**** BEGIN LOGGING AT Fri Jan 20 02:59:57 2012 Jan 20 05:26:49 hi all. I am using release-2011.03 openembedded from git.i am facing a problem with ncurses when it trying to download the patch.But the patch is not available in ftp.invisible-island.com.Anyway to skip that patch ? Jan 20 06:44:59 vicky_: thats a problem Jan 20 06:45:04 which patch is it failing on Jan 20 06:46:13 ncurses-5.7-20110108-patch Jan 20 06:52:28 vicky_: did you upgrade to latest on the release branch Jan 20 06:52:56 vicky_: if not I would suggest to checkout the release branch since that will be uptodate Jan 20 06:55:39 k.let try Jan 20 06:57:05 release-2011.03 is latest.right? Jan 20 07:12:37 vicky_: yes but clone the branch from git Jan 20 07:14:14 git clone git://git.openembedded.org/openembedded;cd openembedded;git checkout -b 2011.03-maintenance origin/2011.03-maintenance Jan 20 07:14:39 will get you the release branch Jan 20 07:22:02 thanks. Jan 20 09:14:24 hi folks Jan 20 09:23:49 anyone knows how much it takes to build opie-image? It already used 32G and I have no idea how much more space I need Jan 20 09:24:56 wooha! massive stats for building something for an embedded device :) Jan 20 09:25:23 are building for each and every arch ever exiswted? ;) Jan 20 09:32:18 hello bluelightning Jan 20 09:32:30 hi ipaqlor, all Jan 20 09:32:52 ipaqlor: yes, building any image does use up a lot of space Jan 20 09:32:56 bluelightning: i had some major problems starting up as it was suggested yesterday here Jan 20 09:33:33 ipaqlor: but it got as far as userspace? Jan 20 09:33:43 bluelightning: no Jan 20 09:33:54 bluelightning: i got as far as damnin the whole oe documentation Jan 20 09:34:06 sobczyk: yes, building any image does use up a lot of space... shouldn't be much more than 32g though Jan 20 09:35:00 sobczyk: you can add INHERIT += "rm_work" to your local.conf to avoid keeping the old tmp/work files, that will save some Jan 20 09:35:11 ipaqlor: so what did happen? Jan 20 09:35:29 bluelightning: i read a third of the usermanual Jan 20 09:36:04 bluelightning: then i was really confused. and i thought. ok, forget the manual. remember what you read in the ircs. Jan 20 09:36:28 bluelightning: take oe-core and a bsp...bla...meta-handhelds... gogogogo Jan 20 09:36:50 ipaqlor: yes, the manual is out of date Jan 20 09:37:13 bluelightning: rm_work doesn't work very well with sstate for multi machine builds.. Jan 20 09:37:19 bluelightning: yes, also, wikis an everything ewlse i could find Jan 20 09:38:05 JaMa: hmm, we should really try to fix that... Jan 20 09:38:24 bluelightning: tahts really annoying. i i dont think reorganizing the whole mess will be done soon whilst there isn even a documentated getting started guide in2 pages or something that tells you how oe community has "descided" its the way to go... Jan 20 09:39:05 the somewhat annoying thing is there are plenty of people who recognise our documentation is out of date but few who are prepared to spend the time updating it :/ Jan 20 09:39:16 bluelightning: i would do it Jan 20 09:39:50 that's great, although I was mainly referring to people who have been around for a while Jan 20 09:39:54 bluelightning: this is really a starnage state Jan 20 09:43:01 bluelightning: the thing is. i have no idea where to start. because noyt even the "start here" readme part is valid when taking the new "layer"shmoo into account. Jan 20 09:44:34 after talking to you guys here yesterday Jan 20 09:44:45 and the usermanual incident Jan 20 09:45:37 bluelightning: I've switched to modified rm_old_work from meta-micro Jan 20 09:47:14 i followed the wiki. and fetched oe-core, then i fetched meta-handhelds, then there was something missing and i got a hint here in the channel pointing me to meta-openembedded , so i got /DEVEL/oe-core/meta, /DEVEL/oe-core/sources/meta-openembedded/meta-oe and /DEVEL/oe-core/sources/meta-handheld in my bblayers.conf Jan 20 09:48:34 i dont even know if im right to this point. there is no "sacred" console image available her. the minimal build thats broken with eglibc, udev and whatnot i havent seen until now seems to be a fully bloated gui image..... Jan 20 09:50:09 bluelightning: i still cant see a "startline" its still, a "you got to do it all on your own" kind of feeling. for everything else just eats my time and i am not comming any inch closer to a build(environment) whatsoever Jan 20 09:50:59 ipaqlor: core-image-minimal should work just fine... Jan 20 09:51:02 so, where is that logfile, memo, whatever where N people decided how future oe-core and meta developpment should be done? some guidlines, anything? Jan 20 09:51:42 ipaqlor: discussions on the mailing list + some face-to-face discussions Jan 20 09:51:56 ipaqlor: there is also http://www.openembedded.org/wiki/OpenEmbedded-Core Jan 20 09:53:03 bluelightning: yeah this is not really theat informative, i read that twice ;) Jan 20 09:53:49 ipaqlor: what's missing? I'd be more than happy to extend it but so far nobody has been able to articulate what questions aren't answered by it... Jan 20 09:53:56 (I wrote that page btw) Jan 20 09:55:45 we are missing a how-to document geared towards the new way of working Jan 20 09:56:17 bluelightning: the how to start. Jan 20 09:56:38 bluelightning: did i do wron, did i do right. i dont know. it doesnt work. Jan 20 09:56:52 JaMa: if rm_old_work is working well, perhaps we should look to bring it into oe-core Jan 20 09:57:31 ipaqlor: don't forget you're resurrecting a quite old machine using the bleeding edge, so you're bound to come across breakage Jan 20 09:58:03 ipaqlor: maybe I should try to replicate your setup Jan 20 09:58:12 bluelightning: no, dont Jan 20 09:58:24 bluelightning: i want to replicate your setup! Jan 20 09:58:32 ipaqlor: :) Jan 20 09:58:36 bluelightning: i nwant to dpo li8ke you do Jan 20 09:58:37 ipaqlor: do you have some patches that enable h3600? Jan 20 09:58:43 bluelightning: noi Jan 20 09:59:08 bluelightning: i just want to setup core layer bla the way you folks here thought it out. but i have no idea how that is Jan 20 09:59:25 bluelightning: lets please start with the basic install :) Jan 20 09:59:51 bluelightning: only issue with that version from meta-micro is that some code in package*bbclass is using last directory name in $WORKDIR to pack sources to -dbg packages, and all sources ends on target in /usr/src/${PV} Jan 20 10:00:37 bluelightning: that's why I had to modify it http://git.shr-project.org/git/?p=meta-smartphone.git;a=commit;h=0d4c17de1ba7343d82bb10c311754e29d573c485 but it should be fixed in that bbclass Jan 20 10:00:42 ipaqlor: if I were to recommend something that's set up out of the box, I'd have to say I recommend poky from the yocto project (which is my day job) Jan 20 10:01:06 bluelightning: also it doesn't save as much workdir space as rm_work does, but better then nothing Jan 20 10:01:29 JaMa: ok... perhaps you or phil could post a patch for oe-core then? Jan 20 10:02:17 bluelightning: thanks Jan 20 10:03:18 bluelightning: i want a starting point thats valid Jan 20 10:03:30 bluelightning: maybe im just asking the wrong questions. Jan 20 10:03:32 pb_: ^ as author, would you send rm_old_work.bbclass to oe-core? Jan 20 10:04:32 bluelightning: is there, or has there, at any time been any consens on how to resurrect a broken target machine, in my case h3600, and make it buildable via oe again? Jan 20 10:04:54 ipaqlor: just work through the issues one by one, that's what I do... Jan 20 10:05:28 bluelightning: i try that. but it seems my english is unusable Jan 20 10:05:32 ipaqlor: it may help if you build for the qemuarm machine first, just to make sure you have a working base Jan 20 10:05:54 bluelightning: what base? Jan 20 10:06:07 bluelightning: how is a base defined? Jan 20 10:06:16 bluelightning: is my setup here a base? Jan 20 10:06:31 bluelightning: tahts the "base" problem Jan 20 10:06:32 :) Jan 20 10:06:56 ipaqlor: I mean, it will make sure there is nothing wrong building a simple OS image for a known-good machine Jan 20 10:07:14 with your individual setup Jan 20 10:07:28 then you can move onto building for h3600 Jan 20 10:08:05 bluelightning: regarding my bblayers.conf content. is this a base setup? Jan 20 10:08:18 bluelightning: and my layer selection Jan 20 10:08:39 ipaqlor: so, I would start by not adding any layers, set MACHINE to qemuarm, and then bitbake core-image-minimal Jan 20 10:08:43 bluelightning: i am fishing here in the dark with a screwdriver for a rod Jan 20 10:09:16 bluelightning. well, ok, that should not fail, but should not bring me anywhere also :( Jan 20 10:09:44 ipaqlor: it will ensure you're not stumbling over issues that aren't at all related to h3600 Jan 20 10:09:56 bluelightning: what would i have learmed after that. that you current git source builds the minimal testoinmg target? Jan 20 10:09:57 ipaqlor: some of the stuff that gets built in that can be reused when you change to h3600 as well Jan 20 10:10:10 on your machine, yes Jan 20 10:10:33 bluelightning: ok, ill do that. but i doubt its sense. Jan 20 10:10:36 good morning Jan 20 10:10:54 ipaqlor: in the mean time I'm going to try to build for h3600 here Jan 20 10:11:31 bluelightning: really not necessary. but you could document your "basic" setup for doing such a thing :) Jan 20 10:12:52 ok, build started Jan 20 10:17:14 ok, out of diskspace :) Jan 20 10:19:13 ok, ill start from scratch and just keep my downloads directory. Jan 20 10:22:01 hi florian Jan 20 10:34:42 ipaqlor: so, we will need to update the machine configuration to use the latest kernel at the very least Jan 20 10:36:25 bluelightning: pardon? Jan 20 10:37:15 ipaqlor: right now, conf/machine/h3600.conf uses the linux-handhelds-2.6 kernel; that is quite old now Jan 20 10:37:38 ipaqlor: that web page I linked to here the other day was talking about 2.6.32 I think Jan 20 10:38:14 so we'd need to dig up / adapt a 2.6.32 kernel recipe and point to that in h3600.conf Jan 20 10:39:07 not too tricky and I'd be happy to do the work, just not today Jan 20 10:40:03 bluelightning: well, thnkas. but, this is a whole different issue. i allready compiled a kernel for my ipaq yesterday. but i couldnt get any images done due to layers upon layers and my magical SUMMARY recursion..... Jan 20 10:40:18 bluelightning: i am all into bootom up now Jan 20 10:40:38 bluelightning: you wont stop me ;) Jan 20 10:40:57 ipaqlor: that problem is puzzling, I've not seen that before here... Jan 20 10:41:03 ipaqlor: so what layers are you using currently? Jan 20 10:41:47 bluelightning: as you told me, the basic core wihich has some /DEVEL/oe-core/meta written inside after ./oe-init-build-env Jan 20 10:42:37 blielightning: and when the quemuarm is built, i rm -rf the whole thingy and start where i am now as far as my motivation goes Jan 20 10:42:46 ipaqlor: what I meant was, what do you have in bblayers.conf? Jan 20 10:43:07 bluelightning: the default Jan 20 10:43:08 LCONF_VERSION = "4" Jan 20 10:43:08 BBFILES ?= "" Jan 20 10:43:08 BBLAYERS ?= " \ Jan 20 10:43:08 /DEVEL/oe-core/meta \ Jan 20 10:43:59 .. Jan 20 10:44:09 ipaqlor: so, how did you actually enable the layers if you didn't add them to that file? Jan 20 10:44:22 git clone git://git.openembedded.org/openembedded-core oe-core ; cd /DEVEL/oe-core ; git clone git://git.openembedded.org/bitbake bitbake ; . ./oe-init-build-env build ; sed -e 's/#MACHINE ?= "qemuarm"/MACHINE ?= "qemuarm"/' -e 's/MACHINE ??= "qemux86"/#MACHINE ??= "qemux86"/' conf/local.conf > conf/local.conf_ ; mv conf/local.conf_ conf/local.conf Jan 20 10:44:28 this is where i am now Jan 20 10:44:53 i cleaned up everything. to make it the way it "maybe yes maybe no" "should" be :) Jan 20 10:45:42 ipaqlor: ok right Jan 20 10:45:43 only manipulation on my side is i keep the downloadsflder moving around. for i would have downloaded glibc 30 times by now if i hadntn :) Jan 20 10:45:59 yes that's fine, I keep my dl dir also Jan 20 10:51:48 bluelightning: the bad thing about keeping the folder i assume is, that if you do it long enough, your buildsetup isnt compareable to otheres anymore due to project/mirrordeath Jan 20 10:52:17 ipaqlor: that's true; it's a minor issue though Jan 20 10:58:55 echo please_please_please_more_performance > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor *thinking about the days when compiling on a dual core machine" was jsut awesome. now my t7300@2gHz barly satisfies me performancewise:( Jan 20 11:00:16 dual core, what am i talking about, dual prozessor, yesss! :) no wimpy die sharing and such... Jan 20 11:00:58 die sharing is good... the wires are shorter... ;) Jan 20 11:01:25 bluelightning: you know what they say about short wires Jan 20 11:02:13 viva la resistance ;) (Ohm) Jan 20 11:04:13 heh Jan 20 11:43:45 cu later Jan 20 13:11:59 florian: you didnt upload that to http://www.linuxtogo.org/~florian like usual? Jan 20 13:12:47 cs_nbp: no not yet... but I can do Jan 20 13:12:48 elo btw Jan 20 13:13:04 is there a way to force removal of the bitbake database without removing tmpdir, normally I touch local.conf or so and it will rebuilt, but not today :-( Jan 20 13:13:11 ah k it yesterday you were babbling about upload in half sleep :D Jan 20 13:15:13 uploading... unluckily calibration does not work yet. Jan 20 13:15:22 np Jan 20 13:15:42 just to see what ur image is like Jan 20 13:15:54 and we couldnt build one for like 4 days now Jan 20 13:16:03 still var/run Jan 20 13:16:25 I really love these error messages: "ERROR: Function 'do_stage' failed (see /dev/null for further information)" Jan 20 13:20:01 apparently nuked shell env vars, new window and everything is singing and dancing again Jan 20 13:37:29 /dev/null now used as manual, too :D Jan 20 13:38:35 anyways resume has problems Jan 20 13:38:55 sometimes it doesnt wake up and says Jan 20 13:39:02 LX require IRQ Jan 20 13:39:40 but i cant trigger it Jan 20 13:39:47 seems random Jan 20 13:40:46 florian: u saw that too? Jan 20 13:42:50 hi is the 2.6.35 kernel too old for udev 175? Jan 20 13:43:07 if so, is backporting sysaccept_4 enough Jan 20 13:43:08 ? Jan 20 13:45:49 florian: is this OE-Core or classic? Jan 20 13:45:56 (the error) Jan 20 13:47:15 I'll look if I've sysaccept_4 Jan 20 13:49:36 hi otavio Jan 20 13:51:00 cs_nbp: not yet... but I didn't try much suspending so far Jan 20 13:51:21 bluelightning: classic Jan 20 13:53:41 florian: ah, right Jan 20 14:04:01 cs_nbp: done Jan 20 14:04:23 bluelightning: Don't tell me that this might be expected behaviour ;) Jan 20 14:04:36 florian: yes, saw it and forwarded the link to Christoph Pulster :p Jan 20 14:05:07 florian: it doesn't look right for sure Jan 20 14:05:41 florian: I only got that resume prob when it was in suspend for a long time, also there is a deactivated (ifconfig eth0 down) PCMCIA card in the slot. Jan 20 14:05:47 cs_nbp: ah cool... its still some way to go. maybe this evening i have a little bit of time left Jan 20 14:06:29 florian: Christoph really wanted to try something actual Jan 20 14:07:05 florian: as he sells those I guess with a working modern Linux they will sell better :D Jan 20 14:07:18 cs_nbp: :-) Jan 20 14:07:37 so you guys have had success with your netbook pro efforts? Jan 20 14:07:53 The best would be to make a really small image for the internal flash Jan 20 14:08:11 you mean a complete sys? Jan 20 14:08:41 bluelightning: yes... linux 3.2 starts to become useful but the internal flash is very small for a modern filesystem Jan 20 14:08:59 cs_nbp: yes... and I admit it is some effort Jan 20 14:09:10 florian: o_O Jan 20 14:09:32 but k wince is completely mummyfied and unusable Jan 20 14:09:36 anyways Jan 20 14:10:18 bluelightning: sure, got python working and can ssh to other hosts, so basically I can work Jan 20 14:10:27 awesome :) Jan 20 14:11:05 now only psion needs to produce a heap of modern hardware version of this Jan 20 14:11:07 :D Jan 20 14:12:22 no but for basic work its fast enough Jan 20 14:12:53 just no 3d shooting rocket cows at godzilla Jan 20 14:13:14 yeah but who needs that... :) Jan 20 14:14:17 and on a mobile device.,. the ppl have no idea how they look when playing on their smartphones in the tram/bus Jan 20 14:18:36 btw there's a pseudo-brick state (had it like 4 times), just remove all batteries and put them back in when it happens Jan 20 14:19:57 cs_nbp: florian: would it be possible for you guys to throw this stuff in a wiki page somewhere if you haven't already? Jan 20 14:20:34 bluelightning: I am keeping log of what I'm doing, wouldn't be too hard to change that to wiki-style Jan 20 14:20:49 florian: speaking of which the ltg wiki is really slow again :( Jan 20 14:21:23 bluelightning: problem is it's changing really fast and probably a wiki will be out of date the moment it's finished :D Jan 20 14:21:40 cs_nbp: still, it's useful to have the stuff documented/tracked Jan 20 14:21:47 y/agreed Jan 20 14:23:31 florian: I may have tracked a build error due to me changing the '-nbpro0' mid-fix string to something else to see if my config was taken. That lead to connman having weird dependency problems, don't know why Jan 20 14:24:02 bluelightning: would be a good idea indeed... we used to have a nbp wiki page somewhere Jan 20 14:24:17 there is one on linuxtogo I think Jan 20 14:25:46 bluelightning: I'll try to find out as soon as I get a usable shell there. Currently everyone here seems to try to flood our external connection :-/ Jan 20 14:26:08 not me .,. Jan 20 14:26:32 florian: :( so you have been able to determine that it's the connection and not the machine? Jan 20 14:26:45 I guess maybe you can't tell yet Jan 20 14:27:07 btw do you know whats wrong w locale? my normal sysadmin actions don't work, there's not even a 'locale' command Jan 20 14:27:21 but can't set LC_ALL manually Jan 20 14:28:02 bluelightning: not yet... but the last time I looked after the wiki was that show the machine was almost idle Jan 20 14:28:07 installed localedef but that didn't help either Jan 20 14:28:44 as it doesn't find any reasonable locale Jan 20 14:29:44 i put 'EGLIBC-GENERATE-LOCALES = "en_US.UTF-8"' into local.conf, might that help? Jan 20 14:30:51 actually with _ and nott - Jan 20 14:49:43 florian: or does by pure luck locale work in ur image^^ Jan 20 15:28:26 otavio, hi Jan 20 16:27:29 03Christopher Larson  07master * r774f59f06e 10bitbake.git/ (7 files in 3 dirs): Jan 20 16:27:29 Add dependency on the backported python 3.2 concurrent.futures Jan 20 16:27:30 Signed-off-by: Christopher Larson Jan 20 16:27:40 03Christopher Larson  07master * rd104f29871 10bitbake.git/lib/bb/cooker.py: Jan 20 16:27:40 cooker: use futures rather than a multiprocessing pool Jan 20 16:27:40 This avoids some silent parser hangs we were seeing which were Jan 20 16:27:40 near impossible to debug as no user feedback was given. Jan 20 16:27:40 [RP: Tweak commit message] Jan 20 16:27:40 Signed-off-by: Christopher Larson Jan 20 16:27:41 Signed-off-by: Richard Purdie Jan 20 16:35:09 03Darren Hart  07master * r8b71370854 10bitbake.git/doc/manual/usermanual.xml: Jan 20 16:35:09 usermanual: Correct "inherit" search to bbclass from oeclass Jan 20 16:35:09 Bitbake looks for bbclass now, not oeclass. Update the docs accordingly. Jan 20 16:35:09 Signed-off-by: Darren Hart Jan 20 16:35:09 Signed-off-by: Richard Purdie Jan 20 16:39:46 oeclass. yikes. Jan 20 16:41:43 yeah I was surprised there was still a reference in there too Jan 20 16:42:29 03Paul Eggleton  07master * r721e986624 10bitbake.git/lib/bb/fetch2/__init__.py: (log message trimmed) Jan 20 16:42:29 bitbake/fetch2: allow resuming interrupted fetch Jan 20 16:42:29 A lot of our checks in the fetcher code assume that if the downloaded Jan 20 16:42:29 file exists, the download finished successfully; this meant that if Jan 20 16:42:29 BitBake was interrupted in the middle of a fetch then it would not Jan 20 16:42:29 resume the fetch the next time, but instead attempt to use the Jan 20 16:42:30 half-fetched file and usually fail as a result. Jan 20 16:42:39 03Paul Eggleton  07master * re44d5be98f 10bitbake.git/lib/bb/runqueue.py: Jan 20 16:42:39 bitbake/runqueue: always log tasks summary Jan 20 16:42:39 It's unclear from the history but at some point we stopped logging the Jan 20 16:42:39 "Tasks Summary" NOTE when tasks failed. Reinstate this for failure, and Jan 20 16:42:39 also make the count of attempted tasks include the failed task. Jan 20 16:42:39 Signed-off-by: Paul Eggleton Jan 20 16:42:40 Signed-off-by: Richard Purdie Jan 20 16:42:40 03Richard Purdie  07master * r7da7dff83e 10bitbake.git/lib/bb/siggen.py: (log message trimmed) Jan 20 16:42:41 siggen: Ensure correct runtask dependency output is shown for diffsigs Jan 20 16:42:41 The actual task names are discounted for comparison of dependent tasks, only Jan 20 16:42:42 the actual hashes are used. This updates the comparison code to account for Jan 20 16:42:42 this change, attempting heuristic matching for more user friendly output but Jan 20 16:42:43 falling back to showing the changed hashes directly. This avoids some confusing Jan 20 16:42:43 output to users where it looked like tasks had changed when they had not and Jan 20 16:43:07 wut Jan 20 16:43:28 03Richard Purdie  07master * r69fd5ad4a8 10bitbake.git/lib/bb/cooker.py: Jan 20 16:43:28 cooker.py: Allow siggen classes to be added by the metadata Jan 20 16:43:28 By calling init_parser which sets up the siggen code after the ConfigParsed Jan 20 16:43:28 event is fired, we can allow the metadata to add siggen classes which Jan 20 16:43:28 was always what the code intended. Jan 20 16:43:28 Signed-off-by: Richard Purdie Jan 20 16:43:38 03Richard Purdie  07master * r164195c068 10bitbake.git/lib/bb/siggen.py: Jan 20 16:43:38 siggen.py: Abstract the runtime task dependency handling code in the generators Jan 20 16:43:38 This means that custom signature handlers can override specific parts Jan 20 16:43:38 of the code without having to reimplement whole functions allowing them Jan 20 16:43:38 more flexibility. Jan 20 16:43:38 Signed-off-by: Richard Purdie Jan 20 16:49:19 03Martin Jansa  07master * r9a57ec705c 10bitbake.git/lib/prserv/serv.py: (log message trimmed) Jan 20 16:49:19 prserv: fix import of sqlite3 Jan 20 16:49:19 * this is used in all other bitbake parts where sqlite3 is used, don't Jan 20 16:49:19 know why it wasn't used here, but it fails e.g. on Gentoo Jan 20 16:49:19 Traceback (most recent call last): Jan 20 16:49:19 File "bin/bitbake", line 39, in Jan 20 16:49:19 from bb import cooker Jan 20 16:49:29 03Martin Jansa  07master * rb4a40c2139 10bitbake.git/setup.py: (log message trimmed) Jan 20 16:49:29 setup.py: install prserv too Jan 20 16:49:29 * otherwise cooker fails: Jan 20 16:49:29 Traceback (most recent call last): Jan 20 16:49:30 File /usr/bin/bitbake, line 39, in Jan 20 16:49:30 from bb import cooker Jan 20 16:49:31 File /usr/lib64/python2.7/site-packages/bb/cooker.py, line 39, in Jan 20 18:28:36 03Richard Purdie  07master * r65c2eb1c09 10bitbake.git/ (bin/bitbake lib/bb/__init__.py): Jan 20 18:28:36 Bump the version to 1.15.1 Jan 20 18:28:36 Signed-off-by: Richard Purdie Jan 20 19:32:29 florian: heh flo Jan 20 19:32:41 florian: tworaz wrote something about locale Jan 20 19:32:56 florian: he said kdrive server might not support xlocale Jan 20 19:33:09 florian: of course after I installed all locale ipks Jan 20 19:35:46 florian: and for touchscreen calibration we should use ts-calibrate from tslib Jan 20 19:36:24 re Jan 20 19:37:19 cs_nbp: it does not support xlocale but imho we do not need it at all. we do not want to localize x11 itself Jan 20 19:37:38 cs_nbp: we have this in the latest image... just need to check why it failed here Jan 20 19:38:33 flroian: just have problems because when I ssh somewhere and do emacs umlauts are broken Jan 20 19:38:46 florian: and french accents too Jan 20 19:39:06 florian: -nox Jan 20 19:43:00 cs_nbp: this should be independent from X11 localisation Jan 20 19:43:15 do u have glibc locales installed Jan 20 19:43:58 but if generating glibc binary locales is disabled... it woul generate them on first boot but this might fail Jan 20 19:44:02 on the ssh-client system? no Jan 20 19:44:28 only eglibc-locale-stuff Jan 20 19:44:50 should I put GLIBC_GENERATE_LOCALES variable in local.conf? Jan 20 19:45:35 lets see if you have proper locale configured for your system Jan 20 19:45:42 is it using C locale by any chance ? Jan 20 19:46:24 in the beginning it had C locale I think, the command 'locale' was not even installed Jan 20 19:46:28 good point... in gpe images we get this "for free" but I have no idea if LXDE cares Jan 20 19:46:52 now I can select from a range of locales and set LANG eg to de_DE.UTF-8 Jan 20 19:47:00 without complains Jan 20 19:47:11 but the weird umlaut over ssh stayed Jan 20 19:47:25 $TERM is xterm Jan 20 19:48:06 on the ssh-'server' seide is ubuntu 10.10 Jan 20 19:50:28 the client is dropbear of course Jan 20 19:50:32 what is what Jan 20 19:50:38 client is dropbear Jan 20 19:50:45 and server is OE based ? Jan 20 19:51:01 nope, server is openssh server from ubuntu 10.10 Jan 20 19:51:15 and on the client i see the umlaut problems Jan 20 19:51:22 client is oe based Jan 20 19:51:31 you see it when you login from another box thats running oe ? Jan 20 19:51:39 ok. Jan 20 19:51:40 when I log in from other linux computers I dont have the weird umlauts Jan 20 19:51:53 and you dont see it when you login to same server from some other machine Jan 20 19:51:58 say another ubuntu box Jan 20 19:52:02 or putty or something Jan 20 19:52:06 i have an openmoko gta02 w shr, i can call emacs on my pc fine w umlaut Jan 20 19:52:35 khem: thats it Jan 20 19:52:37 your pc = server w/openssh Jan 20 19:52:46 ok. Jan 20 19:52:50 y Jan 20 19:54:51 cs_nbp: what does echo $LANG say Jan 20 19:55:43 khem: on client its de_DE.UTF-8 now, same on server Jan 20 19:56:34 cs_nbp: the ssh server is openssh not dropbear right ? Jan 20 19:56:45 khem: yes Jan 20 19:57:34 what does /etc/profile.d/lang.sh have on client machine Jan 20 20:00:29 khem: doesn't exist Jan 20 20:00:38 khem: apparently.,. Jan 20 20:00:55 khem: only tslib.h in profile.d Jan 20 20:01:12 khem: what is the corresponding ipk, maybe I have it? Jan 20 20:02:16 cs_nbp: just create one Jan 20 20:02:54 cs_nbp: add echo "export LANG=\"de_DE.UTF-8\"" > /etc/profile.d/lang.sh Jan 20 20:02:54 k Jan 20 20:02:56 or something Jan 20 20:03:17 no shabang? Jan 20 20:03:33 no Jan 20 20:04:48 * Russ notes the true origin of the phrase, 'the whole shabang' Jan 20 20:04:55 now just reconnect? Jan 20 20:05:25 yes Jan 20 20:05:45 umlaut still dead Jan 20 20:05:50 actually another thing you could retry is try to use strace Jan 20 20:06:04 and see where its getting its locales from Jan 20 20:06:07 btw i have only locale and localedef as commands for locale control Jan 20 20:06:17 I guess the problem is mismatch between server and host Jan 20 20:06:35 so strace on client or host? Jan 20 20:06:41 client kinda slow Jan 20 20:06:43 client Jan 20 20:06:46 -.- Jan 20 20:06:48 thought so Jan 20 20:07:13 ima have to install strace forst :D Jan 20 20:13:09 ok baked and installed Jan 20 20:13:34 and did one ssh and logout Jan 20 20:14:06 try to see what all files dropbear tried to open Jan 20 20:14:54 lines starting with open, right? Jan 20 20:15:11 yes Jan 20 20:15:23 got dem all Jan 20 20:19:25 libnss_nis.so.2 can't be found, but I doubt it's that Jan 20 20:29:21 prolly not Jan 20 20:29:31 does it work if you have openssh client ? Jan 20 20:30:17 on the nbp? Jan 20 20:30:23 yes Jan 20 20:30:35 would have to write a recipe for a client I think, or is there on in oe? Jan 20 20:31:07 is it possible to limit the depth of the bitbake -g command? Jan 20 20:34:14 should I remove dropbear or will opkg take care of everything Jan 20 20:34:26 I think you have to remove it Jan 20 20:34:33 it will complain about conflict though Jan 20 20:34:35 it should Jan 20 20:42:23 ill have to transfer some files, my local ipk is out of date Jan 20 20:42:49 libcrypto is 1.0.0 but should be >= 1.0.0F Jan 20 20:42:51 f Jan 20 20:58:25 anyone has schroedinger compilation error showing that some types are not defined? (the orc/orc.h is used for all intXX_t, but it seems as it's not defined) Jan 20 20:59:37 iirc theres a patch for orc on patchwork Jan 20 21:05:47 hrm... do we need to build distribution feeds? :) Jan 20 21:06:09 ssh produces the same error (weird rectangle) Jan 20 21:06:38 khem: O mean openssh-client Jan 20 21:07:00 cs_nbp: ok so its not dropbear problem as I was thinking Jan 20 21:07:10 its really related to locales I think now Jan 20 21:07:33 im missing locale-gen, could that be it? Jan 20 21:13:32 I think you can add some locales to your image at build time Jan 20 21:13:35 thats should be it Jan 20 21:13:55 what do I need to build to generate locales? Jan 20 21:14:24 if 'locales' is even the right package that contains locale-gen Jan 20 21:14:35 in oe Jan 20 21:17:11 cs_nbp: ENABLE_BINARY_LOCALE_GENERATION = "1" Jan 20 21:17:14 in your local.conf Jan 20 21:17:21 and rebuild eglibc and image Jan 20 21:17:42 I'vo got that, so in the new repo there should be a 'locales' package? Jan 20 21:17:50 aah rebuild eglibc Jan 20 21:22:24 if that generates eglibc-binary-localedata - I already have thoe and they are installed. all. Jan 20 21:22:41 just missing actual locale-gen command Jan 20 21:22:56 building in bg anyways:) Jan 20 21:23:00 the solution is as for old orc recipe, add DEFAULT_PREFERENCE = "-1" Jan 20 21:27:07 cs_nbp: cs_nbp on your client whats does set |egrep 'LANG|LC_' show Jan 20 21:28:16 LANG=de_DE.UTF-8 LC_ALL= Jan 20 21:33:26 what happens if you export LC_CTYPE="de_DE.UTF-8" Jan 20 21:34:25 when I type 'locale' it appears to be set to that value Jan 20 21:34:31 but I can set it again Jan 20 21:34:54 so I'm noticing a number of the tasks that I've prevously seen referenced have disappeared. E.g. the manual refers to task-python-everything or task-sdk-native but there aren't an bb files with these names any longer. Jan 20 21:35:29 spacecolonyone: yes they might have been renamed Jan 20 21:35:44 error remains Jan 20 21:35:55 hmmm Jan 20 21:36:09 I wonder if its your terminal Jan 20 21:36:20 look through old git commits to tell what became what? Jan 20 21:36:21 can you see utf data ok on the client ? Jan 20 21:36:53 cs_nbp: e.g. try to view the same file on client only Jan 20 21:36:56 eg files copied with an sd card? Jan 20 21:37:05 yes Jan 20 21:37:27 Also, I cant seem to find a bb for task_base_extended, though base-image uses it and still works. How does that work? Jan 20 21:37:30 hi khem btw. Jan 20 21:37:47 spacecolonyone: is it oe.dev ? or oe-coe Jan 20 21:37:48 core Jan 20 21:37:50 woglinde: hi Jan 20 21:38:10 I'm not sure what you mean khem Jan 20 21:38:23 khem thanks for fixing gnutls Jan 20 21:38:28 where are looking at Jan 20 21:38:30 woglinde: np Jan 20 21:38:45 woglinde: I am actively fixing uclibc in oe-core Jan 20 21:38:58 woglinde: uclibc git recipes are looking lot promising now Jan 20 21:39:59 woglinde: core-image-sato boots and works on all qemu machines except arm where there seems to be a problem which I dont know and on mips pango-querymodule dies Jan 20 21:40:01 khem: nope, looks weird also (I just made a text only file w Umlauts), afk 10 mins Jan 20 21:40:06 console-images work well on all Jan 20 21:40:26 i ran find ./ -name "task-base-extended*" in sources Jan 20 21:40:31 cs_nbp: try to view the same file in another editor Jan 20 21:41:26 spacecolonyone: well if this is a package then it could be in a recipe which may be named differently Jan 20 21:42:04 spacecolonyone: look in task-base.bb Jan 20 21:42:17 I see. I'm still learning this whole chain so I'm trying to piece together what happens. Jan 20 21:43:12 bingo. Jan 20 21:46:24 Thanks, khem. Jan 20 21:47:12 khem yes mips-pango I read on the uclibc ml Jan 20 22:02:07 khem: if I view the text in 'Notes', Umlauts get displayed correctly Jan 20 22:02:39 less or vim in terminal show squared hell Jan 20 22:03:41 ok your terminal might be an issue Jan 20 22:03:51 can you try with some other terminal Jan 20 22:03:59 xterm or something lxdeterm Jan 20 22:04:18 i think lxterm Jan 20 22:04:51 TERM says xterm Jan 20 22:06:13 install another one then Jan 20 22:06:22 if you have lxterm install xterm Jan 20 22:06:35 its not TERM I Am talking about its the program itself Jan 20 22:07:20 its lxterminal, only the variable is set to xterm Jan 20 22:07:24 ill install xterm Jan 20 22:07:50 it could be that lxterminal did not build its locales properly Jan 20 22:20:21 if I start xterm from lxterm it complains about 'locale not supported by Xlib, locale set to C' Jan 20 22:20:49 'X locale modifiers not supported, using default' Jan 20 22:21:14 xterm: failed to open input method Jan 20 22:21:48 hmm Jan 20 22:21:55 thats the problem I think Jan 20 22:22:26 at least sounds like the core of it Jan 20 22:23:15 has it to do w xserver-kdrive? Jan 20 22:25:12 what does locale -a say ? Jan 20 22:25:29 ouf Jan 20 22:25:38 I think it say everything Jan 20 22:26:19 all xx_XX and xx_XX.ISO-8859-1 are there Jan 20 22:26:37 from a to z Jan 20 22:26:56 probably happened on the course of me trying to fix locale Jan 20 22:29:11 do u have /usr/X11R6/lib/X11 Jan 20 22:29:51 no /usr/X11 anything, R6 or not Jan 20 22:30:05 so no /usr/X11 too Jan 20 22:30:06 launch it like LANG=C xterm Jan 20 22:30:43 complains the same way as before about Xlib Jan 20 22:33:59 better strace it Jan 20 22:34:17 you will be faster to find out what is missing Jan 20 22:36:48 /usr/share/X11/locale/locale.[alias|dir] ... (No such file or directory) Jan 20 22:37:04 sounds like its got to do w it Jan 20 22:41:08 those should come from libx11 Jan 20 22:41:14 do you have libx11 installed Jan 20 22:41:56 libx11-6 and libx11-xcb1 Jan 20 22:42:05 cs_nbp: is /usr/share/X11/locale/ populated at all ? Jan 20 22:42:45 khem: no, in /usr/share/X11 is only XErrorDB and xkb Jan 20 22:42:57 thats a problem then Jan 20 22:43:06 your x does not have localedata installed Jan 20 22:43:30 add libx11-locale to packages to be installed Jan 20 22:43:58 k Jan 20 22:49:45 cs_nbp: you get opkg install it Jan 20 22:50:07 y its almost done Jan 20 22:50:19 baking Jan 20 22:57:05 ok, /usr/share/locale now populated, xterm not complaining at startup, but Umlauts looking weird in sample text Jan 20 22:57:56 where'd you get xterm? Jan 20 22:58:50 bitbake xterm Jan 20 22:59:55 cs_nbp: even in xterm ? Jan 20 23:00:12 goddamnit why can't I get to my compile box Jan 20 23:00:29 khem: yes lookin weird in xterm Jan 20 23:00:37 after not complaining at start Jan 20 23:00:48 can you post the snapshot somewhere Jan 20 23:01:22 yes, but its going to be done by a digicam, no screenshot gadget yet Jan 20 23:01:34 or can I pipe screen output somehow^^ Jan 20 23:01:50 screenshot ? Jan 20 23:01:56 y Jan 20 23:06:43 wait, need to increase font size for digicam Jan 20 23:13:28 ok how the heck do I get opkg to read the repo in opkg.conf Jan 20 23:14:11 I have to specify --conf=/etc/opkg.conf just to get it to update Jan 20 23:15:47 and then it just seems to ignore the conf otherwise Jan 20 23:16:03 doesn't read the package lists Jan 20 23:18:46 of course -fs 12 14 17 18 20 or 25 doesnt work and I have no mouse at the device Jan 20 23:18:53 for accessing the font menu Jan 20 23:25:07 omg it doesn't find installed fonts now-.-,just a sec Jan 20 23:34:08 impossible to make the font biǵger or to make the backlight weaker, ill figure something out in a sec Jan 20 23:40:08 ok got a photo geezus Jan 20 23:41:26 cs_nbp: hey what's your opkg.conf look like Jan 20 23:42:01 are you still having isues with the thing not reading it properly? Jan 20 23:44:30 AH Jan 20 23:44:38 needs OPKG_CONF_DIR set Jan 20 23:45:20 ok, here it is http://imageshack.us/photo/my-images/546/sam0053fo.jpg/ Jan 20 23:45:28 now I get a ton of errors >_< Jan 20 23:45:38 and normally it should look like Käsebrot mit Überbiß Jan 20 23:45:52 don't know what u seeing though :D Jan 20 23:46:34 yeah, those arent' umlauts Jan 20 23:46:49 looks like wrong locale almost Jan 20 23:46:52 nah those are Alienvasion lauts Jan 20 23:47:53 actually it seems the other photos arent bad, only look bad on cam screen :D Jan 20 23:50:02 muriani: I have manual opkg'conf' :D Jan 20 23:50:32 eh? Jan 20 23:50:50 if a package is missing i opkg install it Jan 20 23:50:51 well opkg_conf_dir partialy fixed it Jan 20 23:51:02 yeah, that's not relaly what I'm after Jan 20 23:51:33 need locale 1st so I can read remote mails Jan 20 23:52:29 now it updates Jan 20 23:52:35 but then doesn't list Jan 20 23:52:39 only lists local Jan 20 23:52:43 khem: or do you mean something else w 'snapshot' Jan 20 23:52:46 -.- Jan 20 23:52:54 anyhoo, getting ready for home Jan 20 23:56:30 khem: might have misunderstood, in german 'snapshot' means photo Jan 21 00:55:04 how do I use urls with semicolons in .bb file? **** ENDING LOGGING AT Sat Jan 21 02:59:57 2012