**** BEGIN LOGGING AT Mon May 14 02:59:59 2012 May 14 07:09:23 good morning May 14 07:10:21 i pulled a clean oe-core and did "bitbake meta-toolchain" but get perpetually "IndexError: list index out of range" ? May 14 07:11:51 I need to create a list of packages and licenses for an oe-classic build. Tried adjusting the create_manifest.sh-script from arago without success. Any other ways to do this? May 14 07:48:26 morning May 14 07:56:23 it worked great, once I told it where to find opkg-cl May 14 09:22:46 morning all May 14 09:34:58 good morning May 14 10:07:16 I compile a rootfs and an sdk from oe-classic. Then we make our own applications which link dynamically to 3d-party libraries. Should/must I provide any Legal/License information for the rootfs/sdk part? May 14 10:19:15 gm May 14 10:21:18 I am often getting this error, just after pseudodone stage. It is a config problem, I just forgot to what it is related... Anyone? May 14 10:21:20 ERROR: Nothing PROVIDES 'git-native' May 14 10:21:21 ERROR: Required build target 'u-boot' has no buildable providers. May 14 10:21:23 Missing or unbuildable dependency chain was: ['u-boot', 'git-native'] May 14 10:23:02 ^^^ never mind, I remember again. I had "git" somewhere in my excludes of an rsync command :) May 14 10:59:16 gm mickey_office, all May 14 11:23:19 Is there a common way to attribute OpenEmbedded when I have used it to create my rootfs&SDK? May 14 11:29:48 hi likewise May 14 11:30:15 tasslehoff: no, creativity is required there :) May 14 11:31:34 tasslehoff: you may have requirement regarding the licenses already, some recipes might have requirements regarding mentioning origin. (I think that is what you mean?) May 14 11:38:45 likewise: not really. our own application links dynamically with some lgpl-libraries, and I've made a statement about that. I just wonder if I need to say somewhere that we compile rootfs and sdk with OpenEmbedded. May 14 11:40:26 tasslehoff: I do not think OE has requirements, but the recipes might have requirements regarding attribution. May 14 11:45:21 likewise: ok. May 14 11:48:29 while using UBIFS .... I m getting this error.... http://pastebin.mozilla.org/1638044 May 14 12:21:55 snkt: usually is due to a bug in CONFIG_MTD_NAND_VERIFY_WRITE and subpages May 14 12:22:41 http://www.linux-mtd.infradead.org/faq/ubi.html#L_subpage_verify_fail May 14 12:23:06 for a possible patch: http://lists.infradead.org/pipermail/linux-mtd/2012-April/040737.html May 14 12:23:45 ^ I actualized it for kernel 3.2 but still need to disable the verify or use -O and force full pages May 14 12:24:09 so it seems this patch alone does little :/ May 14 12:27:56 ant_work, i m using linux 2.6.32 May 14 12:28:44 then maybe you'll have better luck May 14 12:30:20 snkt: btw this bug happens also on devices with 512 bytes page and 256 bytes subpage May 14 12:31:30 snkt: so in my case I had to use - O 512 for ubiformat and ubiattach but this is far fromideal... May 14 12:32:09 ant_work, I have 64MB Nand flash May 14 12:32:45 http://paste.debian.net/168731/ May 14 12:32:45 snkt: then is likely512/256 as well May 14 12:33:08 what does make kernel always load initramfs May 14 12:34:30 ant_work, where do configure page and subpage May 14 12:34:53 snkt: mtdinfo -a ubinfo -a May 14 12:35:03 snkt: are defined in kernel driver May 14 12:47:43 ant_work, I m following this step..... and you will notice that while using "rootfs" partition I m getting error ... http://pastebin.mozilla.org/1638072 May 14 12:54:56 you shouldn't use nandwrite to flash ubi volumes May 14 12:55:05 just use ubiformat May 14 12:55:55 or follow the howto-here http://free-electrons.com/blog/creating-flashing-ubi-ubifs-images/ May 14 13:01:49 ant_work, really? I often use nandwrite to flash ubi(not ubifs) volumes May 14 13:05:29 hi, I've that issue with libav: work/armv6-vfp-oe-linux-gnueabi/libav-0.7.4-r4.1/git/libavcodec/arm/ac3dsp_armv6.S:40: Error: selected processor does not support ARM mode `movw r8,#0x1fe0' May 14 13:05:55 I've grepped last ml messages and didn't found the problem in them May 14 13:06:43 I'll try to look what instruction set is movw or in what arm version it's supported May 14 13:12:46 hmmm it says armv6 T2 May 14 13:12:53 what is T2, the new thumb? May 14 13:13:35 right it's that May 14 13:14:02 thumb2 May 14 13:22:12 denisATeukrea: Maybe you restricted your toolchain to thumb1 May 14 13:22:31 in this case there is no encoding for movw May 14 13:22:38 yes I know May 14 13:22:49 but I think it's a bug in the lib May 14 13:22:57 they should do that and they didn't: May 14 13:23:14 #if HAVE_ARMV6T2 May 14 13:23:15 GNUtoo-desktop: http://www.linux-mtd.infradead.org/faq/ubi.html#L_flash_img May 14 13:24:17 GNUtoo-desktop: I just started recently testing UBI and am already a bit scared :/ May 14 13:24:22 ant_work, ahh ok thanks a lot May 14 13:24:34 I use ubi since ages May 14 13:24:55 have you seen the subpages-verify patch I linked above? May 14 13:25:06 no May 14 13:25:18 I looked only at the last link you pasted May 14 13:25:28 I can send you the version for 3.2 but doesn't seem to work :/ May 14 13:25:44 http://lists.infradead.org/pipermail/linux-mtd/2012-April/040737.html May 14 13:25:44 ahh that one May 14 13:25:49 nice May 14 13:26:18 denisATeukrea: yep, probably. So you're using -march=armv6 -mthumb wich restricts it to thumb1 May 14 13:26:21 but we dropped the support for the htcdream in SHR(the htcdream had that problem) May 14 13:26:36 kenws, ok but I'm not sure my armv6 has thumb2 May 14 13:26:50 Features : swp half fastmult vfp edsp java tls May 14 13:26:51 there is the -march=armv6t2 option that would allow the toolchain to generate thumb2 instructions May 14 13:31:05 so I'm tempted to imlpement the instructions for making it work with thumb1 May 14 13:42:00 ah sorry it seem that chromium doesn't really depend on libav May 14 13:42:12 so I'll simply remove the dependency..... May 14 13:42:32 it uses an internal libav if I remember well May 14 13:43:37 denisATeukrea: just out of curiosity. Could you paste the content of /proc/cpuinfo? May 14 13:44:19 I did: May 14 13:44:27 Features : swp half fastmult vfp edsp java tls May 14 13:44:56 here's the more complete version: May 14 13:44:57 http://pastie.org/private/gvjlxs7j35jyttt0jdwq May 14 13:46:29 denisATeukrea: ok, so only thumb1 :) May 14 13:46:41 in your case May 14 13:48:14 yes indeed May 14 13:49:04 hm, I don't see the movw upstream (http://git.libav.org/?p=libav.git;a=blob;f=libavcodec/arm/ac3dsp_armv6.S) May 14 13:49:54 ah, there is an older version May 14 13:50:17 yes May 14 13:51:06 denisATeukrea: http://git.libav.org/?p=libav.git;a=commitdiff;h=9a83adaf346bc5e38183427b2bd11084c11d1c8b May 14 13:51:12 that's the patch May 14 13:52:42 thanks a lot May 14 13:55:35 Does anyone know the semantic differences between =. and _prepend? May 14 13:56:00 * kenws wonders where are these operators are documented May 14 14:05:21 s/are// May 14 14:07:00 kenws: the difference is _prepend happens later in the parsing process rather than being immediate May 14 14:07:05 kenws: historically they've all been covered in the file format info in the bitbake manual, though that hasn't been kept particularly current. _prepend and _append can be used with overrides for conditional append/prepend, for one, there's no syntax for that with =. and .=. the other main difference is that _prepend/_append is lazy. they get applied at the very end of the parsing, not immediately, whereas .= and =. is immediate May 14 14:07:28 what he said :) May 14 14:09:14 bluelightning: kergoth`: thanks! well, that's interesting. I have a bbappend to the yocto kernel recipe where the FILESEXTRAPATHS gets extended to be able to find a defconfig May 14 14:09:28 while May 14 14:09:34 FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}/${MACHINE}:" May 14 14:09:36 works May 14 14:09:41 thats a bit of a special case. we aren't caring about the lazy aspect of _prepend May 14 14:09:43 FILESEXTRAPATHS =. "${THISDIR}/${PN}/${MACHINE}:" May 14 14:09:50 what we need is the immediate expansion, which you removed in switching to =. May 14 14:09:52 := May 14 14:10:01 doesn't (anymore?) May 14 14:10:06 forces immediate expansion of the variable, meaning THISDIR is used for the bbappend, not the recipe itself May 14 14:10:28 ah May 14 14:10:39 that should be documented somewhere : ) May 14 14:10:41 * kenws ducks May 14 14:10:47 it should, yes May 14 14:12:17 It looks like yoctos intel layer is also using FILESEXTRAPATHS_prepend May 14 14:12:40 I suppose that's is the way to do it in this case May 14 14:12:51 it's the recommended way, yes May 14 14:12:57 is it really necessary to add ${MACHINE} to find a defconfig? May 14 14:13:26 it isn't, no, machine is automatically searched, as file:// uris automatically check for overrides May 14 14:13:29 good catch May 14 14:13:34 aah May 14 14:13:44 all my kernels would be broken ... May 14 14:13:45 unless FILESEXTRAPATHS avoids that magic, i don't recall May 14 14:13:50 heh :) May 14 14:14:03 I'll give it a try May 14 14:14:07 btw I had to add FILES to find some shared logos May 14 14:14:18 FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:${THISDIR}/files:" May 14 14:21:53 I'm using a machine called "qemuarmv7a" which doesn't seem to be automatically searched for defconfigs May 14 14:23:29 this is the linux-yocto_3.2.bbappend: http://pastie.org/private/uz0hxoxhnlqcccsjysimq May 14 14:25:46 As soon as the FILESEXTRAPATHS_prepend statement gets removed bitbake won't find the recipes-kernel/linux/linux-yocto/qemuarmv7a/defconfig May 14 14:26:15 kenws: that's what you'd expect May 14 14:26:42 kenws: what was suggested was removing the /${MACHINE} in your line pasted above May 14 14:28:35 aah, my bad. That would work. In this case I should encode the machine into the name of the defconfig... May 14 14:29:20 kenws: that's not needed either May 14 14:29:42 kenws: it will already look in the machine subdir May 14 14:30:04 hmpf, I'm afraid I don't get it : / May 14 14:30:56 I just tried recipes-kernel/linux/linux-yocto/qemuarmv7a/defconfig May 14 14:31:15 alright May 14 14:31:25 I should have read the log more carefully May 14 14:31:37 kenws: whatever path you add to FILESEXTRAPATHS, it will look in subdirectories of that path from OVERRIDES (which includes MACHINE) and then the subdirectory itself May 14 14:32:13 kenws: if you like you can use bitbake -e | grep ^FILESPATH to see the exact search path used after everything is expanded, but you'll find it's quite long... May 14 14:34:47 bluelightning: got it. thank you! May 14 14:36:02 I was somehow thinking that path/to/recipe/$MACHINE is searched for files by default May 14 14:37:04 it is. among others May 14 14:39:45 kergoth`: Then, why would you fiddle with FILESEXTRAPATHS to let bitbake recognize a $MACHINE/defconfig? May 14 14:43:52 huh? May 14 14:43:53 kenws: that is needed because you are in a bbappend May 14 14:43:57 first, the recipe is not the bbappend May 14 14:43:59 as bluelightning says May 14 14:44:28 * bluelightning wonders for the nth time whether this ought to be automatic May 14 14:45:21 nontrivial, unless you assume the same layout across layers, or you inspect __dependsor something to find the parsed files May 14 14:45:42 if you do assume the same layout across layers, like i used to do when using COLLECTIONS, you still end up with just a massive FILESPATH, slows down file:// operations May 14 14:45:45 unfortunately :\ May 14 14:46:23 aaah, true! May 14 14:46:47 kergoth`: I guess __depends would be the logical way May 14 16:46:14 GNUtoo-desktop: pls let me know if you have any success with ubi subpages-verification May 14 16:46:46 GNUtoo-desktop: as I said, I refreshed the patch for 3.2. I'll pastebin it later. May 14 16:47:42 bbl May 14 16:59:31 hi May 14 16:59:48 how can i disable sstate from local.conf? May 14 17:01:23 or would that be a bad idea? May 14 17:01:38 you can't.. it's part of oe-core May 14 17:01:53 are you trying to stop it from being generated or "used" if already generated? May 14 17:02:52 being generated May 14 17:03:24 i had some wierd problems with ccache and thought i'd try disabling sstate, too May 14 17:04:39 can't stop it.. it and the associated hashing are part of the build May 14 17:05:46 ok May 14 17:33:18 I've got a recipe that needs to download two tar.gz files.. how do I specify the checksum of the second? May 14 17:35:33 something like SRC_URL += "http://foo;sha1=abc" should work May 14 17:36:09 thanx May 14 17:36:24 grep for examples, I'm not sure of the exact syntax May 14 17:36:38 I did and didn't find any.. but I may have not grepped correctly.. May 14 17:36:50 grep for sha256sum then May 14 17:37:35 http://foo;name=bar May 14 17:37:41 SRC_URI[md5sum.bar] = "abcd" May 14 17:37:44 also works May 14 17:38:04 yeah, that's the only one I can actually find used May 14 17:38:14 'k May 14 17:38:26 Hi! May 14 17:38:46 Lets state the following scenario: May 14 17:39:32 You are a knight in shining armor. May 14 17:39:44 format was backwards.. :0 should be SRC_URI[bar.md5sum] ;) but anyway.. bitbake DID have helpful error messages.. May 14 17:39:48 You want to attack a castle somewhere. May 14 17:39:57 second question, how do I get that to extract into the directory of the main src file? May 14 17:40:12 When you arrive to the gate of that castle you see a guard at the opening. May 14 17:40:34 The only problem to get past that guard is that you have to speak the same language as him. May 14 17:40:53 How do you solve that situation if you don't know what language he speaks? May 14 17:40:54 fray: something like do_unpack_append() { cp ${WORKDIR}/foo.patch ${S} } May 14 17:41:09 trygvis I thought there was a regular argument like subdir=.. but I couldn't get that working either May 14 17:41:12 That's my problem when I logged into my Home Entertainment System. May 14 17:41:22 I can't get it to listen. May 14 17:41:24 The guard. May 14 17:41:45 It seems to be someone there, but he won't listen. May 14 17:41:50 I hope you get my point. May 14 17:42:11 fray: might be, I'm new to all of this May 14 17:42:24 this isn't normal stuff... and I've not had to do it myself either May 14 17:42:33 Umeaboy: listen to someone else talking to the guard May 14 17:43:10 By sniffing? May 14 17:43:18 yes May 14 17:43:20 I think that do_unpack_append trick is quite normal May 14 17:43:26 Umeaboy: or proxying May 14 17:43:36 Isn't that what tcpdump is for? May 14 17:43:55 yes May 14 17:44:02 though I'd suggest wireshark May 14 17:44:14 I used tcpdump 192.168.0.101, but no luck. May 14 17:44:29 I even tried with -i any option. May 14 17:44:47 I know that this is the right adress. May 14 17:44:56 I can login to it. May 14 17:45:16 Just getting the guard to answer me back is the problem. May 14 17:45:16 figured it out.. it's "subdir=...." May 14 17:45:21 I must have done subdirs= the first itme May 14 17:45:33 ah May 14 17:47:21 All I get is a blank row, but verbose mode is on thou, May 14 17:47:38 So it's like I'm talking to a deaf guard. May 14 17:55:38 hollisb: Hmmmmmmmmmmmmmmmm. wireshark doesn't get anything. May 14 17:55:44 I use eth0. May 14 17:56:14 I know that to connect to since I checked with nmap. May 14 17:56:29 what to connect to I mean. May 14 17:56:48 do you have a switch that isn't promiscuous? do you have an ethernet port that isn't.. etc.. May 14 17:57:07 you need to see whats on the wire to intercept it, if something is keeping it off your wire then wireshark won't help May 14 18:01:37 fray: OK. Do you have a suggestion? May 14 18:04:11 make sure you have a hub, or put your machine into briding mode (w/ two ethernet ports).. and pass the data through your machine May 14 18:06:10 fray: I have a Ethernet->USB-card. Will that suffice ? May 14 18:06:40 as long as Linux can brige it you can do it that way.. you just have to be able to intercept all of the traffic May 14 18:06:50 OK. May 14 18:39:59 ok, i tried a bitbake core-image-minimal just to see if i miss something, it complained about LICENSE and COPYRIGHT.MIT which i assume is "usual", this leaded to -->Attempted 1397 tasks of which 1395 didn't need to be rerun and all succeeded. is this somewhat regular ? May 14 18:40:36 sounds about right.. it's using the pre-built (sstate) cache files May 14 18:40:58 if you want to invalidate the cache files.. either clear the sstate-cache directory (this will make you start over more or less).. or bitbake -c cleansstate May 14 18:41:22 -c clean only cleans the working directory, cleansstate cleans both working and sstate-cache, cleanall cleans those two and also the downloads directory May 14 18:44:19 fray, why do i have to clean anything on a brand-new build ? May 14 18:44:40 you don't.. May 14 18:44:49 it will re-use any files you've previously built and shared.. May 14 18:44:59 I through your were having hte issue where things you expected to build were not May 14 18:46:55 fray, i was probably a bit unclean in my question..........this is the very first start of oe-core regarding to the wiki, just pulled, sourced env and did a bitbake. i assume those files about licensing should be in place since i pulled the oe-core. May 14 18:48:04 if the package uses an unusual license, a warning will be generated that there is no standard license file for the pacakge.. May 14 18:48:09 it's just a warning adn not something that matters May 14 18:50:27 fray, but if there is something delivered, it should contain the appropirate licenses, otherwise it would be illegal. May 14 18:50:41 you are misunderstanding the warning May 14 18:50:55 meta/files/common-licenses May 14 18:51:04 this is a short-cut to many of the common license used in the project.. May 14 18:51:18 but any individual recipe may or may not have additional or unique license requirements May 14 18:51:51 the license for a given recipe is located in the sources for that recipes, and the recipe author has the requirement to best identify the license type and file representative of that package.. May 14 18:52:00 if you find a recipe where that is not true, it is an error.. May 14 18:52:33 the system attempts to map the files in common-licneses to the ones in each recipe processed.. if a common-license is not there, a warning is generated to make sure you are aware that recipe has a unique license and should be further reviewed May 14 18:53:14 if you are in a commercial setting, really your organization should have an IP disclosure process that revolves around inspecting the sources for the packages you are redistributing.. the rest is just trying to be helpful, but is not intended as the ultimate license authority May 14 18:53:26 ok, actually i don't not care that much about some missing text but it ends with 2 tasks need to rerun which will fail again and again. May 14 18:53:39 what is the error for those tasks? May 14 18:54:51 tmp-eglibc/work///temp/log. is the best place for you to look at the error in context.. May 14 18:55:36 there is no error, just 2 warnings about the mentioned licenses but it ends with 1397 tasks which 1395 did'nt need to re-run. this states clearly 2 need a re-run and will fail again. May 14 18:55:57 you need to show me the exact messages May 14 18:56:03 if there is no error, then it worked fine.. May 14 18:56:17 there are generally tasks that need to be rerun each time for various validations, but they are -very- fast tasks.. May 14 18:56:23 only your log can show if it's expected or not May 14 18:56:28 if it worked fine, it should say so :P May 14 18:56:44 it likely did May 14 18:57:31 the last message is telling you how many tasks were evaluated and how many of those tasks it had to build.. May 14 18:57:56 the message says "I was going to build 1397, but I didn't have to build 1395 because they were previously built, i.e. the did not have to be re-run" May 14 18:58:15 so you ran 2 tasks.. I'm guess both are the filesystem tasks, because they always have to be run and intentionally don't cache May 14 19:00:59 ok, for now i will assume that was successful. i used the minimal-console for a dry-run before i will go further. the goal is to build for an older hx4700, which leads me to the next (and last) question : do additional layers (in this case meta-handheld) need to be beside or below oe-core ? May 14 19:02:14 up to you and your configuration.. I add layers -after- meta [oe-core] in my bblayers.conf file.. I want to override meta.. May 14 19:04:49 nah, i meant the location on hdd. i pulled oe-core to /usr/src/oe/oe-core is it assumed to have now /usr/src/oe/meta-handheld OR /usr/src/oe/oe-core/meta-handheld ? May 14 19:05:23 you can put it anywhere.. but I put them parallel to "meta" May 14 19:05:37 i.e., I have meta, meta-selinux, meta-yocto, etc all int he same directory May 14 19:05:45 but you don't have to May 14 19:09:05 ok, will check the layers.conf to see if they are somewhat structured to my understanding ;) May 14 19:10:15 it's not really structured.. just a list of layers.. May 14 19:10:31 BBLAYERS ?= " \ May 14 19:10:31 /home/mhatle/git/oss/oe-core/meta \ May 14 19:10:31 /home/mhatle/wr-tools-profile \ May 14 19:10:31 " May 14 19:10:36 just one example.. May 14 19:10:46 in this case I have two layers.. one is outside of my oe-core directory May 14 19:13:09 i see ( assumed something like that) so how do i know what possibilites bitbake has, where i guess it changes on existing layers. e.g. having meta-handheld leads to some bitbake minimal-bootstrap built for hx4700 where hx4700 is somewhere configured. May 14 19:14:31 bitbake just combines the layers into one global name space.. you can either look at the file structure of all of your layers manually or use bitbake -s (show versions) May 14 19:16:46 fray: I couldn't figure out if the ordering of BBLAYERS actually had any significance so I made a BBFILE_PRIORITY_foo = "20" in my layers.conf May 14 19:17:40 it does.. May 14 19:17:51 BBLAYERS is the order the items are evaluated into bitbake.. May 14 19:18:01 the file priority is in case of conflict who has higher priority.. May 14 19:18:14 (ya, still confusing, but thats the best way I can explain it.. I'm not really sure about all of the little corner cases) May 14 19:22:13 bah, on that amount of configs, where do i save time compared to lfs ;) May 14 19:58:09 * dottedmag wonders what is the difference between pseudo tool OE uses and fakeroot Debian does. Looks the same from the description. May 14 20:06:40 pseudo has a number of advantages over fakeroot.. May 14 20:07:06 it can store more data and it is designed to be persisted.. fakeroot has lots of problem if you persist the data and try to reuse it later.. May 14 20:07:26 pseudo also intercepts, a lot of more APIs.. May 14 20:07:52 pseudo started out as a way to re-write fakeroot, when fakeroot kept failing in large build situations -- or when there was parallel access from numerous build elements.. May 14 20:08:34 pseudo also handles chroot as well as using an alternative passwd and group file for target consistency May 14 20:24:00 fray: oh, so no need for separate fakechroot? Cool. May 14 20:24:53 ok, much work beside that oe, i came back to configs, so far, i found mayn things just to left as is........what's this OLDEST_KERNEL value in angstromv2012.x.conf ? May 14 20:25:55 I haven't used fakeroot for about 2 years now.... but there were parts that were horribly broken for large system development... (but they worked fine on single package development) May 14 20:26:02 also performance was an issue.. May 14 20:26:37 (pseudo has changed so much since we first wrote it about 3 years ago.. it's hard for me to tell the differences these days) May 14 20:32:58 lautriv: I think that is passed into eglibc to determine the oldest kernel it needs to support May 14 20:34:13 yes, i believe you are correct.. May 14 20:34:22 bluelightning, ah. seems the rest would more use 3.2 ;) i thought about glibc but on a regular box this is mostly 2.6.8 instead of the found 2.6.16 May 14 20:34:27 (e)glibc changes behavior to accomodate compatibility with certain kernel versions May 14 20:34:56 wow 2.6.8? I'm not sure I have anything running a kernel that old.. May 14 20:36:32 fray, just check some binary depending on glibc, it'll mostly tell you 2.6.8 ( at least i observed that) May 14 20:37:05 the oldest machine I have, is a Fedora 12 machine.. running 2.6.32 :) May 14 20:37:52 my oldest is 3.3.3 but if i do a simple file /some/bin i see 2.6.8 very often. May 14 20:38:09 /bin/bash: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.18, stripped May 14 20:38:27 just a sample.. May 14 20:38:37 anyway.. doesn't matter.. the point is it's all about compatibility with kernel interfaces. May 14 20:39:43 something must have changed since my last check ... May 14 20:39:47 /bin/bash: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, BuildID[sha1]=0xd3cf5c19bb3b491cf60b2397d4b93584ea2991e6, stripped May 14 20:41:09 ok, back to the build, i did a bitbake -s like you suggested to see what i got so far and it tells me No such file or directory: '/usr/src/oe/meta-oe/conf/layer.conf' BUT that is in place. May 14 20:41:51 did you run the oe-core env init setup, or are you doing it manually? May 14 20:42:26 i sourced that env but wait, i guess i have to re-source after changes........ May 14 20:42:48 you only need to do it once.. as long as it stays in your environment May 14 20:43:13 but a no such... error usually means, well the file simply isn't there, or you don't have read permissions on the directory May 14 20:44:11 Hello all, maybe the wrong place to ask. May 14 20:44:30 i know, surprisingly i have it and using the same user/build-dir like the testing core-minimal. May 14 20:44:41 But I am searching for a zaurus/netwalker like device that is a bit faster May 14 20:45:45 Got a C860 and a Netwalker, but interested in such device with some more oompff :P May 14 20:48:30 RXShorty, don't know that devices exactly but these days any smartphone has more oompff May 14 20:49:27 Yeah I know but those devices lack a decent keyboard... And don`t want to run android. May 14 20:50:02 that are 2 good arguments May 14 20:50:02 lautriv thanks for the responds though... May 14 20:50:21 :) May 14 20:50:59 RXShorty: what about openpandora? assuming you can get one that is... May 14 20:51:00 RXShorty, i'm even on a older thingie but just because it's there. and who said you can't kick android ? May 14 20:51:50 Had an OpenPandora, first batch but the build quality was .. wel not Zaurus like... May 14 20:52:43 fray, meta-oe contains 8 different subdirs, do i have to place them all in bblayers.conf because it seems to stuck here. May 14 20:53:02 RXShorty: that's a pretty high bar unfortunately :/ May 14 20:53:03 lautriv Hmm.. that isn't how it used to be.. May 14 20:53:11 Was looking in a Zaurus SL6000, love the looks but there isnt a wifi version for sale at reasonable price. May 14 20:53:24 lautriv: no, you should only add the one(s) you need May 14 20:54:19 bluelightning: you are right, thought the netwalker would be it! (but working on it right now) May 14 20:54:30 crappy keyboard... May 14 20:54:36 ok, then i found the problem (misread) that is meta-oe/meta-oe where i omitted the second. May 14 20:55:09 (I could be wrong, I haven't used meta-oe in a while "as-is".. I usually only grab one or two recipes from it that I'm working with) May 14 20:55:14 whatfor is meta-efl ? ( ok, i could read it somewhere but you are a bunch faster) May 14 20:55:22 no idea May 14 20:55:54 http://git.yoctoproject.org/cgit/cgit.cgi/meta-selinux/ May 14 20:55:57 oops sorry May 14 20:57:53 lautriv: it's for EFL (Enlightenment) May 14 20:58:16 bluelightning, ok, so no need if i choosed xfce May 14 20:58:33 yay ! parsing. May 14 20:59:36 can a bitbake be piped to less ? ( too much for scrollback) May 14 20:59:51 better to do bitbake ... | tee May 14 20:59:53 less May 14 21:00:04 you can watch it live, but then go back and review it.. May 14 21:00:19 (note if you look in tmp-eglibc/cooker..... are all of the log files for your runs May 14 21:00:33 there just doesn't happen to be a link to the 'last' run to make it easy when you have a million of them in there May 14 21:05:06 ok, fixed anything but one --> WARNING: No bb files matched BBFILE_PATTERN_multimedia-layer '^/usr/src/oe/meta-oe/meta-multimedia/ where that dir has conf/layer.conf May 14 21:30:41 Hmm now looking at the nanoote with wifi adapter. Its 300 mhz but for ssh server mangement it will do as backup May 14 21:31:01 anybody has some handson with the key board May 14 21:31:04 ? May 14 21:36:42 ok, removed that meta-multimedia for now ( there are no receipes in ) what are the magic words for my preconfigured setup --> angstrom/xfce/hx4700 to get something bitbaked ? May 14 21:46:54 ok, this chan is obviously dominated by europeans, anything fellt asleep :)) May 14 21:47:15 just "bitbake" if everything is configured May 14 22:15:33 trygvis, any manual says something about bitbake some-special-version May 14 22:37:57 version of a package? just bitbake foo_1.2.3 May 14 22:38:07 but I think you need a bb file for that particular version May 14 22:38:33 if you want a whole distribution I think you need to checkout a specific git revision. normally at least May 14 22:38:39 I'm still new to this so I might be wrong May 14 23:00:12 if you just need to build package of X version then bitbake blah-VERSION works, if you need you images to contain a certain version then adding PREFERRED_VERSION_blah = "VERSION" to local.conf works May 14 23:00:53 I missed the beginning of this convo as it was earlier than 6am :-D **** ENDING LOGGING AT Tue May 15 02:59:59 2012