**** BEGIN LOGGING AT Thu Nov 12 14:27:36 2009 Nov 12 14:27:52 alecrim: http://cgit.openembedded.org/cgit.cgi/bitbake/log/?h=1.8 Nov 12 14:27:53 ? Nov 12 14:30:31 spaetz, thanks! :) Nov 12 14:36:48 Crofton, Ida got you? :( Nov 12 14:37:30 * mwester-laptop read the backlog, and wonders what awaits him for the next SlugOS build on the dev branch tip :D Nov 12 14:37:50 Good Morning, BTW. Nov 12 14:40:33 morning Nov 12 14:42:49 alecrim maybe you searched for http://developer.berlios.de/project/showfiles.php?group_id=2701&release_id=11520 Nov 12 14:44:11 mwester-laptop: hi Nov 12 14:44:58 mwester-laptop: If there are problems looking like they're from these changes, let me know Nov 12 14:55:14 03Richard Purdie  07org.openembedded.dev * r2f982a8b54 10openembedded.git/conf/bitbake.conf: Nov 12 14:55:14 bitbake.conf: Fix TERMCMD variables (from Poky) Nov 12 14:55:14 Signed-off-by: Richard Purdie Nov 12 14:55:28 I think that is cia "fixed" to only report .dev changes Nov 12 14:56:30 RP what was staging_prefix now? Nov 12 14:56:43 woglinde: ? Nov 12 14:56:49 I mean /usr for angstroem or / for micro Nov 12 14:57:07 wt package needs to set BOOST_DIR to the right Nov 12 14:58:02 RP: excellent Nov 12 14:58:10 woglinde: ${STAGING_DIR_HOST}${prefix} ? Nov 12 14:58:29 rp yes Nov 12 14:58:46 wasnt sure if a shortcut exist Nov 12 14:58:54 RP: what about adding stable/2009 to cia? Nov 12 15:09:10 RP: doh, my proposal for SRC_URI[name:md5sum] doesn't work because you can't have colons in flag names. Nov 12 15:09:22 still, I guess a dot is a reasonable substitute. "name.md5sum" seems to be acceptable to bitbake. Nov 12 15:10:31 03Koen Kooi  07org.openembedded.dev * r1de62e68d1 10openembedded.git/recipes/gstreamer/gst-plugins-good_0.10.15.bb: gst-plugins-good: depend on flac as well Nov 12 15:10:43 I think MD5SUM[name] looks prettier, but hey :) Nov 12 15:12:51 hm FILESPATH was the var where bitbake looks for patches? Nov 12 15:13:02 mickey: pong, let's talk tomorrow, OK Nov 12 15:13:13 Had a big meal, some alcohol and ready to go to bed Nov 12 15:14:09 Laibsch: k, g'night Nov 12 15:18:54 pb_: with a single server we don't need to checksum the metadata, right? Just the sources. Nov 12 15:19:16 in git we trust Nov 12 15:19:49 ant_work: that's the theory. it would be fairly silly to check the metadata against a sum that is stored in the same place. Nov 12 15:20:22 well, maybe not completely silly: it would guard you against some kinds of disk corruption or network-introduced errors, but oe has never really striven for that level of robustness. Nov 12 15:20:48 hm, I mean a checksum for every file, .bb and patches included ( à la Gentoo-Manifest) Nov 12 15:20:49 so yes, "in git we trust" is probably the (unlikely-sounding) watchword. Nov 12 15:20:57 ;) Nov 12 15:21:32 You could do a kernelish automatic flag that the tree is dirty. Nov 12 15:26:13 so, here we are. http://pastebin.ca/1668300 Nov 12 15:26:17 I think that's approximately what's needed. Nov 12 15:42:21 good morning Nov 12 15:43:45 kgilmer: hi Nov 12 15:44:45 hi mckoan. did you see this page? http://wiki.openembedded.net/index.php/ToolingUseCases Nov 12 15:44:58 anyone thought about doing an automatic INHIBIT_DEFAULT_DEPS for recipes that set PACKAGE_ARCH = "all"? Nov 12 15:48:42 pb_: ah, I hadn't realised : was special for that use... Nov 12 15:48:58 03Michael 'Mickey' Lauer  07org.openembedded.dev * r7d2be53866 10openembedded.git/ (2 files in 2 dirs): mobile-broadband-provider-info: new recipe; database for mobile broadband provider info (sic!) Nov 12 15:49:06 03Michael 'Mickey' Lauer  07org.openembedded.dev * r909f2ce8d6 10openembedded.git/recipes/vala/vala-native_git.bb: vala-native: remove git repository; we prefer release versions for it nowadays Nov 12 15:49:09 03Michael 'Mickey' Lauer  07org.openembedded.dev * r3297ada06f 10openembedded.git/recipes/freesmartphone/fsogsmd_git.bb: fsogsmd: depend on mobile-broadband-provider-info and libxml2 Nov 12 15:50:16 kgilmer: yes thx, althought I don't understand how that distincion could be helpful :-) Nov 12 15:51:54 kergoth: That would be interesting and easy to do in base.bbclass Nov 12 15:53:07 mckoan, your tool (i think) is directed to application developers Nov 12 15:53:15 ote is directed to package developers Nov 12 15:53:37 RP brought up the distinction at oedem and i'm trying to explicitly define them Nov 12 15:54:07 so i can say, "if you are X type of user, these tools might be helpful." etc Nov 12 15:54:53 RP: yep.. i can't think of any case where it would be a bad thing. just noticed that it ran off building glibc and stuff just to make zd1211-firmware :) Nov 12 15:56:59 kergoth: We should have kernel.bbclass take out the libc dependency too Nov 12 15:58:56 good idea Nov 12 16:01:06 hrw: nice post on your planet! Nov 12 16:01:08 * mickey|office -> home Nov 12 16:07:33 ok crisis one under control Nov 12 16:08:01 mwester-laptop, Ida + flaky sewer Nov 12 16:08:46 Gaaah! I've had a sewer backup in the basement once. It was bad, really really bad.... :( Nov 12 16:09:10 this wasn't that bad Nov 12 16:09:15 in that sense Nov 12 16:09:31 it was very dilute .... Nov 12 16:09:40 they didn't really notice Nov 12 16:10:01 at first they thought it was from outside, largely due to a sump pump issie Nov 12 16:11:00 Well, I hope it doesn't get worse there. Nov 12 16:11:23 bye Nov 12 16:11:46 the sewer on the street is running now Nov 12 16:11:54 so the worst is over Nov 12 16:12:40 RP: it's not so much that ":" is special, just that the regexp in the parser doesn't admit it. Nov 12 16:13:13 RP: "[a-zA-Z0-9\-_+.]" is the set of allowed characters, it seems. if you deviate from that then you get the time-honoured "unhandled line" diagnostic :-} Nov 12 16:14:13 heh, someone point out the fact that you can't use the url in a variable / flag? Nov 12 16:14:29 * kergoth gets tylenol Nov 12 16:15:21 kergoth: heh. not exactly that, but close. Nov 12 16:15:28 ah :) Nov 12 16:16:18 kergoth: I had wanted to use SRC_URI[name:md5sum] = "..." as the notation for defining tarball checksums. Nov 12 16:16:26 kgilmer: better try Lysergic acid diethylamide Nov 12 16:16:29 but, no problem, I can use SRC_URI[name.md5sum] instead ;-} Nov 12 16:16:46 kergoth: better try Lysergic acid diethylamide (soory ken) Nov 12 16:18:10 kergoth: heh, last time I was in the USA it took me ages to discover that "tylenol" is what you need to ask for to get paracetamol. Nov 12 16:20:53 heh Nov 12 16:21:57 kergoth: any news from Robert? Nov 12 16:22:19 kergoth: wrong nick, sorry Nov 12 16:22:28 that explains why i was so confused Nov 12 16:23:54 mckoan: heh, your success rate with nick completion is not very good at the moment Nov 12 16:24:13 pb_: actually not :-) Nov 12 16:26:30 pb_: Is there a case for making the parser more accepting a wonder? Nov 12 16:26:48 pb_: I carry painkillers with me for that reason... Nov 12 16:27:43 * mwester-laptop interrupts his OE/SlugOS attempt in order to (suddenly) learn more about heating oil leaks :( Nov 12 16:29:03 mwester-laptop: :/ Nov 12 16:29:48 has someone already used the embedded VNC support of qt-embedded ? Nov 12 16:29:49 re Nov 12 16:30:02 longfield zecke long time ago Nov 12 16:31:22 because I need it and I have an error when I try to start the qws with vnc: VNC: driver not found Nov 12 16:32:15 mickey|zzZZzz, ping Nov 12 16:32:35 regarding my python-native problem, I found this in messages Nov 12 16:32:37 Nov 12 09:05:24 elephant kernel: python[2231]: segfault at c8 ip 00007f0f7e4a5aec sp 00007fffd7dd65d0 error 4 in libncurses.so.5.0.4[7f0f7e474000+4d000] Nov 12 16:32:39 RP: yeah, I think a slightly more tolerant parser would probably be a good thing. Nov 12 16:32:51 I wonder if python-native is using the host ncurses Nov 12 16:32:52 I don't think there is any good reason to restrict the set of characters in identifiers to that extent. Nov 12 16:34:51 RP: you did the scheduler, have you given any thought to what would be involved in supporting multiple build nodes? i guess a dumb one that just remotely executes a bitbake -b with a different tmpdir but shared staging via ssh or something might be done in a reasonable period of time, but I'm thinking to really do it well, offload chunks of the dependency digraph perhaps, weight the nodes, probably nontrivial Nov 12 16:38:50 kergoth: isn't that more or less the same problem as the existing BB_NUMBER_THREADS thing? I would have thought that you would just need to swap in a remote implementation of cooker.tryBuild() at the appropriate point in runqueue. Nov 12 16:39:19 kergoth: I did give it some thought - the scheulder is actually plugable Nov 12 16:39:29 hum, I guess the VNC plugin is not installed or built by OE by default Nov 12 16:40:24 kergoth: Half the problem is as pb_ says, just a question of making something run via ssh as a remote command. The other half is teaching the scheduler the idea of node affinity of builds Nov 12 16:40:26 longfield right Nov 12 16:40:29 pb_: if you just want to send over builds of one recipe at a time, thatd probably do. you'd likely want to implement a weighting system for the nodes, though, given there's likely disparity among them Nov 12 16:40:32 * kergoth nods Nov 12 16:40:34 hmm Nov 12 16:40:57 kergoth: So the code doesn't preclude it and should be easy enough to add support Nov 12 16:41:16 qt4-embedded-plugin-gfxdriver-gfxvnc, that's its sweet name Nov 12 16:41:30 kergoth: I always wanted to weight tasks with the CPU/NET/DISK usage with the scheduler accounting for that too Nov 12 16:41:51 yeah, me too Nov 12 16:42:38 kergoth: iirc, the runqueue operates on tasks rather than whole packages at that level already, so it is already relatively fine grained. providing some kind of scheduler weighting would be a good thing although I suspect it is probably not quite as important as you might first think. Nov 12 16:43:19 we could easily remotely run the emitted shell scripts, given a shared tmpdir... python tasks would be a bit more difficult, would probably want a remote xmlrpc bitbake process, or pickle everything and send it over Nov 12 16:43:20 kergoth: to do it really intelligently, you would need to decorate the package metadata with hints as to how much resource is needed so that you didn't, for example, accidentally dispatch glibc:do_compile() to your slowest node. Nov 12 16:44:01 kergoth: remember that bitbake actually reparses state, doesn't save it Nov 12 16:44:15 kergoth: so just running bitbake remotely could be good enough Nov 12 16:44:33 kergoth: Obviously with some "force this task in this file" flag Nov 12 16:44:37 so a bitbake -b, like i was thinking. i don't think I'd trust bitbake to share tmp's cache and everything Nov 12 16:44:42 :) Nov 12 16:45:06 kergoth: yea of little faith :) Nov 12 16:45:20 hehe Nov 12 16:46:55 set up a var with a list of nodes, a few vars to configure those nodes, and offload bitbake -c -b commands in the cooker.. Nov 12 16:47:04 guess an initial pass would be a fair bit easier than i thought Nov 12 16:51:18 Crofton, i mailed my 2cents on the software section of oedem just now. Nov 12 16:51:56 hi ken Nov 12 16:52:55 hey woglinde :) Nov 12 16:54:04 have a nice rest of the day Nov 12 16:56:41 hmm, i need to get better at estimation Nov 12 16:56:53 should start keeping track of estimate vs reality on previous tasks Nov 12 17:01:38 pb_: as arm expert, http://lists.infradead.org/pipermail/linux-arm-kernel/2009-November/003860.html Nov 12 17:02:11 03Jaap de Jong  07org.openembedded.dev * r2ece13c424 10openembedded.git/ (conf/checksums.ini recipes/ptpd/ptpd_1.0.0.bb): (log message trimmed) Nov 12 17:02:11 Precision Time Protocol (PTP) as defined by the IEEE 1588 standard Nov 12 17:02:11 Hi there, Nov 12 17:02:11 attached a proposal for a recipe for building the Nov 12 17:02:11 Precision Time Protocol daemon (ptpd). Nov 12 17:02:13 Suggestions by Michael Smith: Nov 12 17:02:15 > Nov 12 17:02:49 ant_work: that will work for some arches but not all Nov 12 17:03:07 pb_: I'm still trying to see why 2.6.3x won't boot from nand (armv5te-Zaurus) Nov 12 17:03:11 hi, I'm new to oe, and I don't understand something : I wrote MACHINE_EXTRA_RRECOMMENDS = "kernel-modules" in the machine config file (thinking it would put every compiled modules in my image), but a lot of kernel modules are missing in my rootfs... Nov 12 17:03:25 this is, unfortunately, an artifact of the poor way in which linux represents I/O resources internally. Nov 12 17:03:44 RP, have you ever considered to have bitbake run like a daemon? this would be a different approach from xmlrpc... but essentially the state of a given oe tree would remain in memory. Nov 12 17:04:09 there are things i would like to ask of bitbake but not in the context of running a specific task Nov 12 17:04:35 such as, 'where is the tmpdir for a given package' Nov 12 17:04:38 kgilmer: have you tried "bitbake -i"? Nov 12 17:05:11 maybe that could be extended to do what you want. Nov 12 17:05:33 ant_work: I doubt that this issue would be related to non-booting. Nov 12 17:06:19 pb_: sure...Im reading all LAKML posts hoping to find indices... :/ Nov 12 17:07:06 pb_: this on was interesting Nov 12 17:07:23 pb_, that looks about right Nov 12 17:07:41 i will have to investigate more, but seems like it's just what i need. Nov 12 17:07:43 thx! Nov 12 17:10:00 kgilmer: That is what the bitbake xmlrpc system does Nov 12 17:10:29 pb_: I could have more luck with "NAND: failure reported with mtd_oobtest(read delay violated)"... Nov 12 17:29:39 03Michael 'Mickey' Lauer  07org.openembedded.dev * r58b64b5319 10openembedded.git/recipes/ptpd/ptpd_1.0.0.bb: ptpd: stylize Nov 12 17:35:42 CjNr12: I'm new to OE but maybe you want to look at BOOTSTRAP_EXTRA_RDEPENDS ? Nov 12 17:36:45 rednul: thx, I will take alook Nov 12 17:38:27 hi gregoiregentil Nov 12 17:38:32 Hello Nov 12 17:39:12 gregoiregentil would have been cool you would made it to oedem Nov 12 17:39:52 Sorry. It was in UK? Nov 12 17:40:03 A little bit too far away from California. Nov 12 17:40:09 gregoiregentil yes Nov 12 17:58:49 ~seen effem Nov 12 17:58:53 effem was last seen on IRC in channel #oe, 7h 35m 49s ago, saying: 'Karmic does not support armv5te :-('. Nov 12 17:59:06 so wt which uses boost seems to work with boost 1.4.0 too Nov 12 17:59:14 only 11 packages to go or so Nov 12 18:01:33 wt = wesnoth? Nov 12 18:01:37 nope Nov 12 18:01:42 ok Nov 12 18:01:46 webtools Nov 12 18:01:53 ok Nov 12 18:02:01 new wesnoth has 200mb download Nov 12 18:02:14 I'll try wesnoth when I'll have the time...(so after JaMa's branch test) Nov 12 18:02:34 wesnoth right now does not compile Nov 12 18:02:58 but I am thinking its worth to upgrade to the latest version Nov 12 18:03:05 instead of fixing old one Nov 12 18:32:19 03Henning Heinold  07org.openembedded.dev * rc29c0219c0 10openembedded.git/ (8 files in 5 dirs): Nov 12 18:32:19 wt3: update to version 3.0.0 Nov 12 18:32:19 * spilt up with .inc-file and switch to INC_PR Nov 12 18:32:19 * disable qt explicit in CMakeLists.txt Nov 12 18:34:55 he effem Nov 12 18:35:00 I updated wt3 Nov 12 18:35:10 seems to work with boost-1.4 too Nov 12 18:35:15 supper now Nov 12 18:38:12 woglinde: hi Nov 12 18:38:35 nice, didn't try wt3 yet Nov 12 18:40:57 03Khem Raj  07org.openembedded.dev * r9ea5e56d70 10openembedded.git/recipes/klibc/ (klibc-1.5.15/signal-cleanup.patch klibc_1.5.15.bb): Nov 12 18:40:57 klibc-1.5.15: Unbreak the build for x86 with kernels < 2.6.31 Nov 12 18:40:57 Signed-off-by: Khem Raj Nov 12 18:43:26 pb_, please s/and Got/and got/g in your Death to checksums.ini; TIA Nov 12 18:44:09 pb_, oh, and just because it's in that patch to, may i ask to change if type == "http" or type == "https" or type == "ftp" or type == "ftps": Nov 12 18:44:47 is the python module 'glob' available in any of the python packages? Nov 12 18:44:48 pb_, to something like: if type in ["http", "https", "ftp", "ftps"]: Nov 12 18:45:05 python-misc doesn't have it Nov 12 18:49:15 frikker: did you check the python-manifest-2.6.inc? Nov 12 18:49:18 cool - glob is pure python so i dont need to compile it. nevermind :) Nov 12 18:49:37 kergoth, no i did not, didnt think of it. i'm looking at ipkg Nov 12 18:54:22 03Frans Meulenbroeks  07org.openembedded.dev * rb8e6b2627c 10openembedded.git/ (3 files in 2 dirs): libhtml-parser-perl and -namtive, upgraded to 3.64 Nov 12 18:58:51 mmap: Permission denied Nov 12 18:58:58 anyone know how to make that work Nov 12 18:59:05 I think the problem is running qemu Nov 12 18:59:14 F11 w/ selinux Nov 12 19:00:20 http://lists.o-hand.com/poky/0337.html perhaps Nov 12 19:00:45 other than that, it's hard to say from those symptoms. if you could pinpoint which particular mmap() is failing then that might give some clues. Nov 12 19:02:04 I think it is some xll thing Nov 12 19:02:05 er Nov 12 19:02:07 selinux Nov 12 19:02:15 need to go look in my deleted emails Nov 12 19:05:19 RP: ping Nov 12 19:08:51 On a clean build icu is failing. I suspect the SYSROOT_PREPROCESS_FUNCS change to the icu recipe Nov 12 19:11:42 http://pastebin.com/m58010e24 Nov 12 19:12:40 Is there a practical way to track variable modifications in OE? I have a case in wich "bitbake -e" shows me a certain value for a variable and when I print the value of this variable from a recipe, it shows a different value. Nov 12 19:14:05 RP: there is a quite suspicious looking directory structure in the icu work dir: icu-3.6-r4/sysroot-destdir/home/sakoman/ ... Nov 12 19:21:04 pb__, I found the fix, had to set some selinux bool Nov 12 19:21:15 rp day had posted about it a few days ago Nov 12 19:21:23 very good Nov 12 19:21:43 ok, moving on the .stable front again Nov 12 19:30:45 hmm... I get git-native built without curl support which leads to Nov 12 19:30:46 error: git was compiled without libcurl support. Nov 12 19:30:59 when trying to get the rev for a repo via http protocol :| Nov 12 19:40:09 two quick question: 1) 'bitbake virtual/kernel -c menuconfig' shows a terminal that instantly closes. what am i doing wrong? Nov 12 19:41:08 2) what would be the best approach to build a bealgleboard kernel with real-time capabilities. copy the beagleboard recipe and add the patches? or ... dunno .... Nov 12 19:51:43 anyone ? Nov 12 19:56:36 mlip: do you have ncurses development package installed ? Nov 12 19:58:11 i've read this hint, i've installed ncurses on my distro (gentoo), and also built ncurses via oe (dont think that this actually helps though) Nov 12 20:00:05 mlip: hmmm what is your default terminal Nov 12 20:00:58 gnome-terminal Nov 12 20:01:09 khem: gnome-terminal Nov 12 20:01:20 mlip: are you on .dev ? Nov 12 20:02:01 yes, pulled at around 19 o'clock or so Nov 12 20:05:28 khem: just wanted to share that i finally managed to Nov 12 20:05:42 mlip: http://www.gumstix.net/wiki/index.php?title=Kernel_Reconfiguration Nov 12 20:05:44 compile the kernel for my c700, thanks for the help :-) Nov 12 20:05:59 qsp: good. Nov 12 20:06:31 qsp: perseverance is a virtue :) Nov 12 20:06:38 khem: thx .... ;) Nov 12 20:07:24 mlip: if you see anything that you came across can help others .. you could add it to OE wiki Nov 12 20:07:34 khem: also a hint on how the 'right' approach is to get an real-time kernel running using the kernel that is specified for the machine? Nov 12 20:07:38 khem: sure Nov 12 20:07:52 :-) now that i have a working .config, i'll slowly reenable all the features, and if i have something better i'll share :-) so that the c7x0 angstrom supports the c700 itself :-) Nov 12 20:07:53 and also qsp :) Nov 12 20:08:05 yep, sure :-) Nov 12 20:09:29 mlip: you might have to backport rt stuff. I dont know if everything is upstream yet Nov 12 20:10:23 khem: but basically the thing is to fork the appropriate recipe used by my machine and then try to apply the rt patches ? Nov 12 20:11:11 mlip: linux-rt would be best to start of Nov 12 20:11:45 linux-rt_2.6.29.bb would be latest Nov 12 20:11:55 I dont know how well they work though Nov 12 20:12:11 khem: i found the recipce, but I also dont want to miss the patches that are applied for my machine Nov 12 20:12:22 sure Nov 12 20:12:35 you can add them on top Nov 12 20:12:47 oke , thought so Nov 12 20:13:40 and if they are specific to your machine only then safest is to use machine override Nov 12 20:13:45 to add the patches Nov 12 20:14:06 which will make sure that the patches are only applied when this particular machine is selected. Nov 12 20:14:17 that way you dont mess up others Nov 12 20:14:24 oke Nov 12 20:14:52 btw. menuconfig still not working, bitbake.conf is already adjusted by default to the changes stated in the wiki ;) Nov 12 20:14:56 SRC_URI_append_ Nov 12 20:15:01 ok Nov 12 20:16:28 mlip: can you try to use xterm Nov 12 20:20:22 i think the problem is that i dont have any $TERMRCFILE defined Nov 12 20:23:04 RP yet another perl version issue, I get: Nov 12 20:23:20 Cannot find '/usr/lib/perl/5.8/CORE/perl.h' have you installed /home/frans/oe/tmp_angstrom/staging/i686-linux/usr/bin/perl? Nov 12 20:24:05 assuming /usr/lib etc is prefixed somewhere this still cannot work as the dir in staging is .../usr/lib/erpl/5.8.8/CORE/perl.h Nov 12 20:24:10 :q Nov 12 20:24:37 no idea how to fix this yet, open for suggestions :-) Nov 12 20:28:07 mickeyl: ping Nov 12 20:28:11 are you there? Nov 12 20:34:00 khem: xterm fails too Nov 12 20:34:33 i think it's a problem with the $TERMRCFILE not being defined Nov 12 20:37:56 RP: re-hi: dbus-glib-native failed again (from scratch) Nov 12 20:38:45 ant__: are the logs somewhere? Nov 12 20:38:59 and devicekit fails too on a clean build :-( Nov 12 20:39:04 yea, now @home I have all logs Nov 12 20:39:43 PB the odd thing is that the native version of the recipe does build Nov 12 20:39:58 s/PB/RP/ Nov 12 20:40:50 RP: do you need the runlogs too? Nov 12 20:42:13 tarball here http://filebin.ca/unmbtv/temp.tar.gz Nov 12 20:45:57 eFfeM: you've not mentioned which recipe is failing Nov 12 20:46:47 i think I'm going to go ahead and fix some odd quoting issues in some vars. current parser is cool with it, but it probably shouldn't be Nov 12 20:46:49 ant__: The odd thing is do_compile running something from staging /oe/build/tmp/staging/i686-linux/usr/bin/dbus-daemon Nov 12 20:46:55 combined features for example has "" within the var, and the var is quoted with "" too :) Nov 12 20:47:03 * kergoth ponders Nov 12 20:47:18 ant__: This is a clean tmp, right? Nov 12 20:48:41 RP: dbus-glib-native builds for me on a top of tree clean build Nov 12 20:48:58 icu and devicekit do not though Nov 12 20:49:04 RP: yes Nov 12 20:49:28 ant__: Right, so its not dbus-glib-native that has the problem but dbus-native Nov 12 20:49:30 in the meanwhile I tried -c clean but same result :/ Nov 12 20:49:40 ah...deps Nov 12 20:50:06 sakoman: icu or isu-native? Nov 12 20:50:28 RP: are you richart purdie ? Nov 12 20:50:31 -t +d Nov 12 20:50:39 mlip: yes Nov 12 20:50:52 RP found the cause it is again the dreaded perl./config.sh Nov 12 20:51:04 eFfeM: broken contents this time? Nov 12 20:51:19 RP: great! just found your bitbake.conf patch you submited about 6h ago, which actually breaks 'bitbake virtual/kernel -cmenuconfig' Nov 12 20:51:28 RP: icu - I suspect the SYSROOT_PREPROCESS_FUNCS change to the icu recipe Nov 12 20:51:37 RP can't build a package cross which builds native Nov 12 20:51:42 found this Nov 12 20:51:44 ./i686-linux/perl/config.sh:archlibexp='/home/frans/oe/tmp_angstrom/staging/i686-linux/usr/lib/perl/5.8.8' Nov 12 20:51:45 ./armv7a-angstrom-linux-gnueabi/usr/lib/perl/config.sh:archlibexp='/usr/lib/perl/5.8' Nov 12 20:51:45 RP: http://pastebin.com/m58010e24 Nov 12 20:52:13 oke, i will come back in 5 mins, maybe RP is available then ;) Nov 12 20:52:31 kergoth: sounds like a good idea Nov 12 20:52:33 RP the native one has 5.8.8 the arm7 one has 5.8 (and not sure about the lack of prefix in the armv7 version Nov 12 20:53:18 heh, "bitbake -k -g world" takes a long time Nov 12 20:53:36 python has been running for about 20 minutes now. Nov 12 20:53:42 mlip: That sounds bad :/ Nov 12 20:54:43 RP: '....${SHELLCMDS}' seems to be different from '...$SHELLCMDS' Nov 12 20:55:46 pb__: I had created a prototype parser for bb.parse that used pyparsing. Worked, too, but it's a bit more picky about quoting and such :) Nov 12 20:56:07 do_menuconfig exports a SHELLCMDS variable, but so does bitbake.conf, I dont get which overrides which Nov 12 20:56:47 eFfeM: any idea whats braking it? Which recipe shows this breakage? Nov 12 20:56:53 kergoth: oh right, cute. Nov 12 20:56:57 course, it was slower Nov 12 20:56:59 but fun anyway Nov 12 20:57:13 RP, I'm trying to add a recipe for Tk (needed by some xmltv grabbers) Nov 12 20:57:28 ant__: you need to find why dbus-daemon in native staging is a wrapper and not the real thing Nov 12 20:57:45 eFfeM: So other recipes are working at least? Nov 12 20:57:52 it requires CORE/perl.h which in the makefile is prefixed with archlibexp Nov 12 20:58:16 RP I have no problem building others until now Nov 12 20:58:20 kergoth: do you know what became of holger's ast parser? Nov 12 20:58:29 eFfeM: Could we just make it empty? Nov 12 20:58:44 pb_: sitting in a branch waiting for some attention :( Nov 12 20:58:44 RP: I see in dbus-native do_stage + autotools_stage_all Nov 12 20:58:59 RP, hm, no idea Nov 12 20:59:11 pb__: it's been on branches, one based on 1.8 and one based on master, for a while. i think it may be a candidate for inclusion in 1.10 Nov 12 20:59:12 i'll have a look Nov 12 20:59:27 kergoth, pb_: It is intended for 1.10, yes Nov 12 20:59:32 great Nov 12 20:59:55 kergoth: or at least lets rearrange things a bit and make it easier. I still need to understand that code Nov 12 21:00:08 looks like mvl6 will move to 1.8.18 shortly, thankfully. 1.10, who knows... Nov 12 21:00:13 kergoth, RP: ah, very good Nov 12 21:00:13 sakoman: I'm running a build to try and reproduce the problem Nov 12 21:00:32 we're down to 6 bitbake commits that 1) aren't upstream yet, and 2) aren't specific to mvl6 Nov 12 21:00:35 kergoth: great Nov 12 21:00:38 * kergoth will send em to the list shortly Nov 12 21:01:16 RP: fwiw it fails even after compilation of dbus-native Nov 12 21:01:25 now checking the staging of the latter Nov 12 21:01:56 RP: the parser rework is pretty straightforward, really. just changes it to create objects for each thing we parse, and evaluate the objects at the end, in order, rather than operating immediately. intermediate form opens up caching possibilities and the like Nov 12 21:02:12 i think we can still move more into the ast though, i had some changes in that regard in my local work area from when i was toying with pyparsing Nov 12 21:02:27 * kergoth needs more caffeine Nov 12 21:02:28 kergoth: sounds very cool. Do we have performance numbers? Nov 12 21:02:35 good question :) Nov 12 21:03:12 RP: you got some mins to investigate the troubles your patch causes, or should i send it to the mailing list? :) Nov 12 21:03:25 mlip: Its on my todo list now Nov 12 21:03:30 oke Nov 12 21:03:39 mlip: As you'll gather from the above, I'm just doing four things in parallel ;-) Nov 12 21:04:01 mlip: does bitbake busybox -c devshell work for you? Nov 12 21:04:27 RP: i see it's not the best time to ask for help now; I would do it myself, but actually I can't think of what brakes there Nov 12 21:04:48 mlip: Its something to do with the point the data is expanded Nov 12 21:05:09 mlip: How does menuconfig break btw? Nov 12 21:05:51 RP: instantly closing menuconfig Nov 12 21:06:03 RP: shell is opening but returning instantly Nov 12 21:06:04 mlip: ah, I see that too :/ Nov 12 21:06:10 RP removing the env var did not help, the root cause is that config.sh has too many issues Nov 12 21:06:13 * eFfeM is bailing out Nov 12 21:06:29 eFfeM: The perl stuff needs a lot of work :( Nov 12 21:06:38 RP: the devicekit breakage seems to be caused by libudev.pc not being staged Nov 12 21:06:39 I know :-( Nov 12 21:06:52 RP and I don't understand too much about perl and the whole system Nov 12 21:06:55 RP: busybox opens an empty shell in do_devshell; since I dont use it usually I dont know if that is intended Nov 12 21:07:05 hi, does oe_libinstall have a known problem with the recent migration to a different staging method? shr people had an issue with libtcl8.4 that was staged as libtcl8.4.so(and not libtcl8.4.so.0.some_version) without any symlinks Nov 12 21:07:13 basically one seems to need a lot of knowledge on the perl makefile maker Nov 12 21:07:21 mlip: That is the correct behaviour Nov 12 21:07:42 mlip: The trouble is either one works, or the other but not both :} Nov 12 21:08:08 mlip: and the reason it instantly opens/closes is due to the code changing the variables inside the shell function Nov 12 21:08:48 RP: i thought because ${} expands it in a way so bash --rcfile searches for $TERMRCFILE, which is not defined Nov 12 21:09:16 ${} is expanded by bitbake, before it ever gets to the shell, unless its undefined in bitbake/oe Nov 12 21:09:19 RP: instead menuconfig defines $SHELLCMDS, which would work without curly braces because of some earlies/later expanson behaviour Nov 12 21:09:20 RP: if you want, i can see about cleaning up the parser rework patchset a bit and get it onto 1.10. up to you Nov 12 21:09:41 kergoth: I'd really like to take a look and understand it if possible Nov 12 21:09:50 * kergoth nods, figured :) Nov 12 21:10:32 RP: shouldnt menuconfig work, if a $TERMRCFILE is created by do_menuconfig ? Nov 12 21:12:25 mlip: I'm going to just revert that patch, its wrong Nov 12 21:12:50 RP: just killed a few of my commits on 1.8-parser-rework, I fixed some of them on my parser-rework branch on master, better to cherry pick those from there instead if they're worth grabbing Nov 12 21:13:09 not sure which branch will be easier to pull onto 1.10 Nov 12 21:13:28 perhaps the master based one, actually.... given it has a fix for the bbextendclass Nov 12 21:13:32 * kergoth scratches head Nov 12 21:13:48 this is what happens when i hack at something a while and then forget all about it for months.. Nov 12 21:13:57 RP: oke, also fine for me ;) Nov 12 21:14:10 kergoth: I know some of the ones in the branch like dropping __functions__ crept in elsewhere Nov 12 21:16:08 RP: I'd suggest doing a rebase -i from kergoth/parser-rework onto 1.10, and see what the damage is. Nov 12 21:16:13 If I have to type org.openembedded.dev much more I will go mad :/ Nov 12 21:16:17 agreed Nov 12 21:16:53 03Richard Purdie  07org.openembedded.dev * r9aa17d34e2 10openembedded.git/conf/bitbake.conf: Nov 12 21:16:53 Revert "bitbake.conf: Fix TERMCMD variables (from Poky)" Nov 12 21:16:53 This reverts commit 2f982a8b546846d705bdb6e75a58692e224b6726. Nov 12 21:16:53 mlip: reverted, sorry about that Nov 12 21:17:34 03Frans Meulenbroeks  07org.openembedded.dev * r63142c1df8 10openembedded.git/recipes/xmltv/ (files/Makefile.PL.patch xmltv_0.5.56.bb): xmltv: added some grabbers Nov 12 21:17:39 RP: np, sorry that i couldn't help, but didn't have anything to do with bitbake variable expansion till now Nov 12 21:17:44 03Frans Meulenbroeks  07org.openembedded.dev * rb0c5d464e1 10openembedded.git/: Merge branch 'org.openembedded.dev' of git.openembedded.org:openembedded into org.openembedded.dev Nov 12 21:19:16 hi, where oe_soinstall comes from? Nov 12 21:19:23 is it part of the new staging ? Nov 12 21:19:36 new staging = the rfc that there was on mailing list Nov 12 21:21:29 GNUtoo: it's in base.bbclass Nov 12 21:21:52 pb__, ok thanks Nov 12 21:22:02 I don't think it's new. Nov 12 21:22:10 the problem is that oe_libinstall installed libtcl8.4 to libtcl8.4.so Nov 12 21:22:14 oe_soinstall is extremely old Nov 12 21:22:18 and didn't made any symlinks Nov 12 21:22:19 ok Nov 12 21:22:21 so is oe_libinstall Nov 12 21:22:36 its likely you should just be using make install in the buildsystem of the app, rather than manually installing files, nowadays Nov 12 21:22:38 GNUtoo: oe_soinstall or oe_libinstall? they are different and I don't think they use each other. Nov 12 21:22:52 but, yeah, what kergoth said Nov 12 21:22:59 so I looked into the devshell script and found oe_soinstall there too...so I wondered what was soinstall Nov 12 21:23:04 fwiw, I just tried a tcl8.4 build and it seemed to stage fine for me. Nov 12 21:23:15 lrwxrwxrwx 1 pb pb 14 2009-11-12 21:10 ./armv5te-oe-linux-uclibceabi/lib/libtcl8.4.so -> libtcl8.4.so.0 Nov 12 21:23:15 -rwxr-xr-x 1 pb pb 577735 2009-11-12 21:10 ./armv5te-oe-linux-uclibceabi/lib/libtcl8.4.so.0 Nov 12 21:23:23 ok Nov 12 21:23:28 mmm Nov 12 21:23:40 I'm on shr-merge branch Nov 12 21:24:11 yeah, it could be different there Nov 12 21:25:07 I'll look at oe_runmake install....but does it installs the libs too...is there a document on that? Nov 12 21:25:17 the manual says oe_libinstall Nov 12 21:25:29 I bet they get installed with oe_runmake install Nov 12 21:25:33 I'll try Nov 12 21:25:43 but some sed is needed in some script Nov 12 21:25:58 sed -i "s+${WORKDIR}+${STAGING_INCDIR}+g" tclConfig.sh Nov 12 21:26:04 and similar things Nov 12 21:26:48 you don't need to override do_stage or mess with do_install to do those manglings, thats what the sysroot funcs var is for, no? Nov 12 21:27:35 ok I'll look thanks a lot Nov 12 21:27:59 kergoth: right Nov 12 21:29:21 sysroot funcs = oe_runmake install DESTDIR=${D} ? Nov 12 21:29:50 I'll try Nov 12 21:31:16 RP: opkg-native is also failing on a clean build. seems it can't find curl/curl.h which is no longer staged by curl-native Nov 12 21:31:48 sakoman: Are you up to date with OE.dev? Nov 12 21:32:10 sakoman: I supposedly fixed something that sounds very similar to that already :/ Nov 12 21:33:01 RP: heh, at the time I pulled I got the commit just prior to that one :-) Nov 12 21:33:07 sorry for the false alarm! Nov 12 21:33:15 I'll do another pull Nov 12 21:33:15 sakoman: fix for icu pushed Nov 12 21:33:20 03Richard Purdie  07org.openembedded.dev * r25b75bb6dc 10openembedded.git/recipes/icu/icu_3.6.bb: Nov 12 21:33:20 icu: Add missing SYSROOT_DESTDIR to sysroot processing function Nov 12 21:33:20 Signed-off-by: Richard Purdie Nov 12 21:34:10 RP: I understand what you are saying about typing org.openembedded.dev :-) Nov 12 21:34:20 (as I type it a few more times) Nov 12 21:35:05 sakoman: For devicekit, looks like udev's .pc file has gone missing Nov 12 21:35:29 RP: yup that's what I just discovered too Nov 12 21:35:50 moving it there by hand fixes the issue Nov 12 21:35:59 so that much is verified :-) Nov 12 21:37:45 sakoman: why its missing is an interesting question :/ Nov 12 21:38:31 indeed. I don't understand all this new staging stuff :-( Nov 12 21:40:01 sakoman: Hmm :/. Is it anything in particular or the whole thing? Nov 12 21:40:56 wow, i think this is the first time I've ever used the --onto argument of rebase Nov 12 21:41:20 the whole thing :-) I've been trying to get some client work done so I haven't been able to study it, just dealing with the fallout of the change :-0 Nov 12 21:41:44 anybody knows how stable 2.6.31 kernel is for beagleboard right now? don't know whether i should try a 2.6.31 or 2.6.29 real-time kernel Nov 12 21:41:59 conceptually its pretty straightforward. instead of separate do_stage installs, we use the output of do_install, with a var you can use to supply mangling functions Nov 12 21:42:01 sakoman: ok, so this is udev having exec_prefix="" in the recipe Nov 12 21:42:31 sakoman: The core code therefore dutifully ignored the prefix directory Nov 12 21:43:47 kergoth: thanks for the nice simple explanation! Nov 12 21:44:29 the summary of the discussion from oedem covers it too, i think :) Nov 12 21:46:18 kergoth: I haven't looked at the ML lately so I missed that. Just did a pull this AM, then said "Oh sh*t!" Nov 12 21:48:00 sakoman: The benefits of this long term are wonderful but the transition is hard :/ Nov 12 21:48:11 sakoman, I should have warned you not to pull Nov 12 21:48:19 the only thing building for me atm is stable Nov 12 21:48:31 * Crofton has a weird python-native crash Nov 12 21:48:38 Crofton: :/ Nov 12 21:49:11 Crofton: Anything I can help with? Nov 12 21:49:19 but, I do have an fpga bit file to dl Nov 12 21:49:24 tomorrow Nov 12 21:49:30 I need to get this unplugged Nov 12 21:49:45 but, I can't start on it today Nov 12 21:49:50 today has been a mess Nov 12 21:49:55 flooded basement in rental Nov 12 21:50:32 Crofton: ok, np Nov 12 21:51:31 I've been hoping it goes away Nov 12 21:51:42 Crofton: ouch, flooded basement trumps broken build! Nov 12 21:51:43 So OE uses the version of udev with the really broken path handling Nov 12 21:51:46 let me paste what I see in vat/log/messages Nov 12 21:52:27 RP: wonder if we should think about squashing some of zecke's commits. one commit moves a couple things over, hen another moves a couple others... seems worth cleaning up, more about the development process than actual independent steps Nov 12 21:52:32 * kergoth yawns Nov 12 21:52:52 RP, this is the error Nov 12 21:52:56 | /bin/sh: line 1: 19298 Segmentation fault LD_LIBRARY_PATH=/home/balister/oe/tmp/work/x86_64-linux/python-native-2.6.1-ml8.2/Python-2.6.1: CC='ccache gcc -pthread' LDSHARED='ccache gcc -pthread -shared' OPT='-DNDEBUG -g -O3 -Wall -Wstrict-prototypes' ./python -E ./setup.py -q build Nov 12 21:53:05 and this is in var/log/messages Nov 12 21:53:14 Nov 12 14:53:11 elephant kernel: python[19298]: segfault at c8 ip 00007fc84b383aec sp 00007fff0780be40 error 4 in libncurses.so.5.0.4[7fc84b352000+4d000] Nov 12 21:53:22 yikes Nov 12 21:53:26 that's not pleasant Nov 12 21:53:31 no Nov 12 21:55:44 Crofton|work: hope its not loading mixed stuff i.e. from your default installation and from the oe build Nov 12 21:56:00 Crofton|work: not good at all :/ Nov 12 21:56:05 that is what I am worried about Nov 12 21:56:25 I think the python-native stuff has an issue Nov 12 21:56:30 I am on F11 Nov 12 21:56:40 XorA|gone, says it works for him :( Nov 12 21:56:57 Crofton|work: python-native works here too Nov 12 21:57:25 RP: your ICU and curl-native fixes worked fine for me. Many thanks! Nov 12 21:58:44 Crofton|work: may be you should try unsetting LD_LIBRARY_PATH Nov 12 21:58:47 RP: founs smthg.. Nov 12 21:59:06 #define LT_OBJDIR ".libs/" Nov 12 21:59:35 btw dbus-glibc srages correctly Nov 12 21:59:46 eh..dbus-glib Nov 12 21:59:55 ant__: Any context on that such as which file its in? Nov 12 22:00:03 config.h :D Nov 12 22:00:14 ant__: Its dbus-native I'm worried about Nov 12 22:00:24 hmm autofs works much better with tcp for mountproto Nov 12 22:00:34 RP: seems ok..-DDBUS_DAEMONDIR=\"/oe/build/tmp/staging/i686-linux/usr/bin\ Nov 12 22:00:47 problem is the .libs suffix... Nov 12 22:01:27 ant__: Is /oe/build/tmp/staging/i686-linux/usr/bin/dbus-daemon a binary or a script? Nov 12 22:01:59 script Nov 12 22:02:10 # dbus-daemon - temporary wrapper script for .libs/dbus-daemon Nov 12 22:02:10 # Generated by ltmain.sh (GNU libtool) 2.2.6 Nov 12 22:02:51 ant__: hmm, it is here too. You're running rm_work? Nov 12 22:03:06 yea, inheriting Nov 12 22:03:29 ant__: why are you trying to run the dbus-daemon from staging anyway? that doesn't sound like a terribly wholesome thing to do. Nov 12 22:03:59 pb__: Its ok in this case Nov 12 22:04:03 but, clearly, it is bogus for it to be installing the wrapper script there. Nov 12 22:04:17 it's 755 even Nov 12 22:04:28 ant__: right, the difference is that I'm not using rm_work so my wrapper script works Nov 12 22:04:35 omg Nov 12 22:05:54 * kgilmer wonders if it is ok to have TMPDIR not under OEROOT Nov 12 22:06:05 and its broken in poky too Nov 12 22:06:26 kgilmer: yeah, you can put TMPDIR whereever you like Nov 12 22:06:51 barring a couple of curious restrictions like the symlinks thing, and not having any spaces in the name Nov 12 22:07:08 sweet...that's going to make my life a lot easier. freakin eclipse svn plugin dies after trying to scan TMPDIR and I cannot turn it off. Nov 12 22:07:17 heh Nov 12 22:07:53 dbus-native has a funky install function Nov 12 22:08:07 er, stage function Nov 12 22:08:29 you mean /usr/bin/install -c dbus-daemon /oe/build/tmp/staging/i686-linux/usr/bin ? Nov 12 22:08:50 ah Nov 12 22:10:08 ant__: oe_runmake install and autotools_stage_all Nov 12 22:10:25 ah, yes, as I said before Nov 12 22:10:40 autotools_stage_all was the last suspect... Nov 12 22:18:01 Crofton: obviously OE loves me more :-) Nov 12 22:19:02 [dp@xora-build ~]$ getenforce Nov 12 22:19:02 Enforcing Nov 12 22:19:11 so I dont even have SELinux in permissive mode Nov 12 22:20:23 03Richard Purdie  07org.openembedded.dev * rf2d481cd78 10openembedded.git/recipes/udev/udev_141.bb: Nov 12 22:20:23 udev-141: Fix staging of .pc file due to exec_prefix being '' Nov 12 22:20:23 Signed-off-by: Richard Purdie Nov 12 22:23:24 XorA|gone, thanks for checking that Nov 12 22:23:27 saves me trying Nov 12 22:23:41 managed to set the right thing to get qemu to run even Nov 12 22:24:05 * kergoth ponders Nov 12 22:24:12 Crofton|work: its a pretty bog standard F11 install Nov 12 22:24:19 mine also Nov 12 22:26:42 dbus is crazy, it marks dbus-daemon as noinst, then installs it manually Nov 12 22:26:51 and incorrectly Nov 12 22:27:43 sakoman: devicekit should now work Nov 12 22:27:49 03Richard Purdie  07org.openembedded.dev * r12dd61c701 10openembedded.git/recipes/udev/udev_141.bb: Nov 12 22:27:49 udev-141: Fix staging of include files due to exec_prefix being '' Nov 12 22:27:49 Signed-off-by: Richard Purdie Nov 12 22:28:28 RP: ah yeah, funky. I guess this is what fix-install-daemon.patch is in aid of. Nov 12 22:28:55 RP: still fails w/out rm_work Nov 12 22:30:04 this .libs dir resides only in workdir, not in staging Nov 12 22:30:13 (would be strange...) Nov 12 22:31:17 ~seen zecke Nov 12 22:31:19 zecke was last seen on IRC in channel #oe, 12h 36m 30s ago, saying: 'anyway, I have wasted too much time on OE the last two days... i'm really frustrated by all this breakage'. Nov 12 22:31:28 heh Nov 12 22:31:35 pb__: yes, should be applied to the native version too Nov 12 22:31:54 RP: ah, look at dbus-glib-0.76/tools/run-with-tmp-session-bus.sh Nov 12 22:33:26 ant__: insane Nov 12 22:35:12 03Richard Purdie  07org.openembedded.dev * r2c3bb2bf45 10openembedded.git/recipes/dbus/dbus-native_1.2.1.bb: Nov 12 22:35:12 dbus-native: The native version should have the install fix applied else a wrapper is installed into staging which fails with rm_work Nov 12 22:35:12 Signed-off-by: Richard Purdie Nov 12 22:35:14 ant__: Try that ^^^ Nov 12 22:42:52 RP: ok, compiled, thx :) I'm rebuilding all from scratch now... Nov 12 22:42:56 he NOTE: Legacy staging mode for /oe/org.openembedded.dev/recipes/dbus/dbus-glib-native_0.76.bb Nov 12 22:43:21 ant__: You will see that a lot. One step at a time Nov 12 22:43:29 sakoman_: devicekit should be fixed now too Nov 12 22:43:32 argh ;) Nov 12 22:44:48 btw did anyone fix sqlite3 wanting tcl-native as dependency? Nov 12 22:45:05 otherwise will be next stop :=) Nov 12 22:45:46 bluelightning: ^?^ ? Nov 12 22:46:07 haven't looked at it... Nov 12 22:46:19 seems smthg wrong in the recipe Nov 12 22:46:30 gentoo has tcl as useflag iirc Nov 12 22:46:42 should be possible to exclude-it Nov 12 22:46:45 kergoth: Did you submit a GFP for this channel? Nov 12 22:47:12 kergoth: er, a GRF - the freenode channel registration thing Nov 12 22:48:30 RP: ok, survived to rm_work...now real testing from vacuum Nov 12 22:49:27 ant__: Note that this problem was probably not introduced by the staging changes Nov 12 22:51:36 RP: good question Nov 12 22:52:06 fwiw I use since long time rm_work Nov 12 22:52:24 just updated bitbake recently...from svn to 1.8.18 Nov 12 22:57:26 ant__: perhaps the build order just got more efficient Nov 12 22:57:46 well, perhaps... Nov 12 22:58:22 as I briefly discussed with kergoth, old version had a couple of speed-improvements over 1.8.18 Nov 12 22:58:43 some caches Nov 12 22:59:23 ant__: you're saying your old svn version was faster than 1.8.18? Nov 12 22:59:24 would it be possible to have the two last two-three commits in svn for 1.8.19 ? Nov 12 22:59:27 yes Nov 12 22:59:39 it took 45 minutes for me to delete TMPDIR. oof Nov 12 22:59:40 NOTE: Resolving any missing task queue dependencies Nov 12 22:59:40 NOTE: Preparing runqueue Nov 12 22:59:42 here Nov 12 23:00:04 ant__: svn trunk or svn 1.8 ? Nov 12 23:00:22 ant__: I think the speedups you speak of are in 1.10 Nov 12 23:00:32 kgilmer: you want ext4 ;-) Nov 12 23:00:46 RP, yep rebooting now :) Nov 12 23:00:50 brb Nov 12 23:01:09 is it safe to move to 1.8.10 now? Nov 12 23:01:17 er..1.10 Nov 12 23:01:23 ant__: safeish Nov 12 23:01:28 np Nov 12 23:01:31 ant__: The real fun is yet to come Nov 12 23:02:04 ok, let's have a couple of builds with 1.8.18 at first... Nov 12 23:02:48 ant__: I'm using what approximates 1.10 in poky Nov 12 23:02:52 I'm still in the recompiling-phase of natives... Nov 12 23:03:02 ah, I see Nov 12 23:04:57 ant__: allegedly tcl is obligatory (at build time) for recent versions of sqlite3 Nov 12 23:05:32 3.6.20 apparently requires it for generating header files or some such, even when configured --disable-tcl. Nov 12 23:05:46 RP: http://gentoo-portage.com/dev-db/sqlite/Dep#ptabs Nov 12 23:05:57 all are optional Nov 12 23:06:10 there must be some patch there... Nov 12 23:06:28 are those run time dependencies or build time dependencies? Nov 12 23:06:35 runtime Nov 12 23:06:37 ah Nov 12 23:06:39 it says "run time" on that page but I am not all that au fait with gentoo speak Nov 12 23:06:50 right Nov 12 23:07:10 so, if "run time" means run time, that doesn't contradict the situation in oe Nov 12 23:07:26 let me see the recipe... Nov 12 23:07:38 I don't think the oe-built sqlite3 packages have ever needed tcl at run time, they just need tcl-native as a build dependency Nov 12 23:07:41 site is nice but unofficial :) Nov 12 23:07:50 re Nov 12 23:08:17 RDEPEND="icu? ( dev-libs/icu ) Nov 12 23:08:17 readline? ( sys-libs/readline ) Nov 12 23:08:17 tcl? ( dev-lang/tcl )" Nov 12 23:08:17 DEPEND="${RDEPEND} Nov 12 23:08:17 doc? ( app-arch/unzip ) Nov 12 23:08:27 I think is clear enough. Nov 12 23:09:23 perhaps debian is even more atomized...dunno Nov 12 23:09:51 pb__: that USEFLAGS are perversely nice, isn't ? Nov 12 23:16:46 pb__: ewarn "Support for ICU is enabled only when \"tcl\" USE flag is enabled." Nov 12 23:16:58 ewarn "You must enable the tcl USE flag if you want to run the testsuite." Nov 12 23:17:15 ant__: i think just about everyone is agreed that we need something like USE Nov 12 23:17:39 I see there is a patch + some if use tcl; then Nov 12 23:17:40 append-cppflags -DSQLITE_ENABLE_RTREE Nov 12 23:18:35 well, now recompiling gcc...native are done Nov 12 23:18:43 RP: looks good Nov 12 23:20:05 sorry but can't really help with sqlite today...server is busy and only one build is allowed (or...?) Nov 12 23:36:03 RP, didn't we make the layout stuff go away? Nov 12 23:42:39 'nite all Nov 13 00:51:04 trying to do my first build, fails on eglibc-2.10-r7.5 (checking sysdep dirs... configure: error: The geode subspecies of i486 is not supported) is it detecting the i486 config of my gentoo host? Nov 13 00:51:57 rednul read the oe-dev ml for the problem Nov 13 00:52:17 I removed the patches to fixing it because it was not the proper solution Nov 13 00:52:36 this one? http://sourceware.org/ml/crossgcc/2006-09/msg00049.html Nov 13 00:52:57 is this the openembedded-dev mailiglist? Nov 13 00:53:02 I rather guess not Nov 13 00:53:14 oops sorry :) Nov 13 01:17:04 re kergoth Nov 13 01:54:22 03Henning Heinold  07org.openembedded.dev * r14a73529b5 10openembedded.git/recipes/pingus/ (files/boost.patch files/cheader.patch pingus_0.7.2.bb): Nov 13 01:54:22 pingus: make it buildable again Nov 13 01:54:22 * add patches for finding right boost lib and fixing c-headers Nov 13 01:54:22 * pingus has no install target so make do_stage empty Nov 13 01:54:22 * bump PR **** ENDING LOGGING AT Fri Nov 13 02:59:57 2009