**** BEGIN LOGGING AT Sat Mar 17 02:59:57 2007 Mar 17 03:16:43 * cbrake really likes patchy git Mar 17 04:42:33 anybody try building angstrom with MACHINE=native and have problems building perl? Mar 17 05:42:26 03lenehan 07org.oe.dev * re0a22795... 10/ (3 files in 2 dirs): Mar 17 05:42:26 iproute2: Fix up 2.6.16 and 2.6.18 so that the patches are applied. Mar 17 05:42:26 During cleanup the SRC_URI+ was moved to prior to the require Mar 17 05:42:26 resulting in the patches not being added to the base SRC_URI. Use Mar 17 05:42:26 SRC_URI_append to prevent the order for being an issue. Mar 17 05:42:30 03lenehan 07org.oe.dev * rce3a8bd7... 10/ (7 files in 3 dirs): iproute2: Add 2.6.20 to pick up a lot of new commands and bug fixes. Mar 17 05:46:27 ~seen RP Mar 17 05:46:59 rp is currently on #gpe (2d 13m 34s) #oe (2d 13m 34s) #openmoko (2d 13m 34s) #openzaurus (2d 13m 34s) #handhelds.org (2d 13m 34s). Has said a total of 1 messages. Is idling for 20h 28m 37s, last said: 'morning all'. Mar 17 05:49:24 ~seen Laibsc1 Mar 17 05:49:38 laibsc1 is currently on #oe (3h 36m 43s) #openzaurus (3h 36m 43s), last said: 'What does the SECTIONS (not SECTION!) line in packages/opie-taskbar/opie-taskbar.inc do? Is it necessary?'. Mar 17 06:05:43 fun new voip toys for anyone interested... http://bugs.openembedded.org/show_bug.cgi?id=2011 Mar 17 06:18:24 hillct: You should probably use the volatiles stuff for the /var directories, so that it'll work for people like me where /var is a tmpfs filesystem. Mar 17 06:19:04 that's how it's setup Mar 17 06:19:35 volitile stuff is mostly voicemail which all goes into /var Mar 17 06:20:13 hillct: http://www.openembedded.org/user-manual&dpage=recipes_volatiles Mar 17 06:20:35 hillct: That's more what I'm talking about - I haven't finished that section of the manual yet though. Mar 17 06:21:36 So no /var files are installed as part of the package, they are created by install a volatiles file which has instructions on what to create. Mar 17 06:22:09 v8jlene: interesting Mar 17 06:22:37 not sure that's practical for this package though Mar 17 06:23:04 users can move their directories around in openpbx.conf if needed Mar 17 06:23:39 where do you recommend placing semi-volitile items such as voicemail that should survive a reboot? Mar 17 06:24:48 Yeah, that's a good question and one that I don't think there's a good answer for. Mar 17 06:25:38 hillct: I'm not sure about other devices, but on a sharp zaurus /var is created in ram as a tmpfs mount. so /var would not survive a reboot Mar 17 06:27:40 v8jlene: hvontres|home I think it's safe to say you wouldn't want to run this package on a device that's flash-only anyway. I'm sure that's true for other packages as well, but I can't imagine running a PBX on a Sharp Zaurus for example Mar 17 06:27:59 On my distro I bind mount stable storage onto var for the one thing that I need stable storage for. Mar 17 06:28:50 hillct: your right, but I am not sure what other distros, like nslu do for /var Mar 17 06:29:01 well, the file system standards say that /var is persistent Mar 17 06:29:29 v8jlene: Is the populate-volitiles script integrated into every distro? I don't yet see it in SlugOS. When it is added, where will it expect to find the volatiles files? Mar 17 06:29:35 in SlugOS, when you boot from disk, the /var is persistent. when you boot without a disk attached, then it's in ram. Mar 17 06:30:17 populate-volatiles is in SlugOS. Mar 17 06:30:31 I think it's fair to make some assumptions about the robustness of the installation if your'e even contemplating making your NSLU2 into a PBX Mar 17 06:30:38 oh? Mar 17 06:30:42 will check it out Mar 17 06:30:48 it's in /etc/init.d Mar 17 06:30:56 hillct: look at one of the packages that currently use it, such as p3scan or postfix Mar 17 06:31:08 K Mar 17 06:31:21 the FHS says /var is persistant, so you should assume so, and any distro where it is not persistent has to do other stuff to save what needs to be saved. Mar 17 06:32:10 hillct: I currently run asterisk on some of my devices and I have a USB2 storage (usually flash keys) bind mounted into the appropriate directories in /var. Mar 17 06:33:03 it may be valiable to have a mechanism whereby a package post script can collect metadata about where it's being installed so it can make any needed adjustments to target directories based on things like persistance of filesystems Mar 17 06:33:35 v8jlene: yes, and the same applies for openpbx Mar 17 06:34:05 v8jlene: anywhere you can run Asterisk, you should be able to run OpenPBX Mar 17 06:34:48 there's some additional stripping out of dynamic modules that could be done as well, to accomodate even smaller memory devices Mar 17 06:37:19 Heh, I don't have any really small devices - my smallest device has 128M RAM and 64MB flash and some internal and external USB2 ports... so I'm a lot better off than a lot of others ;) Mar 17 06:38:42 as far as I can tell, the only thing in var that really should be non-volitile is /var/lib/* Mar 17 06:38:56 the rest are logs and pid files and such Mar 17 06:41:24 Yeah, I might need to write something up and post to the mailing list to get some clear guidelines on this sorted out. Mar 17 06:43:31 hmm: libol staging is broken Mar 17 06:44:09 also, how should order of execution of volitiles creation work. Onbiously you want 00_core to be first, thus the name, but should each package just try to be next? 01_openpbx ? Mar 17 06:44:19 "install -m 0755 /home/slug/slugosle/tmp/work/armv5te-linux/libol-0.3.18-r5/libol-0.3.18/l\ Mar 17 06:44:20 ibol-config /home/slug/slugosle/tmp/staging/i686-linux/bin/arm-linux" does not look right to me Mar 17 06:44:24 that seems like it could get messy and would be unreliable anyway Mar 17 06:45:00 hillct: Your recipe looks mostly fine apart form that. Might pay to run contrib/oe-stylize.py on it and look at the things it complains about (tabs for example) and the ordering it suggests for things (to try and keep things the same). I normally don't use the output directly. Mar 17 06:45:55 oh, geez. didn't we just have a discussion of that earlier today ;) Mar 17 06:46:13 hillct: Because you'd only depend on things created by core, you can use anything after that. You don't depend on volatiles from other packages so it doesn't matter about what's there. Mar 17 06:46:31 yeah. makes sense Mar 17 06:47:14 hillct: Maybe... I didn't say to use the output, just note what it complains about, and the ordering of some things (ie, I look at the header and go "DEPENDS belongs a few lines down") Mar 17 06:47:22 well, if you depend on the other package you might, depend on it's volitiles, although I can't think of a good example of that at the moment Mar 17 06:47:46 K Mar 17 06:48:01 hillct: p3scan depends on the clamav virus definitions for example... but yeah, that's unusual. Mar 17 06:48:26 looks like lots of packages are installing things into STAGING_BINDIR_CROSS, but that directory is not created by default, and the first package that tries to install something into it creates a file with that name instead of put a file in a directory of that name. Mar 17 06:48:33 I'm putting together a list for the next update. volitiles, a log rotation script, recipe formatting, I'd like to get a sounds sub-package in there maybe Mar 17 06:50:23 gtk+, libnet, libol are all broken in that fashion Mar 17 06:50:56 rwhitby: It's created ok here. If it leaves up to whoever install there first fails to create the directory then you'd have that issue. Mar 17 06:51:19 yeah, it depends which recipe you run first. Mar 17 06:55:32 it's caused some weird stuff to not build here. including openpbx. Mar 17 06:58:52 looks like there is the same problem with STAGING_BINDIR_NATIVE Mar 17 07:02:27 with uboot, uboot-utils, ftdi-eeprom-native, tzcode-native, and dfu-util-native. Mar 17 07:02:49 yeah, none of them are explicitly created by the looks. The binconfig class creates STAGING_BINDIR_CROSS which makes that a bit less visible. Mar 17 07:03:04 yeah, but only if you inherit binconfig Mar 17 07:03:09 I'm fixing them all right now Mar 17 07:04:09 Yep, just happens for me to be one of the first packages built does that... Mar 17 07:05:02 it's an insidious failure, cause it means that pkgconfig files don't get installed for all subsequent packages, and you get weird dependency failures in configure on other packages. Mar 17 07:07:27 hillct: it was before 6.00am my time when you had that talk about oe-stylize on here ;) No wonder I didn't see it! Anyway, I've never had any problems with using the output from it directly for recipes that don't require/include other ones and it's rather handy for keeping things in the same order in recipes. But anyway it's not really important.. Mar 17 07:07:44 03rwhitby 07org.oe.dev * rbcad31fc... 10/ (3 files in 3 dirs): libol, gtk+-1.2: create the STAGING_BINDIR_CROSS directory before installing stuff into it. Mar 17 07:07:49 03rwhitby 07org.oe.dev * r5dae50ef... 10/ (6 files in 6 dirs): dfu-util-native, ftdi-eeprom-native, tzcode-native, u-boot, uboot-utils: create the STAGING_BINDIR_NATIVE directory before installing stuff into it. Mar 17 07:08:18 * rwhitby now blows away all his tmp dirs, cause he doesn't know how insidious the effect of those bugs were ... Mar 17 07:09:06 v8jlene: oe-stylize comes in handy as a reference but taking it as gospel can be dangerous Mar 17 07:09:37 heh - as with most things, including taking the gospel as gospel :-) Mar 17 07:09:40 anyway, I'll be making those adjustments along with a few other changes over the next week or so Mar 17 07:09:52 rwhitby: true Mar 17 07:11:35 well, I'm off to a wedding reception now, so my laptop will have plenty of time to rebuild everything overnight ... Mar 17 07:12:16 Enjoy. Mar 17 07:12:24 * v8jlene thinks it's about beer time Mar 17 08:08:05 morning all Mar 17 08:13:41 good morning Mar 17 08:14:27 morning all Mar 17 08:23:29 hi pH5. Here is a patch to make goffice build. It seems that you succeed to build goffice_0.3.2 but used DEFAULT_PREFERENCE = "-1" for some reasons. but goffice_0.2.1 does not build because of QA issue (missing FILES_${PN}-dbg). Can you please fix it ? Mar 17 08:23:47 pH5: http://rafb.net/p/2Ax4LX78.html Mar 17 08:30:13 cyrilRomain: I just fixed the QA issue in 0.2.1, copied the line to 0.3.2 (without changing the version, doh) and then pushed 0.3.2 instead of 0.2.1 :) Mar 17 08:30:20 Thanks for the patch, I'll apply it in a second Mar 17 08:30:48 pH5: thank you :) Mar 17 08:30:49 The DEFAULT_PREFERENCE=-1 is by koen|away, he wrote in the changelog that there are some api issues. Mar 17 08:31:03 pH5: thanks for the info Mar 17 08:32:02 pH5: instead of using the version, why not useing ${PV} ? ;) Mar 17 08:40:04 cyrilRomain: yeah, here you go. Mar 17 08:40:22 hey koen|recife_ Mar 17 08:41:45 03cyril 07org.oe.dev * r5c577a49... 10/ (3 files in 2 dirs): goffice: make it build (fix QA issues for real). Mar 17 09:52:59 hi all Mar 17 10:02:49 have anybody here use linksys wifi wcf12 card ? Mar 17 10:11:35 Good morning Mar 17 10:12:54 mrdata: hi Mar 17 10:13:10 mr_nice: hi Mar 17 10:14:32 mr_nice: you will transfer /proc/cs3 to sysfs, is this right? Mar 17 10:15:36 mrdata: I will try to do so Mar 17 10:16:17 mr_nice: need you help or will you do it alone Mar 17 10:16:48 mrdata: help is allways welcome. I am not that c professional Mar 17 10:18:13 mr_nice: i also not professional, so we must read a lot of documentation first Mar 17 10:18:34 mrdata: yes Mar 17 10:18:51 mrdata: I think it would be very cool to get apm working. but I have no clue how to start with it Mar 17 10:19:44 mr_nice: okay, i have also no plan was going wrong with apm Mar 17 10:20:07 mr_nice: but, i will try to understand this Mar 17 10:21:18 mr_nice: you could try the /proc/cs3 stuff and i will fight against apm, is this okay? Mar 17 10:21:35 mrdata: ok Mar 17 10:22:02 mr_nice: then let us do the work Mar 17 10:22:27 mrdata: :) Mar 17 10:23:13 mr_nice: is your mq200 stuff right compiled now ? Mar 17 10:23:48 mrdata: yes, it compile and works fine! Mar 17 10:25:28 mrdata: I have hand it in as an bugreport (1971) one bugs.openembedded.org so if some of the devs have time it will be included in oe. also I have added it to the slackpad svn repro. Mar 17 10:29:04 mrdata: the original mq200 driver from zecke included also a simpad_bl.c and a simpad_lcd.c file which both didn't compiled. maybee you can look over it? Mar 17 10:29:48 mrdata: I think the _bl is for backlight control Mar 17 10:30:12 mr_nice: where can i find the original driver set from zecke? Mar 17 10:31:13 mrdata: I can send you the files. else you can get his git tree but this is overkill I think Mar 17 10:31:24 mrdata: I send you an email Mar 17 10:31:36 mr_nice: thx! Mar 17 10:34:34 mrdata: mail is sent Mar 17 10:37:36 mr_nice: your mail is arived Mar 17 11:14:10 mrdata: hm, it seems to be a good idea to have the battery state also done via sysfs. Mar 17 12:00:33 mickey_away: it seems that python-2.4.4-manifest.inc need to be regenerated by generate-manifest.py. I've had QA issues http://rafb.net/p/erArqx12.html that disappeared with a renegerated manifest (python-misc is indeed not here anymore) Mar 17 12:05:27 CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include/libffi Mar 17 12:06:16 I have this nice error, apparently this kind of error happend a few times in the past, I see bugzilla entries Mar 17 12:06:46 stuff like that http://bugs.openembedded.org/show_bug.cgi?id=1826 how could I see how to fix it? Mar 17 12:08:05 keesj: you have to fix the package configure options, or fix autoconf files if it cannot be fixed with configure option Mar 17 12:08:26 keesj: on which package do you have such error ? Mar 17 12:09:12 gcc itself I think oe/tmp/work/armv4t-linux/gcc-cross-initial-4.1.1-r11/gcc-4.1.1/ Mar 17 12:09:52 gcc ? ouch ! Mar 17 12:10:28 I am really and oe n3ebie. chances are bigger I have a bug :p Mar 17 12:12:02 * cyrilRomain is quite surprised there is such a bug in gcc... and never encoutered it in gcc Mar 17 12:12:27 this is the "full" log http://paste-it.net/1639/raw/ Mar 17 12:13:52 I have set MACHINE to smdk2440 and the DISTRO to generic Mar 17 12:14:27 I do run bitbake as normal user and mu install is in /home/keesj/oe/ Mar 17 12:14:36 03nail 07org.oe.dev * r4dcbd44e... 10/ (1 packages/images/foonas-image.bb): foonas: Update image support Mar 17 12:17:16 keesj: have you also set TARGET_OS, and more generally what should be set in conf/local.conf ? Mar 17 12:19:57 keesj: hmm, I have no clue on what is wrong :-/ Mar 17 12:28:01 cyrilRomain: thanks for the help Mar 17 12:28:10 keesj: do you have recent metadata ? Mar 17 12:29:52 keesj: at least I can't find any related bug. Please check your config, and file a bug if you cannot get ride of the error Mar 17 12:31:45 I have found a possible problem (disk almost full), I will free some spavce and try again Mar 17 12:35:53 apaulsen, slapin: I still have trouble to reproduce yesterday's error reliably. If you can provide me with some config and command as a test-case I'd appreciate it. Mar 17 12:37:06 For example, I use "which glibc" in bitbake shell. But I always get glibc version 2.5 as a response. When I actually build glibc over night yesterday, it appears that indeed 2.3.5 was chosen. Mar 17 12:37:25 I am looking for a faster method to reproduce this error. Mar 17 13:00:27 nope it was no a problem with disk space Mar 17 13:01:19 hey koen|recife_ Mar 17 13:02:15 03pfalcon 07org.oe.dev * rb9d87867... 10/ (3 files in 2 dirs): xserver-common 1.14, 1.15: Apply calibrate-only-if-ts.patch (from 1.13). Mar 17 13:08:01 03marek.vasut 07org.oe.dev * re6d5b26a... 10/ (11 files in 3 dirs): Mar 17 13:08:01 libsdl-qpe 1.2.9: Add version of libsdl for OPIE, matches versions of other libsdl Mar 17 13:08:01 implementations in OE. Mar 17 13:24:04 03pfalcon 07org.oe.dev * r073563c6... 10/ (1 packages/libsdl/libsdl-qpe_1.2.9.bb): (log message trimmed) Mar 17 13:24:04 libsdl-qpe 1.2.9: Make sure OPIE package has distinct name from other libsdl packages. Mar 17 13:24:04 * The fact that all libsdl variant packages have the same package name, is Mar 17 13:24:04 a major issue we have, causing various issues with iser support. Mar 17 13:24:04 * It is caused by Debian package renaming, which ignores real package names Mar 17 13:24:07 and names them instead based on formal rules of pressing package's files, which Mar 17 13:24:09 has issue like this in general. Mar 17 13:28:24 koen|away: can you please push this monotone do_install fix http://rafb.net/p/7xswn529.html ? Thanks in advance :) Mar 17 13:49:18 ~seen mvo Mar 17 13:49:46 mvo <~egon@ip161.168.1211H-CUD12K-02.ish.de> was last seen on IRC in channel #debianppc, 1101d 22h 56m 59s ago, saying: 'Bor_Ed: and I can get the deivce listing of OF via /proc/device-tree?'. Mar 17 13:54:09 'morning Mar 17 13:57:59 hi koen|recife_ Mar 17 13:59:35 Humm, I've tried twice to install a gpe-image on my husky, and the two times, I only get a lot of jffs2_scan_eraseblock() errors. Mar 17 13:59:52 Something about Magic bitmask 0x1985 not found at ..... Mar 17 14:00:27 I expect that the problem is the jffs2 image I've generated. Could the flash be damaged? Mar 17 14:03:04 psokolovsky: you broke libsdl runtime deps for everyone, please revert that cset Mar 17 14:04:02 koen|recife_, what you mean? I didn't touched anything except libsdl-qpe Mar 17 14:04:35 someone not using it will not notice any effects Mar 17 14:05:31 Hi. I'm wondering: How does bitbake (or ipkg for that matter) compare versions? Especially in the uboot-gta01-1.2.0+svn${SRCDATE} case. If I put a number in SRCDATE, will it do the right thing? (E.g. "uboot-gta01-1.2.0+svn10" is more than "uboot-gta01-1.2.0+svn2") Mar 17 14:08:01 03svolpe 07org.oe.dev * r234f2f3a... 10/ (4 files in 3 dirs): Directfb 1.0.0: Add 1.0.0 and delete RC3, update pkgconifg patch to fix all .pc.in files Mar 17 14:11:04 koen|recife_, so, it's OPIE-local, the reason why it was done that way is exctly your concern that it is easy for users to confuse different libsdl variants, which causes headache for OE mainatiners. Mar 17 14:11:52 koen|recife_, I dicussed the issue with hrw, and he didn't support idea to extend it to other libsdl immediately Mar 17 14:12:40 koen|recife_, so, I decided to test in on local OPIE level, it works, I fully sign off of it, and ready to handle issues connected with it Mar 17 14:17:16 psokolovsky: it breaks every non-opie application Mar 17 14:17:23 psokolovsky: so revert it ASAP Mar 17 14:18:08 koen|recife_, how can I reproduce that? I'm running with it for 5 days w/o issues. Mar 17 14:19:53 every sdl app you build after libsdl-qpe depends on libsdl-qpe instead of libsdl12 Mar 17 14:24:03 disapprove pushed Mar 17 14:24:10 03pfalcon 07org.oe.dev * r98be4107... 10/ (1 packages/libsdl/libsdl-qpe_1.2.9.bb): disapproval of revision '073563c6fb1aad528e3d0c4f974eea7a71e1ad87' Mar 17 14:24:21 koen|recife_, so, I didn't lie that it affcects *only* opie users, right? Mar 17 14:24:35 libtool-native-1.5.10 has stopped working for me, configure complains that it is already configured Mar 17 14:24:43 no, it affects everyone without PREFERRED_PROVIDER_libsdl set Mar 17 14:24:45 koen|recife_, what do your recommend as solution? Mar 17 14:24:58 psokolovsky: keep PV and PR for all libsdls in sync Mar 17 14:25:01 koen|recife_, we have that set to x11, right? Mar 17 14:25:47 koen|recife_, but I want to solve that "different things named the same way" problem Mar 17 14:26:07 koen|recife_, I see not staging qpe as good solution, what do you think? Mar 17 14:26:15 no Mar 17 14:26:22 the 'problem' is in the shlibs code Mar 17 14:26:54 koen|recife_, let's use x11 only for staging Mar 17 14:27:05 that will make no difference at all Mar 17 14:27:14 or yet better, make generic libsdl package for staging Mar 17 14:27:23 ok, I will look closer into this issue Mar 17 14:27:28 again, staging is not the problem Mar 17 14:28:14 it doesn't matter which libsdl you build against, it matters what ends up in ipkg's Depends: field Mar 17 14:28:59 koen|recife_, well, I want it to stay libsdl-1.2-0 still, just that dependncy be provided by package with different name Mar 17 14:30:19 To answer my own question: there is an ipkg-compare-versions which I can use to experimentally find that out Mar 17 14:31:07 Henryk, be wary that there're bugs here and there in ipkg stuff Mar 17 14:31:36 psokolovsky: I see Mar 17 14:41:12 Hello. Anybody installed gpe-image from openzaurus-unstable lately? Mar 17 14:48:08 The problem is that I'm able to mount using mtdram the generated gpe-image, but after flashing it to my husky, I only get: Mar 17 14:48:09 jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00000008 Mar 17 14:48:19 Could be a problem of my internal flash? Mar 17 15:06:20 re Mar 17 15:10:26 Well, I'm going to try with a NAND restore Mar 17 15:19:42 I'm afraid the jffs2 generated for gpe-image is not good Mar 17 15:20:17 After a NAND restore (original Sharp ROM working), I've flashed again with this gpe-image and again the same errors. Mar 17 16:32:23 is there a policy or guidelines with regard ot using arbitrary variables in recipes? Mar 17 16:33:18 for example, I want to add te variables 'instuser' and 'instgroup' so I can more reasonably manage file ownership Mar 17 16:34:36 If I allow for this file ownership flexibility, I'd also want to dynamically create te volitiles file for my package, rather than including a static one Mar 17 16:40:33 hillct, you can use them, if it makes sense. just follow conventions. INSTUSER. (or maybe there's already standard var name for that) Mar 17 16:41:40 as far as I can tell, there isn't, but I've actualy reconsidered and I'm sticking with the static file for volitiles Mar 17 17:04:33 03coredump2 07org.oe.dev * rf2a15f2f... 10/ (3 files in 3 dirs): webcam-server: Add webcam-server Mar 17 17:04:37 03coredump2 07org.oe.dev * r321bc832... 10/ (8 files in 5 dirs): netbase: Add support for custom if-*.d directories. Also fix failing dhcp mode on SlugOS due to slow auto-negotiation (dhcp fix courtesy mwester & NAiL, thanks!) Mar 17 18:03:40 03Laibsch 07org.oe.dev * r34ccdf81... 10/ (1 packages/qof/qof_0.7.2.bb): qof: beef up the bb file with some further entries Mar 17 18:03:44 03Laibsch 07org.oe.dev * r13300dbb... 10/ (1 packages/gpe-expenses/gpe-expenses_0.0.6.bb): Mar 17 18:03:44 gpe-expenses: fix dependencies, closes 2010. Mar 17 18:03:44 Beef-up bb file with some further entries Mar 17 19:03:49 03nail 07org.oe.dev * r6389cf32... 10/ (4 files in 3 dirs): linkstation: Add support for linkstation-devices. Thanks to timtimred Mar 17 19:03:52 03nail 07org.oe.dev * r687185a0... 10/ (8 files in 3 dirs): uboot: Add support for linkstation target Mar 17 19:03:56 03nail 07org.oe.dev * r408f7747... 10/ (4 files in 3 dirs): linux: Add linkstation kernel Mar 17 19:04:04 * timtimred wiggles Mar 17 19:04:21 thanks NAiL Mar 17 19:10:27 np Mar 17 19:21:20 evening all Mar 17 19:32:04 mr_nice: email sent to you Mar 17 19:52:15 mrdata: thx ! Mar 17 19:54:38 mr_nice: no problem, but the problem with .config file is not linked with packages_dir stuff Mar 17 19:58:53 mr_nice: i delete only in ../tmp/stamps/simpad-angstrom-linux/ the files after linux-2-6.20-r3.do_qa_configure Mar 17 19:59:56 mr_nice: then rebuild kernel stuff begin with do_compile and .config should not touched there Mar 17 20:12:44 mrdata: best way is to just replace the defconfig file in packages/linux/linux/simpad/defconfig Mar 17 20:17:05 mr_nice: prepare packages dir and will test it now Mar 17 20:20:46 v8jlene: what's the volatiles cache as distinct from the volatiles listed in /etc/default/volatiles ? Mar 17 20:45:00 mr_nice: its funny, when i insert CONFIG_BACKLIGHT_SIMPAD=y and CONFIG_LCD_SIMPAD=y in defconfig Mar 17 20:46:29 mr_nice: and start a clean new compile, then in .config my new 2 lines are removed and Mar 17 20:47:18 mr_nice: simpad_bl.c + simpad_lcd.c would not compile Mar 17 20:50:43 is anyone using packaged-staging from scratch? it doesn't seem to be staging the kernel headers for me Mar 17 20:52:12 Hi All, Mar 17 20:54:35 mrdata: hm, you could try bitbake -c rebuild virtual/kernel but i am not that familiar with oe Mar 17 20:56:56 I want to compile libgnomeui but it is not successful, as it depends on gtk+-2.8.9 however configure fails since it needs gtk+ > 2.9.0 Mar 17 20:58:26 I compiled gtk+-2.10.x. Can I force libgnomeui to use gtk+2.10.x somehow? Mar 17 20:59:19 Anyone know why thttpd's "htpasswd" binary is statically linked? It takes 410kb. thttpd itself takes 84kb... Mar 17 20:59:51 LOL Mar 17 20:59:57 no idea Mar 17 21:00:43 while I'm thinking about it, there appears to be a problem with linking against libpcre Mar 17 21:02:05 when I tried to build lighttpd it worked but was unable to use libpcre so URL rewriting was not functional. I recall seeing discussion of problems on other OE distros where libpcre was being built was not dynamically linkable Mar 17 21:03:43 hi Mar 17 21:24:32 mr_nice: problem solved, i must declare the two new parameters also in a Kconfig file Mar 17 21:24:47 03coredump2 07org.oe.dev * r9ac358a3... 10/ (5 files in 2 dirs): netbase: Add forgotten directories. Mar 17 21:26:21 mr_nice: -> /drivers/video/backlight/Kconfig and tata all will compile Mar 17 21:31:31 mrdata: :) Mar 17 21:32:05 mrdata: you can also do a obj-y in the Makefile but creating a Kconfig entry is better Mar 17 21:33:34 mr_nice: with new parameter only in a Makefile no success Mar 17 21:34:09 mr_nice: and obj-y is realy hardcore Mar 17 21:34:41 mr_nice: but thx for your help Mar 17 21:38:19 the $(CONFIG_XX) in the gets y m or n depend on the state of the Kconfig entry (e.g tristate) Mar 17 21:38:48 mrdata: thx for your work Mar 17 21:38:57 mr_nice: in the cs3-sysctl-patch on sf is a entry in the Makefile for sysctl.o missing Mar 17 21:39:13 mr_nice: so it would not comiled Mar 17 21:39:34 mr_nice: be compiled Mar 17 21:39:52 mrdata: I will change it Mar 17 21:40:40 mrdata: thx for hinting on it. the sysfs is quite cool. I hope I will be able to complete it tomorrow Mar 17 21:42:21 03nail 07org.oe.dev * r079169fa... 10/ (3 files in 3 dirs): thttpd: Build htpasswd shared too. Reduces the total size from 600kb to 110kb Mar 17 21:43:05 mr_nice: fine, i must read the linux device driver book 3th edition first Mar 17 21:43:52 mr_nice: i hope, that i will understand sysfs then Mar 17 21:44:54 mr_nice: and the for my mmc driver i must look for usage of udev Mar 17 21:46:36 mr_nice: tomorrow i will try to understand the suspend problem we have Mar 17 21:46:58 mrdata: :) Mar 17 21:48:30 mrdata: I have to read my C book some struct stuff the kernel drivers are using still puzzles me. Mar 17 21:50:43 mrdata: but i will try to do my best to help getting the kernel complete Mar 17 21:52:03 mr_nice: ;-) , same by me and sometime (in the near future i hope ...) we will complete it Mar 17 21:53:15 mrdata: hehe, yes Mar 17 21:55:14 mr_nice: i need some sleep now and wish you a good night, tomorrow will be a long day again Mar 17 21:57:19 good night all Mar 17 21:59:06 mrdata: good night Mar 17 22:03:56 is busybox built for all distros with the same applets ? Mar 17 22:05:01 every time packaged-staging fails I sigh...it was such a good idea.... Mar 17 22:05:25 in that we can assume busybox is available, can we also assume that it has all te same applets (beyond the defaults)? In my case, I'm talking about gzip, but I imagine the same could apply for any of the optional applets like run-parts, etc Mar 17 22:05:37 JustinP: what error? Mar 17 22:06:14 hillct: actually some (most?) distros use their own BB config Mar 17 22:06:23 oh Mar 17 22:06:25 wonderful Mar 17 22:07:00 with that in mind, is there any method for conditional RDEPENDS ? Mar 17 22:07:35 of course. With some serious python-foo that should be doable.... Mar 17 22:08:09 I mean, I can test for the existance of /bin/gzip, but can I then update RDEPENDS based on the result, from within my recipe? Mar 17 22:09:01 it wouldn't seem to take great python-foo, depending on exactly how RDEPENDS are handled within a built package Mar 17 22:09:16 hillct: yes, with a python function inside the reciepe that is doable. You can manuipulate very large parts of BB/OO via python functions Mar 17 22:09:56 my python-foo is not strong :( Mar 17 22:09:56 hillct: see my comment on http://bugs.openembedded.org/show_bug.cgi?id=354 Mar 17 22:10:35 03nail 07org.oe.dev * r642d13b0... 10/ (1 conf/distro/foonas.conf): foonas: Add kernel modules to the linkstation depends, remove _sparc defines Mar 17 22:10:41 I was hoping to focus on Ruby next, but I may have to go with python :( Mar 17 22:15:52 JustinP: unless I misunderstand your error, it's not using headers from staging at all but in fact using them from your build system Mar 17 22:16:24 or is that portion of the error message not an accurate representation of the actual paths being used Mar 17 22:17:43 that's just an informational message Mar 17 22:18:38 if I look in tmp/staging/arm-angstrom-linux-gnueabi/include I see nothing at all Mar 17 22:18:54 however if I look there in a non-packaged-staging build dir I see all sorts of things Mar 17 22:19:04 it looks like packaged-staging is....not installing any includes Mar 17 22:20:48 isn't that a function of specifying the files using FILES_${PN}-dev += "/include/*" Mar 17 22:21:11 I'm not an expert by a long shot though Mar 17 22:26:05 yes and no Mar 17 22:26:12 the -dev packages probably get created file Mar 17 22:26:15 JustinP: when you say packaged staging, you mean the directory represented by ${D} right? Mar 17 22:26:25 the problem is that packaged-staging just doesn't work as it should.... Mar 17 22:26:35 no, what I mean is INHERIT += "packaged-staging" Mar 17 22:26:54 ah Mar 17 22:27:10 now you're beyong my knowledge of bitbake Mar 17 22:27:18 haven't looked at that class Mar 17 22:27:22 yet Mar 17 22:29:13 it is supposed to use packages to populate tmp/staging Mar 17 22:47:40 hi, all! Mar 17 22:51:28 * Ifaistos wonders why his OE build mysql dies on startup.... Mar 17 22:53:20 could anybody explain me tinderbox use policy? Mar 17 22:53:53 hillct: The cache a script of commands that creates all the volatiles. If enabled it's a *lot* faster than parsing each of the files in the /etc/default/volatiles. Mar 17 22:55:27 Ifaistos: It doesn't build at all for me Mar 17 22:56:20 v8jlene: what does my package need to do to utilize it? Mar 17 22:56:49 hillct : hmm strange. it builds but i think somthing is wrong with the configs. Mar 17 22:57:05 hillct : trying to get freepbx going Mar 17 22:57:32 feeePBX hiss boo hiss boo :( Mar 17 22:58:05 Ifaistos: what device are you trying to deploy on? Mar 17 22:58:20 a custom powerpc Mar 17 22:58:31 but using efika as test-bench Mar 17 22:58:37 hillct: Nothing really. You need to call "/etc/init.d/populate-volatile.sh update" in postinst. That will create your volatiles and if the cache is enabled it will regenerated the cache at the same time. Then on next boot the cache will be used instead of reparsing the files (assuming it's enabled of course). Mar 17 22:59:07 ok Mar 17 22:59:52 in my case, the standard make install, sets up the directories listed in /var for the first time Mar 17 23:00:38 I guess running populate-volatiles.sh with the directories already there won't make a difference Mar 17 23:00:39 hillct: The cache is in /etc/volatile.cache - you can take a look it to see what it does. Pretty obvious. It's enabled/disabled via a setting in /etc/default/rcS. (Guess I should finished of that chapter in the manual!) Mar 17 23:00:49 hillct : got all the perl, php stuff build, trying to run install_amp, passes all tests bur can not connect to mysql because the server dies Mar 17 23:00:56 Has anyone tried setting up a MACHINE=native build, and have problems compiling perl? Mar 17 23:01:22 hillct : on start up, so there is nothing to connect to Mar 17 23:01:40 and no mysql error log messages? Mar 17 23:02:03 edit /etc/my.cnf to increase the logging level Mar 17 23:02:09 so see what's going wrong Mar 17 23:02:51 hillct : only mysql started and mysql ended messages... Mar 17 23:03:26 increase the logging level Mar 17 23:04:10 SirManiac: Is you host system x86_64? Mar 17 23:04:19 no just x86 Mar 17 23:04:23 hillct : will look at it tomorrow. i am too tired to do any real work now Mar 17 23:04:49 v8jlene: a pentium 4 1.4Ghz to be exact Mar 17 23:05:13 hillct : i've seen your bug/patches. hopefully will give them a try to see how things go Mar 17 23:05:29 SirManiac: Hmm.. dunno. I seem to remember issue with it, but can't find anything related at the moment. pastebin some details Mar 17 23:05:36 Ifaistos: OpenPBX is looking pretty good Mar 17 23:06:09 v8jlene: i've noticed the do_configure changes include pathnames to the staging area, and most of those includes aren't built using native, wondered if there was already a special way to get around that. Mar 17 23:07:28 SirManiac: I've never used MACHINE=native, so I don't really know. I guess that means with native it's assuming a lot of stuff is already provided... Mar 17 23:08:56 v8jlene:yes, that's primarily all that native seems to do, just adds several ASSUME_PROVIDED lines Mar 17 23:10:48 night everyone Mar 17 23:10:58 night Mar 17 23:11:29 v8jlene: still want pastebin info, or do you think this merits a new bug report? Mar 17 23:12:58 y'know...CROSS COMPILE BADNESS checks are great and all....but I have no idea how to fix this...it's all autoconf Mar 17 23:14:17 I have an idea how to fix the perl CROSS_COMPILE stuff, I modify do_configure to check if MACHINE = native. Mar 17 23:14:44 SirManiac: No need for pastebin, I see from your explanation why it wouldn't work. You probably need to ask the question at a time when native users are around. But yeah, if you add a bug then someone who knows might answer. Mar 17 23:15:16 SirManiac: When I get some time I'll try and see if I can fix up perl. Mar 17 23:15:16 v8jlene: thanks Mar 17 23:15:57 v8jlene: I'm playing with that myself, and jsut might come up with a solution, so check my bug-report Mar 17 23:16:34 SirManiac: I will! Perl is a total mess, so good luck ;) Mar 17 23:16:50 v8jlene: thanks again :) Mar 17 23:21:47 Who has CROSS COMPILE BADNESS issues? I've been meaning to write a chapter for the user-manual on how to go about fixing that issue... generally it's reasonably easy... Mar 17 23:27:39 v8jlene: mpd is having CROSS COMPILE BADNESS Mar 17 23:27:52 v8jlene: although it seems to work fine.... Mar 17 23:28:00 v8jlene: but angstrom won't build it now... Mar 17 23:29:13 JustinP: I'll take a look at it. Mar 17 23:29:51 v8jlene: much thanks, I'd like to know how to fix these... Mar 17 23:42:30 I hae a recipe update for the logrotate package that implements some best practices and general cleanup. http://bugs.openembedded.org/show_bug.cgi?id=2014 Mar 17 23:46:26 hillct: Wow, that patch style is hard to read ;) I would prefer if you used unified diffs for patches (ie, diff -du) since I think it makes them a lot more readable! Mar 17 23:46:47 ok Mar 17 23:46:50 one sec Mar 17 23:47:00 hillct: Or generate them with monotone - do a "mtn diff ." in the directory of the package you have modified. Mar 17 23:47:38 hillct: The monotone way needs you to do a "mtn add" for any new files of to make them show up. Mar 17 23:47:46 can't do that given my buildenviroment, for the moment Mar 17 23:47:53 but patch-du I can do Mar 17 23:48:19 er diff -du I mean Mar 17 23:48:38 hillct: No problem. Mar 17 23:51:55 updated Mar 17 23:59:44 hillct: Ah, that's more like it ;) Thanks. Mar 17 23:59:55 np Mar 18 00:00:13 updated openpbx recipe momenterily Mar 18 00:00:54 hillct: Just checked logrotate's history and see that you added it, so that'd make you the maintainer so I can push that change now for you. Mar 18 00:01:36 thanks Mar 18 00:03:05 hillct: Do you want me to use your email address from bugzilla as the author id? Mar 18 00:03:37 hillct@pinnacledigital.com Mar 18 00:03:40 er Mar 18 00:03:44 make that Mar 18 00:04:03 hillct-oe@pinnacledigital.com Mar 18 00:06:30 are pkgconfig files like confuse-config meant to be in staging/i686-linux/bin or staging/i686-linux/bin/arm-linux ? Mar 18 00:06:37 03hillct-oe 07org.oe.dev * r4b0ecf8b... 10/ (1 packages/logrotate/logrotate_3.7.1.bb): Mar 18 00:06:37 logrotate 3.7.1: Eliminated hard coded paths, removed recommendation of gzip Mar 18 00:06:37 since it's typically provided by busybox which we can expect to always be Mar 18 00:06:37 present. Also added proper config file specification. Closes #2014. Mar 18 00:10:36 rwhitby: http://lists.linuxtogo.org/pipermail/openembedded-devel/2006-December/001082.html Mar 18 00:11:13 Those pkgconfig scripts would be cross files... so STAGING_BINDIR_CROSS ? Mar 18 00:11:54 actually, this is an ftdi-eeprom-native build looking for a confuse-native library. Mar 18 00:13:51 so that means confuse-config should go directly in STAGING_BINDIR_NATIVE, when put there by confuse-native.bb and ftdi-eeprom-native.bb should be looking in there for it. Mar 18 00:14:53 so it's being put in the right place, but ftdi-eeprom-native's configure is not finding it. Mar 18 00:17:19 Is it not finding it or finding the wrong one? Looks like STAGING_BINDIR_CROSS is always in the path before STAGING_BINDIR_NATIVE, so it should fine the cross one first and the native one otherwise? Mar 18 00:20:03 Oh, native.bbclass rewrites STAGING_BINDIR_CROSS to point at the STAGING_BINDIR_NATIVE, which is then appended to the path... so that should be ok. Mar 18 00:20:45 ah, looks like the configure script doesn't even look for confuse-config, even though it looks for libftdi-confing. Mar 18 00:22:38 so ftdi-eeprom-native causes OE to have a libconfuse host library dependency. Mar 18 00:31:38 JustinP: Finally got mpd built... it depends on a lot! What was your CROSS COMPILE BADNESS error? (It build fine here) Mar 18 00:33:44 v8jlene: insane.bbclass quit after the configure due to -I/usr/include. Looks to me like AC_CHECK_HEADER for iconv.h was including it somehow Mar 18 00:33:52 v8jlene: are you building for angstrom? Mar 18 00:34:18 * rwhitby gives up on ftdi-eeprom-native - my autofoo is not strong enough Mar 18 00:34:24 What is the deal with STAGING_BINDIR/STAGING_BINDIR_NATIVE? I'm ending up with this path: /tmp/staging/x86_64-linux/bin/arm-linux, which frankly doesn't make any sense at all. Mar 18 00:34:49 why does the arm-linux directory end up there at all? Mar 18 00:35:10 NAiL: http://lists.linuxtogo.org/pipermail/openembedded-devel/2006-December/001082.html Mar 18 00:38:08 JustinP: I'm building for my own distro no angstrom. I tested on my i486 build which doesn't include insane.bbclass yet. I'll retry on my sh4 build which does have insane.bbclass enabled. Mar 18 00:43:30 v8jlene: ok Mar 18 00:45:48 Hey peeps :) Mar 18 00:46:38 I ran into a problem compiling Perl using the MokoMakefile and since it's not OpenMoko-specific I was guided here to see if you guys had a clue what it's about Mar 18 00:47:00 http://pastebin.ca/399471 is the end of the log Mar 18 00:53:13 rwhitby: thanks ;) Mar 18 00:55:41 03nail 07org.oe.dev * r85d31d9c... 10/ (1 packages/php/php_5.2.0.bb): php: Unbreak php 5.2.0 build Mar 18 01:08:37 so close... Mar 18 01:08:50 forgot to update volatiles cache... Mar 18 01:09:15 JustinP: Ok, I see it. It's searching host headers for iconv, so just need to talk it out of doing that. Mar 18 01:09:53 v8jlene: ir have it look in the right places....I think the same is happening for other headers too. Please let me know what you do to fix it :-/ Mar 18 01:10:08 v8jlene: will populate-volatiles update the cache the first time it runs without the arg 'update' ? Mar 18 01:11:01 Adding a "--disable-iconv" to EXTRA_OECONF shuold fix it... but it may need iconv if being built with uclibc. Mar 18 01:11:30 hillct: No, without update the cache is not touched. Mar 18 01:13:16 v8jlene: well, sure, --disable-iconf might work but it doesn't fix what's actually happening.... Mar 18 01:14:12 JustinP: Well, it does. It's looking to see if iconv is installed, and it's looking in /usr/include. So what you need to do is either stop it looking (if it's not needed) or make it look in the staging area instead. Mar 18 01:15:30 v8jlene: yes. And I would have no idea how to make it look in staging since it's AC_CHECK_HEADER that's doing it Mar 18 01:16:14 the same error is happening for mpcdec.h Mar 18 01:16:36 JustinP: If you look in configure.ac and search for iconv you'll see if has options to change where to look and or to disable looking for those headers/libraries. Mar 18 01:20:42 v8jlene: you seem to be suggesting fixing the symptom instead of the problem.... Mar 18 01:22:27 JustinP: Err, well I think of the problem as being "autoconf is searching host directories", the solution is to talk it out of that? That's how it's handled in all of the current recipes. Mar 18 01:27:16 v8jlene: ok, it sounded like you meant to just disable any feature that cuased the problem. Mar 18 01:30:45 yeah....I still haven't found anything... Mar 18 01:34:42 JustinP: Oh, no. In general you actually want to configure where to search for it, not disable it. If its not actually needed/wanted then it can be disabled instead. Mar 18 01:38:58 v8jlene: the openpbx package recipe has been updated to address your concerns as well as NAiL's and addition of log rotation support Mar 18 01:39:07 v8jlene: http://bugs.openembedded.org/show_bug.cgi?id=2011 Mar 18 01:45:04 bleh....I had to disable flac and mpc...I have no idea why autoconf is looking in host includes... Mar 18 01:45:13 I'll keep trying Mar 18 01:46:55 ok....well....it's because it's using includedir for including which OE sets to /usr/include.... Mar 18 01:49:50 You can add "mpcdec_prefix=${STAGING_LIBDIR}/.." to EXTRA_OECONF to fix mpc. That flac version test looks like it needs changes to configure.ac. Mar 18 01:50:30 I think it sets it /usr/include because that's what the include dir will be when it is installed into the target... for things that don't care about that you could probably change the default. Dunno. I didn't make any of it work this way ;) Mar 18 01:56:20 *python* has QA problems? Mar 18 01:59:54 JustinP: QA issue about .debug file for python-misc ? Mar 18 02:00:12 * cyrilRomain|zZz half asleep Mar 18 02:02:47 cyrilRomain|zZz: yes Mar 18 02:02:55 JustinP: 13:01 < cyrilRomain> mickey_away: it seems that python-2.4.4-manifest.inc need to be regenerated by generate-manifest.py. I've had QA issues http://rafb.net/p/erArqx12.html that disappeared with a renegerated manifest (python-misc is indeed not here anymore) Mar 18 02:04:01 cyrilRomain|zZz: I don't see a generate-manifest.py in packages/python.... Mar 18 02:04:19 nm, it's in contrib Mar 18 02:04:24 my fault, in contrib/python/ Mar 18 02:07:22 ok, I'll try that Mar 18 02:07:28 off to pick up a friend Mar 18 02:07:50 ok, 'night Mar 18 02:07:54 * cyrilRomain|zZz back to sleep **** ENDING LOGGING AT Sun Mar 18 02:59:56 2007