr 29 12:27:05 Laibsch: but those are *not* machine specific, that's the point. that's machine class-specific, like big for all qvga devices. and as OE strives to support different classes of machines, that's global OE issue. Apr 29 12:27:15 of course they Apr 29 12:27:16 are Apr 29 12:27:54 I take the view point of a user with a problem who wants to know about known problems for his machine Apr 29 12:28:10 He does not care about OE internals Apr 29 12:28:17 He should! :) Apr 29 12:28:26 So he is similar to me in that I might care but I don't understand. Apr 29 12:28:28 Laibsch: and I'm - developer's who's tired of device feudalism ;-) Apr 29 12:28:36 polyonymous: No, a user is a user Apr 29 12:29:01 psokolovsky: explain "device feudalism" Apr 29 12:29:11 Laibsch, to absolutely not care about OE internals he should keep out of bugzilla, anyway :) Apr 29 12:29:19 no Apr 29 12:29:26 we want user feedback Apr 29 12:29:31 Laibsch: so, there must be "raising awareness" actions ;-). and ping me when Debian or Ububtu bugtracker will provide fields for monitor size ;-) Apr 29 12:29:42 yes, but we want them to care at the same time, don't we? Apr 29 12:29:50 them! I *am* a user after all :) Apr 29 12:30:06 devs are almost always users Apr 29 12:30:12 this is not true the other way round Apr 29 12:30:18 But I'm not a dev. Not OE dev at least :) Apr 29 12:30:21 maybe yet :) Apr 29 12:30:24 psokolovsky: I still don't understand where you are getting at Apr 29 12:30:41 polyonymous: Let's not mince words Apr 29 12:30:48 Laibsch: people in embedded world get used to think that their own device is a special, and needs special treatment. wrong. all machines are the same, and all must be supported well. there should be as less machine-specific stuff, including "fixes", as possible. Apr 29 12:30:49 You are compiling with OE Apr 29 12:31:05 Let's take that as distinction between dev and user if you want Apr 29 12:32:36 psokolovsky: I am all for the angstrom-approach Apr 29 12:32:55 Laibsch: well, backing to start of discussion, having a specific field for machibe model is not really a necessity (and possibly even slightly evil). user submits bugs, devs categorize them. and "specific to device" should be very rare categorization in reality. Apr 29 12:33:06 Laibsch, well, you got a point here. Apr 29 12:34:52 psokolovsky: Let's say it this way. I do a lot of bug categorization and bug triaging. And it would help *me*. At least now. It might change in the future. Apr 29 12:35:48 Laibsch: yep, so I exactly hint about sustainability of that approach. anyway, my 2 cents. Apr 29 12:37:48 Sustainability should never be an issue. You can just ignore the field and be where we are now. Apr 29 12:38:05 But again, it would help me if it was available. Apr 29 12:38:41 s/should never be an issue/should not be an issue here/ Apr 29 12:40:40 Laibsch: but you will admit that you won't look into ipaq tickets, right? and I won't look into z. and that's bad. we just got unity of OZ and Ang, and now wnat to lose it again - in device feudalism, now pretty clearly defined, I guess. Apr 29 12:41:13 Laibsch: I'd consider approach you (and I) use - having "parent" ticket with dependencies - to be superior. ymmv. Apr 29 12:43:37 psokolovsky: yes, and that is exactly what this is useful for. Determine what you want to look and what you don't want to look at. Apr 29 12:44:06 The setting for MACHINE is of course not set in stone. If I realize this is not a collie specific problem, I mark it as such. Apr 29 12:44:22 You could talk about distro feudalism the same way Apr 29 12:44:29 And it exists. Apr 29 12:44:39 The key is fluidity - updating stuff. Apr 29 12:44:56 But I understand and respect your point now. Apr 29 12:46:07 ok, just hope that we'll get to better geberal PDA/handheld Linux support ;-) Apr 29 12:46:58 Laibsch: anyway, what machines you work with besides collie? Apr 29 12:52:10 I have a spitz Apr 29 12:52:13 again missing dependency... NOTE: Checking Package: linux-libc-headers-dev Apr 29 12:52:13 ERROR: Can not check RPATH, scanelf (part of pax-utils-native) not found Apr 29 12:52:39 Laibsch: ok. and another collie q: do you have any idea about collie & usbnet? Apr 29 12:52:42 psokolovsky: and I'd like to see support for the Linksys WRT but I think that is some time out Apr 29 12:53:40 Laibsch: yep, me too ;-) we really should try to bring openwrt kernel recipes into oe ;-) Apr 29 12:53:40 psokolovsky: Furthermore I intend to be an owner of an openmoko device eventually (but I might go the HTC route if that is feasible and suits me better hardware wise) Apr 29 12:53:55 psokolovsky: From what I understand the problem is the flasher Apr 29 12:54:12 psokolovsky: What do you want to know about usbnet? Apr 29 12:54:17 I have it working Apr 29 12:54:48 * Laibsch looks around to find more OE devices Apr 29 12:55:41 Laibsch: you have usbnet working with 2.6 collie? AFAIK, there's no modern gadget USB driver for SA. Apr 29 12:56:08 Laibsch: well, nevermind, I just wanted to know what Zs you work with, just in case ;-) Apr 29 12:56:17 * Laibsch finds none Apr 29 12:56:20 Laibsch: Lots of OE-devices w/o display. turbostation, storcenter, linkstations, kuroboxen ;-) Apr 29 12:56:23 psokolovsky: no, that was with OZ Apr 29 12:56:35 psokolovsky: I barely made a login into my Z about an hour ago Apr 29 12:56:46 my Z on angstrom Apr 29 12:56:57 * polyonymous envies Laibsch :) Apr 29 12:57:24 polyonymous: I have an already broken collie to spare ;-) Apr 29 12:57:47 If I had two devices I wouldn't suffer this kernel dilemma :) Apr 29 12:58:03 nothing to be envious of. No GUI for collie yet. And once there is, this collie won't have a working touchscreen I am afraid (it broke mysteriously) Apr 29 12:58:18 polyonymous: Can I suggest something else? Apr 29 12:58:25 This is what I always did Apr 29 12:58:30 You are still on OZ? Apr 29 12:58:40 Laibsch, yes, but on the one from .dev branch. Apr 29 12:58:44 Laibsch: ok, so if you'll be doing any investigations about usbnet, just note that h3600 has the same issues and could use any knowledge ;-) Apr 29 12:59:00 That is on the one which is no more... Apr 29 12:59:06 psokolovsky: I'll be in contact with you. And watch the BTS ;-) Apr 29 13:00:04 hmm... maybe I should just find out when oz went off of .dev and revert to the revision... Apr 29 13:01:20 polyonymous: I don't understand Apr 29 13:01:29 This is what I did. Apr 29 13:01:30 Laibsch, well, anyway, what was your suggestion? Apr 29 13:01:49 What you did was reverting to revision? Apr 29 13:02:48 polyonymous: Install an image to flash (and never care about it since). Use altboot to boot into a loop image on SD. Take out the SD card before doing anything dangerous. Thus I could flash the Z to my hearts content, reflash to original OZ and be sure to come back exactly the same environment. Apr 29 13:03:04 polyonymous: Doing that on OZ than doing it on angstrom, now. Apr 29 13:03:14 You'd still be able to test out all the latest goodies. Apr 29 13:03:53 polyonymous: Doing that on OZ than should be easier than doing it on angstrom, now. Apr 29 13:04:03 * Laibsch thinks faster than he types Apr 29 13:04:12 Which is a good thing (in a way) ;-) Apr 29 13:04:16 I don't understand... The problem I have is that I need incompatible kernel to run angstrom. How having a few root filesystems would help? Apr 29 13:04:51 polyonymous: You don't want to lose the ability to work with your spitz, right? Apr 29 13:05:17 Use altboot on OZ, do the daily work from an OZ loop image Apr 29 13:05:24 exactly. Ah, do you suggest that I just reflash kernel back and forth instead of kexecing it? Apr 29 13:05:50 Then you can flash the angstrom images all you want without worrying. I am not aware of bricking which would be the only risk Apr 29 13:06:14 polyonymous: Your last sentence sounds like what I am suggesting. Apr 29 13:06:28 Laibsch, got it. Maybe that's the way to go, indeed. Apr 29 13:06:49 Well, I'm doing a fresh rebuild of angstrom now - determined to attempt on fixing those QA fatal failures. Apr 29 13:07:47 Hi, anyone have experience with MIPS? Apr 29 13:08:25 NAiL: Is there a simple NAS device for 50-100$ (can be 2nd hand)? Further requirement: I want to attach a Fritz ISDN card to it (can be USB) and a parallel printer. Apr 29 13:08:31 NAiL: Never found such a thing. Apr 29 13:10:24 Laibsch: I have Asus WL-HDD dusting here, was $60 when I bought it a year ago. it has only 1 ethernet though, but can do nas of course ;-) Apr 29 13:14:32 USB port? Apr 29 13:14:41 OE device? Apr 29 13:16:14 Laibsch: yep, though usb1.1 iirc (and 1 port). no, not OE device. but openwrt one. mips Apr 29 13:17:41 Laibsch: you actually can treat it as mobile server companion for pda. youc annot put it into pocket, but would fit into a small bag, together with standalone accu. and builtin hdd will serve you (people around) via 802.11g lotsa stuff ;-) Apr 29 13:18:35 comes with lotsa stuff preinstalled? :) Apr 29 13:19:44 polyonymous: even if so, would you do anything to it execpt rm -rf ? ;-E. nope, sure, at $60, it doesn't include HDD, you put your own from the shelf, left from a laptop upgrade ;-) Apr 29 13:20:03 psokolovsky: Well, the portability is nice. I am looking to replace a LAN server, though. I wonder if the USB port would drive the Fritz ISDN card Apr 29 13:20:19 psokolovsky, of course. It was my feeble attempt on making a joke. Apr 29 13:21:22 I wish I had hdds from laptop upgrades... I have quite a few from deceased laptops. Apr 29 13:23:26 Laibsch: depends on driver availability of course. people attach usb-com dongles and then old phones to get backup GPRS channel even with default crappy software ;-) Apr 29 13:27:23 03pH5 07org.oe.dev * rcb611997... 10/ (1 packages/avahi/avahi.inc packages/avahi/avahi_0.6.18.bb): avahi: remove libavahi-ui (gtk+ dependency) as long as there are no users. Apr 29 13:29:37 03lenehan 07org.oe.documentation * r7d6083f8... 10/ (1 usermanual/chapters/recipes.xml): recipes: Clean up some of the spelling and grammar in the DEFAULT_PREFERENCE section. Apr 29 13:30:31 ~lart building custom kernels for Ubuntu Apr 29 13:30:31 * ibot turns building custom kernels into a lifesized tux doll for Ubuntu Apr 29 13:30:55 hello all Apr 29 14:12:01 mplayer packaging fails with an RPATH error because it calls the sdl-config from libsdl-native instead of the libsdl-x11 one during configure. any idea how to fix that? Apr 29 14:13:21 hmmm, util-linux doesn't like newer kernel-headers Apr 29 14:16:22 Laibsch : Misdn has issues with big endian machines Apr 29 14:16:39 Laibsch : working on that at the moment Apr 29 14:20:24 * Ifaistos thinks that he needs more time at the beach Apr 29 14:20:44 it was full today, like mid summer Apr 29 14:21:42 koen: how new? I just built it against 2.6.20 headers Apr 29 14:22:03 pH5: 2.6.20 Apr 29 14:23:12 umount.c:45: error: expected declaration specifiers or '...' before 'umount2'umount.c:45: error: expected declaration specifiers or '...' before 'path'umount.c:45: error: expected declaration specifiers or '...' before 'flags' Apr 29 14:23:19 that's the sympton Apr 29 14:23:21 m Apr 29 14:24:32 ouch Apr 29 14:25:03 I found some thread in a polish mailinglist, but I can't decode the solution Apr 29 14:25:06 (armv4t) Apr 29 14:25:07 I should stop forgetting to checkin changes. Cany you test http://en.pastebin.ca/464418 Apr 29 14:26:53 This patch is taken from the debian packages. Apr 29 14:29:12 pH5: works, could you commit that ? Apr 29 14:29:28 ok Apr 29 14:30:04 Can I add 'PATH =+ "/usr/local/arm/2.95.3/bin" ' to conf/distro/sharprom-compatible.conf and expect it to work? Any objections to doing so? I guess none of the core devs still compiles for Sharpm ROM, right? Apr 29 14:31:20 hi tmorning Apr 29 14:35:44 ltg.org seems to be down Apr 29 14:36:09 I have working kexec binary now :) Apr 29 14:36:42 Now I need to get to build OE so that it works. For now it's missing libgcc_s.so Apr 29 14:37:22 What is the "proper" way to get along with these rpath failures in gcc-cross, for instance? I'm not sure how to fix it... Apr 29 14:37:59 any kind word from devs? :) Apr 29 14:38:29 pat me on the head, tell me I'm a good boy :) Apr 29 14:40:15 polyonymous: one secret way is to find a place where it pukes and comment it out ;-E. hush, don't tell anyone! ;-) Apr 29 14:40:33 psokolovsky, Well, yesterday I did comment out one line in insane.conf Apr 29 14:40:44 polyonymous: Please document your work in the angstrom wiki Apr 29 14:40:45 Thought there are better ways :) Apr 29 14:40:46 Thank you Apr 29 14:41:09 polyonymous: Well, done Apr 29 14:41:23 * Laibsch pats polynonymous Apr 29 14:41:43 Laibsch, Not much to document yet. I do not feel like documenting what psokolovsky suggested, because it comes naturally into minds of the likes of us, anyway and isn't for general public. Apr 29 14:41:44 polyonymous: the right way is to have it fixed. it's that maybe not on your list for today ;-) Apr 29 14:42:16 no, such things may not be documented - indeed, they are secret ;-D Apr 29 14:42:33 psokolovsky, Well, I think the way to fix it is to make it omit -rpath when compiling libstdc++, but the .bb part of gcc is even more complicated for me than gcc itself :) Apr 29 14:43:57 03pH5 07org.oe.dev * ra6299c22... 10/ (3 files in 3 dirs): util-linux: fix build failure with newer kernel headers, patch from debian Apr 29 14:44:28 and yes, thanks, Laibsch :) Apr 29 14:45:48 * polyonymous starts fresh build with the secret measure in effect. Apr 29 14:47:42 pH5: thanks Apr 29 14:56:19 03koen 07org.oe.dev * r62ddcead... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom: prefer a newer cairo and update linux-libc-headers to 2.6.20 for HAL Apr 29 14:56:48 tinymail compilation fails with error: tny-gpe-password-dialog.h: No such file or directory Apr 29 15:14:18 hmm i think i found a snag when build different versions of uboot on a multi-machine setup. the mkimage binary stored in staging is always from the last u-boot build.... Apr 29 15:15:18 so if you build a board that needs uboot 1.1.2 and then one that needs a different version (i.e 1.1.4) mkimage gets overwriten Apr 29 15:15:35 * mwester thanks pH5 for fixing the troublesome util-linux / kernel headers bug! Apr 29 15:16:06 so if you go back to make a virtual/kernel for the fist board its using the wrong mkimage Apr 29 15:16:36 Ifaistos: uboot changed file-format between micro releases?!?!?!? Apr 29 15:17:54 i think there are differences but not sure how critical they are Apr 29 15:18:22 will check the changelog Apr 29 15:24:33 polyonymous: I think you should document it Apr 29 15:24:59 So that others like hvontres can incorporate your work Apr 29 15:25:08 Laibsch, commenting stuff in insane.conf ? Apr 29 15:25:13 Or what? Apr 29 15:25:13 Where you document it is up to you Apr 29 15:25:27 polyonymous: The stuff you did to get kexec to work Apr 29 15:25:48 Laibsch, mtn checkout older revision and bitbake kexec-tools? :) Apr 29 15:26:09 It's just that I need to build kexec against my kernel. Apr 29 15:26:15 needed Apr 29 15:26:30 polyonymous: Oh, it was just the breakages that kept you from it? Apr 29 15:26:42 It was kernel change, yes. Apr 29 15:26:57 And I had OZ built from .dev tree, where it is no more. Apr 29 15:29:01 hello all Apr 29 15:34:50 hey likewise Apr 29 15:35:11 hey koen Apr 29 15:36:02 koen: survived the race? Apr 29 15:36:13 likewise: yes Apr 29 15:36:17 I didn't run :) Apr 29 15:36:25 koen: from behind, or in front of the bar? Apr 29 15:36:56 drinking can be more hazardous then running, mind you :-) Apr 29 15:37:07 I did things like this: http://www.flickr.com/photos/koenkooi/476864511/ Apr 29 15:37:34 Koen: I cannot debug why this msg appears when I build angstrom-console-image for efika: NOTE: preferred version 2.5 of glibc not available (for item virtual/powerpc-angstrom-linux-libc-for-gcc) Apr 29 15:38:07 because glibc 2.ancient has a provider for being glibc-intermediate or something Apr 29 15:38:29 bitbake is considering to build glibc instead of glibc-intermediate Apr 29 15:38:36 so that message is safe to ignore Apr 29 15:38:39 koen: gesponsord tuinsetje? http://www.flickr.com/photos/koenkooi/476615863/in/photostream/ Apr 29 15:39:21 likewise: yes :), people could have their pictures taken at the rabobench Apr 29 15:39:45 likewise: that's nynke behind jelmer, btw Apr 29 15:41:36 koen: yes, seen on multiple pictures already, I had no doubt about that. Apr 29 15:41:58 who is jelmer? Apr 29 15:42:04 and what is the strlen status? Apr 29 15:42:13 zecke: the guy who plans most of my jobs Apr 29 15:42:23 zecke: lennert is working on the strlen thing Apr 29 15:43:04 OE the project that has someone who can talk to Lennert :) Apr 29 15:43:04 * likewise has grabbed his efika board to toy with. Apr 29 15:43:38 koen@bitbake:~/OE/monotone/org.openembedded.dev/conf/distro$ cat ~/OE/build/conf/auto.conf Apr 29 15:43:38 MACHINE=efika Apr 29 15:43:39 :D Apr 29 15:44:43 * koen is still annoyed with the rpath issues Apr 29 15:45:45 * likewise whistles Apr 29 15:45:45 :) Apr 29 15:46:37 I spent 8+ hours on them this last few days, but I cannot find the root of the problem in the packages. I've filed them though... Apr 29 15:47:18 In gnutls, I suspect some cross-over from a rpath-link configuration to a rpath variable. Apr 29 15:47:35 I could reproduce Crofton's report on libusb being insane. Apr 29 15:47:44 s/I could/I could *not*/ Apr 29 15:48:07 freetype was an easy fix (which I messed up anyway :-) ) Apr 29 15:48:40 likewise: start looking at .la files for bad signs :) Apr 29 15:49:01 likewise: e.g. if they have installed=yes this can explain bad rpath's in binaries Apr 29 15:49:05 ~seen pb_ Apr 29 15:49:16 zecke: I really need to learn auto[conf,make, ...} stuff first because I cannot tell head from tail. Apr 29 15:49:28 pb_ was last seen on IRC in channel #handhelds.org, 17d 22h 13m 55s ago, saying: 'bd2: hi'. Apr 29 15:49:48 likewise: well, libtool needs understanding, I don't know what it is doing Apr 29 15:50:01 likewise: I think this an element just to cause issues :} Apr 29 15:50:29 e.g. KDE kicked out libtool and it still works on Linux,BSD,Solaris,Windows :} Apr 29 15:51:16 zecke: please don't spead the KDE heresy ... ok cmake is a lot better than autocrap/libtool :) Apr 29 15:54:17 03ifaistos 07org.oe.dev * r4e27e897... 10/ (1 packages/linux/linux-magicbox_2.6.18.6.bb): packages/linux/linux-magicbox_2.6.18.6.bb : Fix a wraped-line creating an error during bb execution Apr 29 15:54:24 03ifaistos 07org.oe.dev * r2087be80... 10/ (3 files in 3 dirs): packages/uboot/u-boot_1.1.2.bb : Add suport for Magicbox boards Apr 29 15:54:57 Ifaistos: Hi. What are those magicbox boards anyway? I could only find a DVB-T tuner named such? Apr 29 15:56:15 likewise: They are an illusion... that's why the call them Magicbox :) Apr 29 15:56:17 http://www.magicbox.pl/ Apr 29 15:56:38 Everything from .pl is an illusion :-) Apr 29 15:57:05 likewise: I thought .pl was the illusion :) Apr 29 15:58:17 Xilinx stuff is next.... nighmare Apr 29 16:00:32 Ifaistos: what do you mean by Xilinx stuff PPC405 in the Virtex? Apr 29 16:00:54 likewise: yes Apr 29 16:01:00 koen: which rob is this? http://www.flickr.com/photos/koenkooi/466250844/ Apr 29 16:01:15 mcqueen Apr 29 16:01:18 aka robot101 Apr 29 16:01:44 koen: ok, seems a look-a-like to my brother :-) Apr 29 16:02:00 koen: the guy on the left, that is Apr 29 16:02:24 likewise: http://www.flickr.com/photos/qgil/465379116/ Apr 29 16:02:43 (bonus point for spotting a famous OE hacker in the back ;) ) Apr 29 16:02:59 * chouimat is listening to The Dark Eternal Night by Dream Theater on The Dark Eternal Night [Amarok] Apr 29 16:03:51 likewise: http://www.flickr.com/photos/koenkooi/466250844 now has subtitles :) Apr 29 16:04:29 koen: :-) Apr 29 16:04:49 great cookies at ELC, btw Apr 29 16:04:57 * likewise is listening to his girlfriend [NOT] Apr 29 16:05:34 likewise: you should ... because in 10 years she will remember this conversation ;) Apr 29 16:06:02 they also gave out digital recorders at ELC :) Apr 29 16:06:11 chouimat: there is no conversation. (uh oh, is that bad? :-) ) Apr 29 16:06:28 chuimate : You are *SO* right :) Apr 29 16:06:37 koen: audio, video? samsung, by any chance? Apr 29 16:06:49 4 minutes audio thingy Apr 29 16:07:01 the usualy $0.05 taiwanese stuff Apr 29 16:07:17 the sony camcorders were running linux, though Apr 29 16:07:28 koen: do you know if free-electrons or similar where video-ing the presentations? Apr 29 16:07:39 Tim Bird made some videos Apr 29 16:10:33 grrr Apr 29 16:10:43 my ppc603e toolchain is broken Apr 29 16:10:49 * koen stabs libgcc problems Apr 29 16:16:09 koen : 4.1.2 ? Apr 29 16:17:53 koen: that's probably an rpath problem. Apr 29 16:18:04 koen: did you build with insane.bbclass enabled? Apr 29 16:18:49 koen: libgcc*so* missing by any chance? (in cross, IIRC) Apr 29 16:33:55 cross/powerpc-angstrom-linux/lib/libgcc_s.so.1 Apr 29 16:34:00 I have my path munger enabled Apr 29 16:45:12 koen: fyi: without path munger but with insane enabled, my cross/powerpc-angstrom-linux/lib/libgcc_s.so.1 did not get installed, even though a subsequent re-run of bitbake gave no errors. Apr 29 17:15:31 hey all Apr 29 17:16:06 can somebody help me with the intel wcf2011beww wireless cf card? I don't know the chipset n'either I do know how to determine in it. Apr 29 17:35:29 likewise: every once in a while OE builds a toolchain that doesn't link stuff to libgcc or libstdc++ Apr 29 18:29:46 to set the size of the generated jffs image we use ROOTFS_IMAGE_SIZE or IMAGE_SIZE ? Apr 29 18:37:43 Ifaistos: let me check Apr 29 18:38:08 IMAGE_ROOTFS_SIZE Apr 29 18:38:15 Ifaistos: yes :-) Apr 29 18:38:39 i tried all permutation :) Apr 29 18:39:01 Ifaistos: and you could pad the fs w/: EXTRA_IMAGECMD_jffs2 = "--pad=0x600000 --big-endian --eraseblock=0x20000" Apr 29 18:40:07 setting a PREFERRED_PROVIDER_blabla will not make blabla a target, right? Apr 29 18:46:21 Those flickering dependencies seems to be pretty random. First image failed with ipkg-cl: command not found and I had to bitbake ipkg-native, now it fails fakeroot: command not found (and later ipkg-cl too) Apr 29 18:48:24 polyonymous: all come from base.bbclass Apr 29 18:48:41 zecke, anything I can help you with? Apr 29 18:48:55 polyonymous: yes, debug it Apr 29 18:49:18 pity, I'm not a .bb-king, yet. But I'll try. Apr 29 18:50:16 polyonymous: well, print bb.data.getVar('DEPENDS', d) at some spots in base.bbclass Apr 29 18:50:44 zecke, yup. First, I'll read it to get some idea of what is there to debug (never looked into it yet) Apr 29 18:51:12 the idea would be. for any bb file parsed Apr 29 18:51:25 ipkg-native fakeroot-native quilt-native gets added to DEPENDS Apr 29 18:51:50 aha... I'll go read the file while angstrom-x11-image is building. Apr 29 18:57:03 zecke, I see the logic there. Okay, I'll let you know when I try to dig it. Right now I'm waiting for bake to complete, but then I'll play around. Apr 29 19:02:06 Hello Apr 29 19:11:27 hey sirfred Apr 29 19:11:56 koen: Hey. I've found my performance issues with snes9x Apr 29 19:12:11 what was it? Apr 29 19:12:18 koen: bug 2154 Apr 29 19:13:03 koen: static uint8 Buf[MAX_BUFFER_SIZE]; seem to be not aligned with gcc-4.1 Apr 29 19:13:21 koen: I added a __attribute__((aligned(4)));, and everything started to work fine. Apr 29 19:13:51 I've opened that bug, with this fix and also a fix for some linkage annoyances I've found. Apr 29 19:23:10 COMBINED_FEATURES="alsa ext2 usbhost", but task-bluetooth selects bluez anyway? Apr 29 19:24:11 I have no idea what are you talking about, but why would you expect task-bluetooth not to select bluez? Apr 29 19:25:00 likewise: sounds like you want to roll your own image using the task-base subpackages Apr 29 19:25:23 koen: Thanks for committing the fix. :) Apr 29 19:25:34 koen: I'm stripping efika/angstrom a li'l bit. Apr 29 19:25:34 sirfred: thanks for fixing it :) Apr 29 19:26:08 03manuel.teira 07org.oe.dev * r2ead7798... 10/ (3 files in 3 dirs): snes9x: fix linkage and gcc4 problems, closes #2154 Apr 29 19:26:12 I have another x-file with scummvm. It has libmpeg2 in its depends, but the package depends on libmpeg2-0 Apr 29 19:26:25 And that's not an existing package. Any hint of what could be happening? Apr 29 19:26:33 sirfred: debian shlib renaming? Apr 29 19:26:47 koen: Don't know what is that. Apr 29 19:26:58 koen: How is that related with oe ? Apr 29 19:27:28 it renames library packages to their SO version, so you can have 2 versions installed in parallel Apr 29 19:27:45 sirfred: lots of OE distros do 'INHERIT += "debian"' Apr 29 19:27:51 I'm talking about .ipk and oe generated packages. How a debian ... Apr 29 19:27:56 koen: Even angstrom ? Apr 29 19:28:03 yes Apr 29 19:28:11 sirfred: see http://www.angstrom-distribution.org/repo/?action=details&pnm=libcairo2 for example Apr 29 19:28:18 koen: But there's some inconsistence then. Apr 29 19:28:31 libmpeg2 is generating a libbmpeg2 package. Apr 29 19:28:47 And scummvm depends on libmpeg2-0 Apr 29 19:29:34 But that's at runtime, I mean, the depends written in the ipk file. Apr 29 19:29:46 The DEPENDS says "libmpeg2" Apr 29 19:30:20 if I say DEPENDS = "cairo" the ipk will have Depends: libcairo2 Apr 29 19:30:52 Hmm Apr 29 19:31:06 So, libmpeg2 should perhaps emit a libmpeg2-0 package? Apr 29 19:31:16 I think so Apr 29 19:31:23 Let's see Apr 29 19:36:56 It seems there are libmpeg2 and libmpeg2-0 in angstrom: http://www.angstrom-distribution.org/repo/?name=libmpeg2&action=search Apr 29 19:45:59 koen: http://rafb.net/p/fvgio337.html Apr 29 20:08:57 03koen 07org.oe.dev * rf3e00b7c... 10/ (1 packages/libmpeg2/libmpeg2_0.4.0b.bb): Apr 29 20:08:57 libmpeg2: fix shlib packaging Apr 29 20:08:57 * OE can't add a proper RREPLACES, so people installing generated packages will proably have to a form of -force-overwrite Apr 29 20:09:44 koen: So, bitbake libmpeg2 should now generate a libmpeg2-0 package? Apr 29 20:09:51 yes Apr 29 20:09:59 koen: Great! Thanks. Apr 29 20:10:00 and libmpeg2-bin with the utils Apr 29 20:11:09 there, finished updating my irc client Apr 29 20:11:23 Hmm, bitbake contacts (o-hand pim) is building cairo-1.4.4. iirc there were some performance issues with that cairo version. Are they fixed? Apr 29 20:11:57 sirfred: see packages/cairo/cairo-1.4.4/cairo-workqueue.patch Apr 29 20:13:02 I understand the work was ported to 1.4.4 and so the answer is yes. Apr 29 20:13:27 they implemented it in a completely different way Apr 29 20:14:33 sirfred: http://lists.freedesktop.org/archives/cairo/2007-April/010381.html Apr 29 20:14:44 koen: Thanks for the info Apr 29 20:14:45 the new patch saves a lot of X traffic as well Apr 29 20:15:25 Some of the numbers are awesome Apr 29 20:17:05 indeed Apr 29 20:21:17 koen: Your fix for libmpeg2 worked fine. Thanks again. Apr 29 20:21:55 sirfred: the caveat in the commit message still applies, though Apr 29 20:23:28 koen: But scummvm already had a dependency on libmpeg2-0, so, no problem. Apr 29 20:24:46 Is there any recipe in oe for the poky sato gtk engine? Apr 29 20:24:57 not yet Apr 29 20:25:53 koen: ok. I will try Apr 29 20:26:20 D'oh Apr 29 20:26:28 * koen stabs mpeg2dec and libmpeg2 Apr 29 20:26:58 Hmm, scummvm complains about not libstdc++.so.6 Apr 29 20:27:09 * likewise hugs 2.6.21-cfs-v7 Apr 29 20:40:27 03koen 07org.oe.dev * r2cac8b9e... 10/ (1 packages/libmpeg2/libmpeg2_0.4.0b.bb): Apr 29 20:40:27 libmpeg2: generate the same packages as mpeg2dec Apr 29 20:40:27 * can we remove either mpeg2dec or libmpeg2, they seem a bit redundant..... Apr 29 20:45:40 03koen 07org.oe.dev * r0e85d248... 10/ (1 packages/scummvm/scummvm.inc): scummvm: point to the right path for the sdl config Apr 29 20:45:57 nite Apr 29 20:46:03 'night zecke Apr 29 20:46:33 koen: I remember to have added a bug for scummvm sdl-config related problem Apr 29 20:46:54 #2000 Apr 29 20:47:37 koen: Right. But my solution was just avoid using --sdl-config Apr 29 20:47:48 As it should be in the path for the configure program. Apr 29 20:48:49 Oh, I see your comments. Right. Apr 29 20:49:01 we want to use OE's libsdl-config, since you could have more sdl-configs in $PATH Apr 29 20:49:13 Yes, that's better Apr 29 20:49:50 I'm afraid my scummvm not finding libstdc++.so.6 is related with the same problem I had with libgcc_s.so.1 Apr 29 20:50:00 yes Apr 29 20:50:11 I don't know why, but gcc-cross didn't generate for me neither libgcc nor libstdc++ Apr 29 20:50:18 And so, no package is depending at runtime from those. Apr 29 20:50:23 tried rebuilding gcc-cross? Apr 29 20:50:30 koen: Yes, with no luck. Apr 29 20:50:36 I'm going to try again now. Apr 29 20:51:38 Curiously, I tried to run the install script directly, and it seemed to populate correctly at least the libgcc install directory Apr 29 21:23:34 Good night Apr 29 23:37:52 zecke? Apr 29 23:38:13 oh well, tomorrow then :) Apr 29 23:38:23 he isn't even here :) Apr 29 23:38:28 nite all **** ENDING LOGGING AT Mon Apr 30 02:59:56 2007