**** BEGIN LOGGING AT Wed Jul 06 23:59:56 2005 Jul 07 00:53:47 morning people Jul 07 00:58:46 morning all Jul 07 01:01:03 <__law__> has anyone updated monotone repo (monotone.vanille) successfully today? Jul 07 01:05:22 motning Jul 07 01:07:31 I saw some screenshot today of an app that lets the user choose between starting Opie and GPE... is that publicly available? Jul 07 01:07:56 is it a bootloader or a standalone (fb) app, and is it in OE yet? Jul 07 01:13:30 morning Jul 07 01:18:05 hi XorA Jul 07 01:18:26 XorA: it was long dinner.. Jul 07 01:19:28 hrw|work: Well I got so fat I couldnt make it back to computer :-) Jul 07 01:22:05 XorA: soon you will be able to put computer on your stomache then Jul 07 01:22:15 hrw|work: probably Jul 07 01:23:50 Hi Jul 07 01:24:19 XorA: what does your wife think about it? some things can became harder to do.. Jul 07 01:24:24 hi eJumbo Jul 07 01:24:35 hi hrw Jul 07 01:25:12 hrw:caught in a major problem since 2 days; my oe is not working need ur help. Jul 07 01:25:18 ok. I will jump to 2.6.13-rc2 Jul 07 01:26:08 eJumbo: PREFERRED_VERSION_gcc-cross="3.4.4" PREFERRED_VERSION_gcc-cross-initial="3.4.4" and rebuild? Jul 07 01:26:25 eJumbo: or set it to gcc-csl version Jul 07 01:26:56 the local.conf file worked earlier ?. but why is it not working now ? Jul 07 01:29:33 vanille.de down again? Jul 07 01:31:04 XorA: it lokks like Jul 07 01:31:17 eJumbo: pastebin it Jul 07 01:35:35 http://pastebin.com/308802 Jul 07 01:39:18 NOTE: package glibc-2.3.2+cvs20040726: completed Jul 07 01:39:24 so it is good Jul 07 01:39:47 eJumbo: add gcc* 3.4.4 into local.conf and test Jul 07 01:40:12 gcc-cross and cross-initials ? Jul 07 01:42:22 initial Jul 07 01:42:23 yes Jul 07 01:43:31 will the combination of glibc and gcc, binutils work ? Jul 07 01:43:34 you should not get gcc-snapshot building Jul 07 01:43:53 how old repo you have? Jul 07 01:44:11 last repo i made on June 24th Jul 07 01:44:36 i have no problem in starting all over again. Jul 07 01:44:49 shall i delete TMPDIR and sources ? Jul 07 01:45:37 which gcc-cross-initial was built? Jul 07 01:45:52 you can try with current tmpdir Jul 07 01:46:19 and no need to drop sources - my dl_dir is 0.8G Jul 07 01:46:43 deleted tmpdir yesterday Jul 07 01:50:04 hrw: is there any problem with my local.conf file Jul 07 01:50:13 no Jul 07 01:50:19 why does it go and fethes from cvs ? Jul 07 01:51:24 hrw: which other files i must check. all conf files not changed after last successful built. Jul 07 01:53:45 what fetch? Jul 07 01:54:46 why does it download snapshot ? Jul 07 01:56:16 ~lart glibc for not building Jul 07 01:56:16 * ibot wallops glibc with a main rotation server that needs rehubbing. It won't take long for not building Jul 07 01:58:22 eJumbo: be specific Jul 07 01:58:39 ok Jul 07 01:58:49 07 10:54 < eJumbo> why does it download snapshot ? Jul 07 01:58:53 it means nothing Jul 07 01:59:16 snapshot of glibc? of gnu-config? or umbaumba? Jul 07 01:59:24 hi mickeyl Jul 07 02:00:44 oh, hi mickeyl Jul 07 02:03:52 morning all Jul 07 02:05:28 i mean why does it try to download gcc-snapshot-cross Jul 07 02:05:40 if my local.conf file is correct. Jul 07 02:06:12 if it works once; it has to work the next time right ? Jul 07 02:07:24 Something funny is going on with some of the glibcs. They seem to need the headers copying over before they'll configure... Jul 07 02:07:47 The 2.3.2+cvsxxx is ok. 2.3.2 and 2.3.5 don't seem to be though :-/ Jul 07 02:10:02 my downloaded gcc-cross-initial is 3.4.4 Jul 07 02:12:12 i am a bit scared to give preferred versions ; because i dont know if they will work or not ?. gcc-cross="3.4.4" not sure if it works with 2.3.2 glibc Jul 07 02:12:28 try Jul 07 02:12:43 RP: 2.3.5-cvs builds on my machine at work, but not on my real build machine Jul 07 02:12:45 2.3.2+cvs20040726 builds fine with 3.4.3 3.4.4 Jul 07 02:12:47 for arm Jul 07 02:12:52 RP: and they are both running same debian vesion Jul 07 02:13:30 so shall i give gcc=3.4.4 and gcc-initial=3.4.4 Jul 07 02:13:33 hrw: ? Jul 07 02:13:57 http://pastebin.ca/17125 Jul 07 02:13:58 eJumbo: gcc-cross Jul 07 02:14:24 XorA: Could it be you had a previous glibc in staging? Jul 07 02:14:38 RP: Ive delete /tmp Jul 07 02:14:54 * XorA will re-attempt again just to make sure Jul 07 02:14:54 XorA: Strange... Jul 07 02:15:08 pb_: I think something funny is going on with the glibcs... Jul 07 02:15:25 where funny is defined as getting the preprocessor error I had last night Jul 07 02:15:28 good morning pb_ Jul 07 02:15:47 pb_: It seems to affect both 2.3.2 and 2.3.5 Jul 07 02:15:57 but not 2.3.2+cvsxxxx Jul 07 02:16:26 yes gcc-cross; Jul 07 02:16:36 If I had glibc headers already in cross/arm-linux/include, 2.3.2 then configures and built fine Jul 07 02:16:40 mickeyl: bitbake updated - will check "rebuild opie-today*" Jul 07 02:16:59 i have tested gcc-cross:3.4.1 bin utils 2.15.90 and glibc 2.3.2 it works fine ? Jul 07 02:17:08 hrw|work: build, not rebuild. Jul 07 02:17:13 rebuild doesn't feature it yet Jul 07 02:17:33 well, actually clean doesn't feature it yet Jul 07 02:17:39 which is getting called by rebuild Jul 07 02:17:49 ok Jul 07 02:18:14 i'm pretty confident it won't introduce problems, but i just want one other confirmation before adding it to the other commands Jul 07 02:18:23 morning Jul 07 02:18:26 morning proti Jul 07 02:18:32 hi proti Jul 07 02:18:35 morning proti Jul 07 02:18:57 mickeyl: http://www.frankengul.org/~seb/monotone-0.19+.tar.gz Jul 07 02:19:03 mickeyl: good morning! Jul 07 02:19:07 hi pb Jul 07 02:19:08 proti: ah excellent. thanks Jul 07 02:19:09 mickeyl: maybe do it in a way: "command regexp" -> split regexp to list and then do one-after-one? Jul 07 02:19:21 Don't know if it works. Compiling it at the moment. Jul 07 02:19:24 pb_: morning btw :) Jul 07 02:19:30 hi pb Jul 07 02:19:35 pb_: morning Jul 07 02:19:40 hi woglinde, proti, hrw|work, RP Jul 07 02:19:50 RP: remind me what the preprocessor error was? Jul 07 02:20:14 mickeyl: Your make.py refactoring makes the patch requested by zecke much harder. Jul 07 02:20:38 hrw|work: ya, that may be better than handling it in the individual commands, however splitting it to the list depends on the type of command (i.e. providers vs. files) so I'd end up adding more knowledge about the commands to the loop Jul 07 02:20:39 today is good day we will have a barbecue for the students sponsored bei IBM Jul 07 02:20:53 proti: why's that? what kind of patch did he request? Jul 07 02:21:01 checking how to run the C preprocessor rp... arm-linux-gcc -E Jul 07 02:21:01 configure: error: C preprocessor "arm-linux-gcc -E" fails sanity check Jul 07 02:21:01 FATAL: oe_runconf failed Jul 07 02:21:28 pb_: Ignore the rp bit - that was just me hacking configure to check exactly where it was... Jul 07 02:21:41 woglinde: sounds good Jul 07 02:21:49 mickeyl: He requested the manual marshalling and unmarshalling of the cfg for removing the deepcopy of linkdataSet. Jul 07 02:22:09 mickeyl dont if zecke already talked about the big truck Jul 07 02:22:13 +know Jul 07 02:22:22 woglinde: not to me Jul 07 02:23:10 proti: i don't see how the reintegration of make.py makes this any harder. besides, the refactoring was really necessary to reduce arbitrary complexity. Jul 07 02:23:21 mickeyl: Have a look at the pickle_prep and unpickle_prep in http://www.frankengul.org/~seb/cow/cow8a.patch Jul 07 02:23:24 pb_: Sadly I've lost the config.log as my last build worked (after copying the header files over). Basically, that test couldn't find assert.h Jul 07 02:24:01 pb_: The paths it was searching were the same as other working glibc builds. There were no headers in cross/arm-linux/include Jul 07 02:24:17 mickeyl: http://pastebin.ca/17126 (nice output - shouldn't it die after first failure?) Jul 07 02:24:24 mickeyl: I need to replace the reference of the parent dict in the data_smart by a string that reference it when pickling the dict. Jul 07 02:24:39 And the reverse when unpickling. Jul 07 02:24:56 RP: you could check with the nslu2 people. I think they use 2.3.2. Jul 07 02:25:17 Problem is that in data_smart, I don't know anything about cfg or the parent dataSet. Jul 07 02:25:29 hrw|lecture: hehe, oops. Jul 07 02:25:51 hrw|lecture: dunno. how would you like it to behave? Jul 07 02:25:59 pb_: I saw the same problem with 2.3.5... Jul 07 02:26:07 pb_: Should it need headers in cross/arm-linux/include ? Jul 07 02:26:13 Before, make.py was doing the pickle and unpickle calls. cfg was defined inside so no problem. Jul 07 02:26:35 mickeyl: it should stop after failure - like it is done when I do "build something" Jul 07 02:26:46 hrw|work: ok, i'll do it like that Jul 07 02:27:23 Now, pickle and unpickle are done in the dataSmartPackage. So I need a way to name dict references and to get them back. Jul 07 02:27:48 ~lat quilt for need of gawk Jul 07 02:27:53 ~lart quilt for need of gawk Jul 07 02:27:53 * ibot puts quilt into a headlock and administers a mighty noogie, rubbing half of quilt's hair of for need of gawk Jul 07 02:28:21 I was thinking of creating the equivalent of pkgdata inside dataSmart, a private var that holds all the dict of every bbfiles and the cfg. Jul 07 02:29:07 proti: hmm, do we really need yet another collection of these data? I have the feeling we already use too many different duplicated variables Jul 07 02:29:19 mickeyl: SHELL: Building opie-today-todolistplugin opie-today-fortuneplugin opie-today-addressbookplugin opie-today opie-today-mailplugin opie-today-weatherplugin opie-today-stocktickerplugin opie-today-datebookplugin... Jul 07 02:29:26 mickeyl: what for "..." at the end? Jul 07 02:29:37 hrw|work: ... = that'll take a while :) Jul 07 02:29:38 mickeyl: This time, it will be private. Just a way of tracking all dataSet we create. Jul 07 02:30:02 mickeyl: all bitbake users know that it can a while Jul 07 02:30:49 hrw|work: heh, ok. I remove the dots. Jul 07 02:32:53 mickeyl: BB>parse;build opie-today;setenv CVSDATE other-date-then-it-was;parse;build opie-today == nothing to build Jul 07 02:33:10 mickeyl: My problem is if I link dataSet A -> B, if I want to pickle dataSet B but not dataSet A, I need to break the link A -> B and replace it by the info that dataSet A was linked to B explicitely. And then when unpickling the dataSet B, I need to reestablish this link with the info provided in the dataset B (which is this dataSet is linked to A). Jul 07 02:33:41 hrw|work: ah. there's an update of the outer conf files necessary. could you add that to bt pleaes? Jul 07 02:34:08 ok Jul 07 02:34:13 thanks Jul 07 02:34:43 proti: i see. can't you do that with a dictionary using a hashvalue of a dataset? Jul 07 02:35:46 hrw|work: oops, i think wildcards already work in the version I've commited for all things that build providers Jul 07 02:36:07 grr Jul 07 02:36:09 * pb_ stabs linux Jul 07 02:37:34 I hope nobody's in London :-/ Jul 07 02:38:01 RP: whats happening? Jul 07 02:38:27 mickeyl: Exactly. So I need a table to track all the dataSet that are used. Jul 07 02:38:28 XorA: Reports of mutliple bus and train explosions Jul 07 02:38:53 XorA: Although no real info and other claim its a power surge etc... Jul 07 02:41:07 sounds like a terrorist attack :( Jul 07 02:42:59 mickeyl: monotone 19+ doest not compile. It barfs on decode_gzip. I've too little knowledge in C++ to fix that. Jul 07 02:43:12 i'll try Jul 07 02:43:23 mickeyl: #136 Jul 07 02:44:56 hrw|work: That's because pkgdata of the bbfile holds a copy of cfg and you are modifiing the cfg itself I think. Jul 07 02:45:44 ya. i would need to update pkgdata after a call to setenv Jul 07 02:45:59 Weel, maybe not. something to do with modifiying the copy and not the value really used. Jul 07 02:46:07 proti: perhaps - but that should work Jul 07 02:46:41 03mickeyl 07org.openembedded.dev * r2109e112... 10/conf/distro/ (openzaurus-3.5.2.conf openzaurus-3.5.3.conf): remove old openzaurus distribution conf files Jul 07 02:46:52 hrw|work: Modifying the cvsdata implies rebuilding the depgraph and providers. Jul 07 02:47:17 s/cvsdata/cvsdate/ Jul 07 02:47:18 proti: I can do 'parse' after setenv Jul 07 02:48:11 hrw|work: And what the findBestProvider says after ? Jul 07 02:48:39 proti: how to test it? (I'm not familiar with bitbake internals) Jul 07 02:49:03 the problem is that when you reparse, the information about CVSDATE gets reloaded from local.conf, instead of your local copy Jul 07 02:49:19 local copy in memory, that is Jul 07 02:49:31 ah.. Jul 07 02:50:30 so i would need to inject the new local environment into the pkgdata Jul 07 02:50:35 which sounds doable Jul 07 02:50:40 just something i need to think about Jul 07 02:50:50 RP: well removed /tmp glibc still doesnt compile :-( Jul 07 02:52:24 mickeyl: Why not do an update_data, update_keys after the setenv ? Jul 07 02:52:30 i had a problem with 2.6.12 kernel. the file system was getting corrupted. but i am not getting the problem now. the earlier file system i was using was the filesystem i got when i did gpe-image. now i have made a small filesystem by hand and testing it and i am not getting any filesystem corruption. has anyone got jffs2 file system corruption (u can see those CRC errors and #ino messages during the next reboot) Jul 07 02:54:08 proti: sounds like an idea. like i told you previously, i have no idea about the smartData data type. Jul 07 02:54:22 all i have been doing is to poke into dictionaries until now Jul 07 02:54:43 i hope it's documented Jul 07 02:55:28 mickeyl: It's not smart data. It is the way bb is working. After parsing a bbfile, you have a update_data and update_keys called. See the parse/BBHAndler.py. Jul 07 02:55:49 ah the BBHandler. black magic to me as well. could you please add that suggestion to #136? Jul 07 02:58:54 once bitbake 1.x has settled down a bit i'll start with a graphic frontend again Jul 07 02:59:43 eJumbo: we have "This is not a distribution support channel" in topic Jul 07 03:00:49 Ah, sorry, it was expandKeys. update_data translate the ${azer} the dict to the real value. The expandKeys do the same thing for the keys of the pkgdata it expands the name and create a new entry if the name is different. Jul 07 03:01:07 mickeyl: OK. will do. Jul 07 03:01:42 thanks Jul 07 03:03:06 mickeyl: I was thinking about putting the name of the bbfile in the bbfile entry. Does it sound reasonnable ? Jul 07 03:04:03 proti: definitly. Jul 07 03:04:04 In the pkgdata dict, I mean. Jul 07 03:04:25 there are times when you need to quickly find out which file the data are in Jul 07 03:04:36 s/are in/are derived from/ Jul 07 03:04:50 Ok, I removes a lot of hurdle. picklinkg a linked dataSet comes really simple with this assumption. Jul 07 03:04:57 s/I/it Jul 07 03:07:44 hi Paul Jul 07 03:08:01 mickeyl: Ok comment added. Will try a patch if I have time. Jul 07 03:08:33 hi all Jul 07 03:08:35 hi hrw|work Jul 07 03:09:45 proti: excellent Jul 07 03:09:48 hi bluelightning Jul 07 03:09:58 hey mickeyl Jul 07 03:10:14 database.cc:1059: error: no matching function for call to 'decode_gzip(std::string&, std::string&)' Jul 07 03:10:21 that's indeed odd Jul 07 03:10:34 sure you applied all patches? Jul 07 03:12:56 mickeyl: look, decode_gzip is defined as template in transforms.hh. Jul 07 03:13:35 Which is included in database.cc. Jul 07 03:14:57 Why would bitbake build glibc-2.3.2, then tell me 2.3.2 wasn't available and build 2.3.5 ? :-/ Jul 07 03:15:32 sounds like a glibc-initial vs. glibc thing Jul 07 03:16:19 RP: Sure glibc-2.3.2 was build correctly ? Jul 07 03:16:37 mickeyl: in bitbake shell when I'm "build one_thing" then it is not needed to say "SHELL: Building one_thing..." - that SHELL: info is usefull if many are getting build... or no - ignore that.. if I want to build opie-sysin* then only opie-sysinfo is available so it has sense Jul 07 03:16:39 proti: Yes :-/ Jul 07 03:18:20 RP: Don't know then. mickey's explanation should be the right one. Jul 07 03:20:46 hrw|work: i can change it to only print SHELL: building, if we actually have received a glob expression Jul 07 03:21:03 RP: can you paste the exact output from bitbake here? Jul 07 03:24:04 03mickeyl * r278 10bitbake/lib/bb/shell.py: Jul 07 03:24:04 Shell: Jul 07 03:24:04 - remove '...' in output for "SHELL: Building " Jul 07 03:24:04 - don't continue if building a number of providees and one fails Jul 07 03:24:08 pb_: Will do once it repackages 2.3.2 and hopefully does it again... Jul 07 03:24:34 okay Jul 07 03:26:31 Should it be running glibc-initial before glibc? Jul 07 03:26:44 Yes. Jul 07 03:27:23 That would be problem number one then... Jul 07 03:27:25 The toolchain build goes glibc-initial -> gcc-initial -> glibc{-intermediate} -> gcc { -> glibc } Jul 07 03:27:56 For non-TLS builds, glibc-intermediate isn't used and the final glibc is built after gcc-initial. Jul 07 03:28:30 I have no glibc-initial being built... Jul 07 03:28:51 glibc itself doesn't require glibc-initial. Jul 07 03:29:13 glibc does require gcc{-initial}, though, and glibc-initial is normally required to build gcc-initial. Jul 07 03:29:31 gcc-initial or gcc-cross-initial ? Jul 07 03:29:36 gcc-cross-initial Jul 07 03:29:59 That is built... Jul 07 03:30:35 mickeyl: #136 depends on #135. If I do expandkeys on every packages in pkgdata, it will change nothing in the pkdata since the value will be picked from the cfg copy embedded in the dict, not the one that has been modified. One need to modify the original and propagate the modification in every package. Jul 07 03:30:39 (I'm using gcc 3.4.3, not that it should make much difference) Jul 07 03:33:08 proti: *nod*. I added the dependency Jul 07 03:37:06 pb_: http://pastebin.ca/17131 is the output I could catch from bitbake Jul 07 03:37:32 (Except 2.3.5 does start from the beginning) Jul 07 03:38:53 Which the log does show. I'm just getting confused :) Jul 07 03:39:06 okay, that must be a bitbake bug. Jul 07 03:39:27 I think zecke and proti have been poking at that code most recently. Jul 07 03:41:06 I suspect there are two bugs here - 1. glibc-initial isn't getting triggered and 2. the above provider version issue... Jul 07 03:41:24 Regarding the glibc-initial thing, you should check gcc-cross-initial's DEPENDS. Jul 07 03:57:05 gcc-cross-intial's depends get trampled on somewhere... Jul 07 03:57:59 RP: fixed my glibc problem finally :-) Jul 07 03:58:25 XorA: What was it? Jul 07 03:58:55 RP: ubuntu gcc, I thought Id replaced it with debian etch, but I hadnt Jul 07 04:02:30 XorA: Aha :) Jul 07 04:08:35 hi xora Jul 07 04:08:54 okay, i'm poking around in OE and have some questions, this is the right place to ask? Jul 07 04:09:12 hi woglinde Jul 07 04:09:33 giel here? Jul 07 04:09:43 giel oh yes Jul 07 04:09:48 okay Jul 07 04:10:06 so, i'm trying to build an as-simple-as-possible root filesystem for a generic arm device Jul 07 04:10:24 bitbake bootstrap-image Jul 07 04:10:27 i'm bitbaking bootstrap-image, that seemed reasonable to me Jul 07 04:10:41 pb_: An illustration of the problem is: http://www.rpsys.net/openzaurus/temp/rp3 Jul 07 04:11:07 That bit of inline code doesn't generate the glibc-inital dependency for some reason... Jul 07 04:11:12 an now, when it's almost done, ipkg starts acting up Jul 07 04:11:21 + ipkg-cl -f /home/giel/oe/build/tmp/work/bootstrap-image-1.0-r0/temp/ipkg.conf Jul 07 04:11:21 -o /home/giel/oe/build/tmp/rootfs update Jul 07 04:11:21 ipkg: Cannot create directory `/usr/lib/ipkg': Permission denied Jul 07 04:11:36 You can ignore that error Jul 07 04:11:43 okay Jul 07 04:11:47 Morning Jul 07 04:11:50 but still, it doesn't work Jul 07 04:11:54 Look for another one Jul 07 04:11:59 hi Sirfred Jul 07 04:12:01 An error ocurred, return value: 1. Jul 07 04:12:01 Collected errors: Jul 07 04:12:01 ERROR: Cannot satisfy the following dependencies for task-bootstrap: Jul 07 04:12:01 base-files base-passwd busybox initscripts netbase sysvinit sysvinit-pi Jul 07 04:12:01 dof tinylogin modutils-initscripts fuser setserial linux-hotplug ipkg-collateral Jul 07 04:12:03 ipkg ipkg-link module-init-tools-depmod Jul 07 04:12:17 giel: That looks more likely Jul 07 04:12:21 konqueror-embedded-20030705 failed in compile stage. Jul 07 04:12:23 args the depmod error Jul 07 04:12:39 I think the problem is: Jul 07 04:12:39 sirfred with pcre error? Jul 07 04:12:41 *** No rule to make target `-lpcre', needed by `konqueror'. Stop. Jul 07 04:12:45 woglinde: Exactly. :) Jul 07 04:12:53 does it have something to do with depmod? 'cause i was fighting with that earlier Jul 07 04:13:02 it tried to compile modutils instead of module-init-tools Jul 07 04:13:10 sirfred yes i have same problem but it cant be reprduced on other machines Jul 07 04:13:19 giel: what MACHINE? Jul 07 04:13:32 sirfred go to the Makefile.in an delete the LIBPRCE entry Jul 07 04:13:36 woglinde: I havent updated my tree in a pair of days. Jul 07 04:13:40 woglinde: OK Jul 07 04:13:55 hrw|work: it's only tune-xscale for now Jul 07 04:14:02 hrw|work: maybe i should just remove that as well Jul 07 04:14:10 sirfred whats your buildsystem? Jul 07 04:14:21 hrw|work: i'm building for a machine not known to OE Jul 07 04:14:37 woglinde: My build machine? Jul 07 04:14:42 RP: maybe it's seeing the wrong TARGET_ARCH. Jul 07 04:14:44 woglinde: This line? Jul 07 04:14:47 LIBPCRE = @LIBPCRE@ Jul 07 04:15:08 giel: so it does not know which kernel to use so does not know does need modutils or module-init-tools Jul 07 04:15:22 sirfred right Jul 07 04:15:33 pb_: TARGET_ARCH is set to arm... Jul 07 04:15:35 sirfred yes your machine Jul 07 04:15:39 woglinde: I'm on a ubuntu hoary linux Jul 07 04:15:43 hm Jul 07 04:15:58 hrw|work: hm, sounds reasonable, but i've kinda fixed the modutils error, but now it gives me the error shown above when doing do_rootfs Jul 07 04:16:41 RP: at the point where the expansion runs? Jul 07 04:18:16 RP: oh, sorry, I was reading the code wrong. Never mind. Jul 07 04:18:37 giel: that packages was even build? Jul 07 04:19:20 hrw|work: yep, they're all in deploy/ipk Jul 07 04:19:27 RP: it looks like glibc-initial is not required for arm and mips. Jul 07 04:20:56 Ah, right. So it doesn't need to build that afterall... Jul 07 04:21:01 giel: hmm.. never tried that kind of build Jul 07 04:21:24 yeah, indeed. I guess linux-libc-headers suffices on those platforms. Jul 07 04:21:28 hrw|work: just to be sure i'll remove my tmp dir and start all over again, i've been messing around a bit Jul 07 04:22:13 or maybe even that isn't needed for the -initial build, actually. Jul 07 04:22:41 pb_: Except it appears some headers are needed to keep configure happy :-/ Jul 07 04:23:05 On some versions of glibc anyway... Jul 07 04:24:19 That's a separate issue. If glibc needs those headers, it should say so in its own DEPENDS. Jul 07 04:25:14 So its probably a missing header depends... Jul 07 04:25:25 Probably. Jul 07 04:39:02 zecke__: are you there? Jul 07 04:44:33 zecke__, mickey|lecture: can you confirm that collections are working properly in bitbake at the moment? They seem to be broken for me. I suspect that some of the changes that zecke123 made around r236/r238 may be to blame, but I haven't been following that code very closely. Jul 07 04:50:44 woglinde: Which one of the one thousand makefiles konqueror has, need I to change? Jul 07 04:51:16 Sirfred: install libpcre-dev on host - it helps sometimes Jul 07 04:51:29 hrw|work: Thanks, I'm gonna try. Jul 07 04:54:07 pb_: The linux-libc-headers are included and that doesn't seem to be enough :-/ Jul 07 04:56:09 hrw|work: It worked! Thanks. Jul 07 04:57:28 hey mickey|lecture, you about? Jul 07 05:01:09 np Sirfred Jul 07 05:03:49 RP: okay. well, feel free to adjust the depends to call in glibc-initial, if that turns out to be necessary. Jul 07 05:04:08 I'm not sure why it is inhibited for those two architectures; maybe just to reduce build time or something. Jul 07 05:12:14 mickey|lecture: you have email :P Jul 07 05:21:38 okay, rebuilding everything kinda worked Jul 07 05:21:46 i have a rootfs now, but bbimage is complaining Jul 07 05:23:45 and although i know a reasonable amount of python, this error seems kinda strange to me: Jul 07 05:24:12 | + bbimage -n bootstrap-image--20050707112412 -t ext2.gz -e /home/giel/oe/org.openembedded.dev/packages/meta/bootstrap-image.bb Jul 07 05:24:16 | Traceback (most recent call last): Jul 07 05:24:18 | File "/home/giel/oe/bitbake/bin/bbimage", line 23, in ? Jul 07 05:24:21 | from bb import * Jul 07 05:24:23 | AttributeError: 'module' object has no attribute 'make' Jul 07 05:36:16 okay Jul 07 05:36:40 apparently, the line "make" in bitbake/lib/bb/__init__.py shouldn't have been there Jul 07 05:36:53 weird, is this known or is something wrong with my setup? Jul 07 05:39:53 ah, nm, checked out a newer version and it was fixed Jul 07 05:39:58 the "make" was removed Jul 07 05:41:18 giel: update bitbake Jul 07 05:41:52 uhu... don't mind me too much, i always solve stuff by just typing out my problems... maybe i should join an empty channel :) Jul 07 05:42:43 giel: you have to update bitbake - make in bbimage was fixed Jul 07 05:44:20 hrw|work: yeah, i just found out, thanks Jul 07 05:44:38 hrw|work: modified local.conf so that gcc and gcc-intial takes version 3.4.4 but still it is taking gcc-snapshot-cross Jul 07 05:44:55 eJumbo: pastebin it Jul 07 05:47:35 http://pastebin.com/308870 Jul 07 05:50:25 eJumbo: local.conf Jul 07 05:53:56 I've made a 'bitbake opie-image' Jul 07 05:54:28 I found that there's an image in ~/tmp/work/opie-image-1.0-r14/temp/opie-image-c7x0-20050707063648.rootfs.jffs2 Jul 07 05:54:35 Is that the correct location for it to be generated? Jul 07 05:54:42 what filesystem should use in a CF? XIP or a compressed filesystem? Jul 07 05:54:53 i think it's supposed to be in tmp/deploy/images Jul 07 05:55:02 giel: Thanks, I'm going to see Jul 07 05:55:34 giel: You're right. Thanks. Jul 07 05:55:40 Sirfred: np :) Jul 07 05:56:06 I suppose than the updater.sh is the same from older openzaurus releases. Jul 07 05:56:23 Sirfred: yes - as c7x0 have magic header added into it Jul 07 05:56:38 Sirfred: updater.sh should be in tmp/deploy/images/ Jul 07 05:56:48 hrw|work: It's not there. Jul 07 05:56:55 hrw|work: Perhaps I need to bitbake something to generate it? Jul 07 05:57:03 Sirfred: machine='c7x0' and not generated it? Jul 07 05:57:09 hrw|work: Let me see. Jul 07 05:57:12 03tmbinc 07org.openembedded.dreambox * rc8335c77... 10/packages/dreambox/ (15 files in 5 dirs): add dreambox specific tools Jul 07 05:57:13 03tmbinc 07org.openembedded.dreambox * rab3db230... 10/ (69 files in 34 dirs): Jul 07 05:57:13 propagate from branch 'org.openembedded.dev' (head 305e6fa72a587cdf917f2be1399e3d7119271662) Jul 07 05:57:13 to branch 'org.openembedded.dreambox' (head c8335c776e4c1fcccf598ef3c341c835da20efcf) Jul 07 05:57:29 Sirfred: build zaurus-updater then Jul 07 05:57:39 hrw|work: Yes, MACHINE="c7x0" Jul 07 05:59:42 strange.. it should built it Jul 07 06:01:07 mickey|lecture: why zaurus-updater is not in BOOTSTRAP_DEPENDS_EXTRA of >collie Z confs? Jul 07 06:01:26 hrw|work: Thanks, that made it. Jul 07 06:02:03 Sirfred: ok Jul 07 06:02:33 hrw|work: I suppose this is just a binary, I mean, we have no source code for this. Jul 07 06:03:41 hrw: http://pastebin.com/308882 Jul 07 06:04:01 Sirfred: its a shell script which is encoded by encoder which we have in OE Jul 07 06:04:03 was doing a bitbake clean and take the new output at http://pastebin.com/308882 Jul 07 06:04:19 hrw|work: Oh, I see, the encdec-updater-native Jul 07 06:08:36 mickey|lecture: ping Jul 07 06:11:28 hrw|work:ping Jul 07 06:11:40 eJumbo: pong - trying with your local.conf Jul 07 06:17:00 hi drw Jul 07 06:17:08 re Jul 07 06:17:45 eJumbo: http://pastebin.ca/17145 thats what I have with your local.conf (bitbake -n gcc-cross) Jul 07 06:18:07 hi zecke Jul 07 06:18:07 hi hrw|work Jul 07 06:19:14 pong Jul 07 06:21:10 mickeyl: spitz.conf has IMAGE_FSTYPES = ;( Jul 07 06:21:32 mickeyl: and none of updater.sh machines have it in BOOTSTRAP_DEPENDS_EXTRA. Jul 07 06:24:36 mickeyl: both looks strange for me Jul 07 06:25:14 w00h00, success! Jul 07 06:25:21 i booted my device with the rootfs i just built Jul 07 06:25:29 giel: great Jul 07 06:25:30 now i can identify stuff for in the machine.conf Jul 07 06:25:48 that'll teach all those montavista adepts :) Jul 07 06:25:51 hrw: it worked on ur system. what should i do it make it work ? Jul 07 06:26:01 i dont know where the problem is ? Jul 07 06:27:33 eJumbo: I just get your local.conf and tried Jul 07 06:27:47 it does not work for me. Jul 07 06:28:15 lets analyse first: what packages i should clean first Jul 07 06:28:59 i did bitbake -cclean gcc-snapshot-cross and bitbake -cclean gcc Jul 07 06:30:03 hrw|work: ok, i'll take care about that Jul 07 06:31:27 i still cant understand why bitbake behaves so, on my machine Jul 07 06:36:40 mickeyl: o coremaster, did you see my question about collections? Jul 07 06:37:53 ~hail pb_ Jul 07 06:37:54 * ibot bows down to pb_ and chants, "I'M NOT WORTHY!!" Jul 07 06:42:39 03tmbinc 07org.openembedded.dreambox * rffd827ab... 10/packages/links/links-dream_2.1pre15.bb: add links plugin for dreambox Jul 07 06:46:28 I think i fixed a bug in libpng-1.2.8.bb Jul 07 06:47:12 as in, the one from the repo didn't compile for me, and now it does Jul 07 06:47:40 giel: good Jul 07 06:47:47 what did you change? Jul 07 06:47:56 bugs.openembedded.org Jul 07 06:48:28 tmbinc: well, it couldn't find zlib.h because the include dir wasn't set correctly because the CFLAGS got overridden by the bb Jul 07 06:48:45 tmbinc: unset CFLAGS did the trick, but dirty, so I guess a CFLAGS_append is more in order Jul 07 06:50:25 giel: isn't zlib.h in STAGING_INCDIR? Jul 07 06:50:43 zecke__: yes, it is Jul 07 06:50:55 zecke__: but STAGING_INCDIR didn't get included Jul 07 06:52:34 that "ZLIBINC=${STAGING_INCDIR}" doesn't do anything? Jul 07 06:53:04 well, it only really gets used in the CFLAGS=-I{ZLIBINC} in the makefile Jul 07 06:53:10 but CFLAGS get overridden Jul 07 06:53:15 ah Jul 07 06:53:37 hm, still, why isn't -I{STAGING_INCDIR} always in my cflags? Jul 07 06:53:40 i know it's not really relevant, but libpng builds for me. what am i doing wrong? Jul 07 06:54:12 i think i might be doing something wrong all along Jul 07 06:54:20 hi tmbinc Jul 07 06:54:29 'cause i got the same sort of problem with the next package Jul 07 06:54:40 hi wogi :) Jul 07 06:55:46 bye all Jul 07 06:58:25 ah, i think it's me Jul 07 06:58:29 trying again... Jul 07 07:03:15 yep, it was me, sorry guys Jul 07 07:03:30 i had a TARGET_CFLAGS = "-O1" in my local.conf :S Jul 07 07:03:48 changed it to _append and it works now Jul 07 07:04:09 i needed that for compiling glibc succesfully, i should look into that as well Jul 07 07:04:22 hail zecke__ Jul 07 07:05:14 (in CF card) i'm in doubt of use a compressed fs. or write directly the system to CF? Jul 07 07:05:32 Do I need to delete tmp every I make changes in org.openembedded.dev/packages ? Jul 07 07:06:55 micropal: no Jul 07 07:13:12 zecke__: do you have any thoughts on my collections problem? Jul 07 07:13:56 zecke__: the code involving "tmp_pn" in findBestProvider() looks somewhat suspicious to me. Jul 07 07:14:51 hrw|meeting: Finally my gcc-cross worked. i dont know why ?. But when i do a bitbake bootstrap-image i get a new error:http://pastebin.com/308897 Jul 07 07:15:06 pb_: first asking mickeyl, now coming back to me... Jul 07 07:15:07 *fine* Jul 07 07:15:14 zecke__: heh. well, mickeyl seems to ignore me. Jul 07 07:15:21 so do I Jul 07 07:15:23 drat Jul 07 07:15:40 pb_: It could be broken Jul 07 07:15:58 pb_: and if it does not work for you it is definately broken Jul 07 07:17:22 pb_: can you spare a minimal collection? One additional (patched) package and a config file? Jul 07 07:17:58 eJumbo: please add space before http:// Jul 07 07:18:00 zecke__: I'll construct a testcase Jul 07 07:18:10 sorry Jul 07 07:18:48 base-passwd???? eJumbo - you have screwed system or something.. Jul 07 07:18:49 hrw|work: my gcc-cross worked dont know how and why ?. now have a new problem : http://pastebin.com/308897 Jul 07 07:20:24 eJumbo: seriously, you should "mt revert" your whole org.openembedded, erase all your cache files, re-setup your environment and retry. Jul 07 07:20:36 and erase tmp/, of course :) Jul 07 07:21:08 eJumbo: you have TMPDIR on noexec or nosuid etc partition? Jul 07 07:22:52 i am not following you hrw: cant make out what you are saying Jul 07 07:23:23 shall i delete TMPDIR and start all over again or do you want to to delere Jul 07 07:23:39 openembedded, packages, sources too Jul 07 07:25:15 this is the second time i would be doing it Jul 07 07:25:42 eJumbo: do yo have public IP and can get me account there? I can try then at your system - I'm out of ideas now Jul 07 07:26:10 no public ip Jul 07 07:27:03 just check the error url at paste.bin it is showing build-type --build=i686-linux is it ok ?. Jul 07 07:27:13 my target is arm : omap 5912 Jul 07 07:28:49 Opie looks great with the new fonts. Jul 07 07:30:26 Sirfred: ttf ones? yes Jul 07 07:31:00 hrw|work: Yes, I've just installed a today snapshot. Jul 07 07:31:29 RP: Why not the fastsysclk is set to 1 as default? I think the image is better. Jul 07 07:33:30 What I don't have is the eye boot picture, just plain text on boot. Jul 07 07:33:52 I have a kernel 2.6.12-mm1, is that correct? Jul 07 07:34:19 Sirfred: if you want ELPP then you have to enable it in local.conf Jul 07 07:34:46 hrw|work: Is that only affecting to the kernel or also to the rootfs ? Jul 07 07:34:56 Sirfred: kernel Jul 07 07:35:47 hrw|work: How could I enable ELPP ? Jul 07 07:36:13 Sirfred: look into linux-openzaurus_2.6.12-mm1.bb Jul 07 07:36:34 hrw|work: OK, thanks! Jul 07 07:37:27 pb_: sorry, i didn't see your collection question. i'll scroll up Jul 07 07:37:47 ENABLE_ELPP = ${@bb.data.getVar("OZ_KERNEL_ENABLE_ELPP",d,1) or "no"} Jul 07 07:37:58 So, I assume that I have to put: OZ_KERNEL_ENABLE_ELPP=1 ? Jul 07 07:38:21 Well, OZ_KERNEL_ENABLE_ELPP = "1" Jul 07 07:39:07 mickeyl: when do we upgrade to monotone .20 on the server? Jul 07 07:39:12 pb_: ok, i've read it now. I have no idea frankly, I never tried using collections. Could you construct a small test case? Jul 07 07:39:36 zecke__: I'm not sure yet. I need time to test if all my scripts still run with .20 Jul 07 07:39:45 pb_: now you prefer mickeyl again? screw you ;) Jul 07 07:39:54 heh, he doesn't Jul 07 07:40:04 actually i've just read the backlog Jul 07 07:40:18 Sirfred: yes Jul 07 07:40:46 zecke__: did you use collections yet? Jul 07 07:40:57 I'm impressed with the fonts. The console is great. Jul 07 07:41:08 But it seems to have some bug with the tab management. Jul 07 07:41:23 Sirfred: yeah, console's tab management is broken. could you submit a bug please? Jul 07 07:41:29 opie-bugs.oszine.de Jul 07 07:41:32 mickeyl: OK. Jul 07 07:41:34 thanks Jul 07 07:41:41 mickeyl: no problem. Jul 07 07:42:29 Also, I think that the icons in konqueror need a revision. Jul 07 07:42:39 all icons need a revision Jul 07 07:42:41 That's not a bug, but... Jul 07 07:42:59 mickeyl: But konqueror ones are more needed, I think. Jul 07 07:43:21 true enough, however no one is working on konqueror/e 2.x anymore Jul 07 07:43:37 wb zecke Jul 07 07:43:47 mickeyl: Is there any other alternative for a web browser on opie? Jul 07 07:43:49 lol our own mail server is black listing us... Jul 07 07:43:57 Sirfred: lynx, links, w3 Jul 07 07:44:11 zecke: On opie, I mean. Jul 07 07:45:07 Sirfred: http://hrw.one.pl/tmp/sc_Thu_Jun_9_22.04.00_2005.png you mean (tabbar) Jul 07 07:45:12 Sirfred: ? Jul 07 07:45:19 ya, that's it Jul 07 07:45:23 i've seen that before Jul 07 07:45:31 it doesn't occur on collie Jul 07 07:45:34 hrw|work: Yes. Jul 07 07:45:44 mickeyl: it need >qvga Jul 07 07:47:57 OTabbar is small changed QTabbar - so bug is in qt/e? Jul 07 07:48:20 i don't think opie-console is using otabbar Jul 07 07:48:26 or is it? Jul 07 07:48:31 aha Jul 07 07:48:37 didnt lok Jul 07 07:49:10 neither otabbar nor qtabbar Jul 07 07:49:18 something hand crafte Jul 07 07:49:18 d Jul 07 07:49:28 mickeyl: Console is a core or a non core app? Jul 07 07:49:31 OCTabbar ;) Jul 07 07:49:33 mickeyl: A non core I guess, isn't it? Jul 07 07:49:34 Sirfred: non Jul 07 07:49:35 ah QTabWidget Jul 07 07:49:43 opie/cvs/noncore/apps/opie-console Jul 07 07:49:50 Sirfred: noncore, but this distionction no longer makes sense, really Jul 07 07:49:54 distinction, ven Jul 07 07:49:56 even Jul 07 07:50:01 ~lart my keyboard Jul 07 07:50:01 * ibot pries my keyboard's back open with a screwdriver and flashes a new bootldr to my keyboard Jul 07 07:50:37 someone should look if things change when OTabWidget is used Jul 07 07:50:49 *sigh* soo many bugs, so little time and devels ;) Jul 07 07:55:13 mickeyl: Why don't you clone yourself ? mickeyl.createCopy() or mickeyl.newthread() ? Jul 07 07:56:20 #ifdef FSCKED_DISTRI Jul 07 07:56:23 :)) Jul 07 07:56:40 of course :) Jul 07 07:57:08 heh Jul 07 07:57:28 hrw|work: oops :) Jul 07 07:58:03 some strange contractor work is visible in opie-console Jul 07 07:58:18 yep.. ifdef EAST Jul 07 07:58:46 10east hired zecke and harlekin to work on opie-console Jul 07 07:58:58 iirc Jul 07 07:59:07 mickeyl: opie-console use TabWidget which is derive from OPIE::UI::OTabWidget Jul 07 07:59:33 mickeyl: monotone, I found another patch I didn't apply but seems to modify the problem you and I encountered. Jul 07 08:00:06 proti: ah. sorry, i didn't get back to you ealier, but I got it to compile Jul 07 08:00:21 proti: do you have an updated tarball? Jul 07 08:00:30 Harg!!, how ! Is it working ? Jul 07 08:00:56 proti: well it's probably missing a patch then, but it seems to work Jul 07 08:01:02 vanille.de/temp/ Jul 07 08:01:11 mickeyl: No, the package is not working. It complains about a sqlite3_complete_last missing. Jul 07 08:02:56 This function is defined in a directory sqlite in the monotone distribution but not in the debian sqlite. Jul 07 08:03:32 mickeyl: It's the bug 000168 Jul 07 08:03:33 mickeyl: It's the bug 0001682 Jul 07 08:03:37 k Jul 07 08:03:54 mickeyl: I've just generated an opie-image and opie is not starting. Jul 07 08:04:09 proti: I'm on a pretty recent Mandriva. it compiled there without problems after patching the gzip issue Jul 07 08:04:11 mickeyl: I'm afraid that perhaps my patches have broken it. Jul 07 08:04:51 * france is away: Away Jul 07 08:04:58 Sirfred: I will add comment and snaps to it Jul 07 08:05:08 hrw|work: I put an snapshot. Jul 07 08:05:19 hrw|work: Didn't get uploaded? Jul 07 08:05:22 mickeyl: maybe the sqlite dir is a private copy that is used whenever the libsqlite3.h is not found. I'm trying to rebuild it without the libqlite3-dev package. Jul 07 08:05:42 hrw|work: Yes, it is Jul 07 08:05:49 proti: yeah Jul 07 08:06:38 Sirfred: it is - I want to add my shots so comment from me will be complete Jul 07 08:06:47 hrw|work: of course Jul 07 08:06:57 My problem is that the sun machine is so slooooowwwww. c++ is even worst. Jul 07 08:08:05 Sirfred: look now Jul 07 08:08:29 mickeyl: Yes, I'm getting a SIGSEGV. Jul 07 08:09:25 Sirfred: sounds like we need a traceback Jul 07 08:09:42 mickeyl: I'm going to build a gdb Jul 07 08:09:53 hrw|work: Good work. Jul 07 08:10:03 excellent. + a debug build of libopie2 + qte Jul 07 08:10:12 and possibly the launcher as well Jul 07 08:10:18 mickeyl: Hmm. Jul 07 08:10:30 mickeyl: I had bad experiences with that in the past. Jul 07 08:10:51 qte with debug is huge Jul 07 08:10:52 how else could we obtain a traceback? Jul 07 08:10:56 mickeyl: gdb running a debug version of qte + qpe + libopie can be sloooow Jul 07 08:11:17 mickeyl: Well, we don't need a full debug version. Jul 07 08:11:23 mickeyl: Only a non stripped one. Jul 07 08:13:28 deleted TMPDIR and started all over again : got an error : http://pastebin.com/308933 Jul 07 08:13:34 hrw|work: ping Jul 07 08:14:11 mickeyl: It's strange, my "development" version of libqte don't SIGSEGV Jul 07 08:14:24 mickeyl: Neither using the Transformed screen. Jul 07 08:16:43 eJumbo: I give up - your system is fucked or broken in some strange way Jul 07 08:19:41 Sirfred: do you know gdb-cross? Jul 07 08:19:53 Sirfred: and catchsegv can be handy as well Jul 07 08:20:00 zecke: No. Jul 07 08:20:05 zecke: #135 I don't get the 'base' replacement backend thing. could you elaborate ? Jul 07 08:20:32 proti: What we do now is the following Jul 07 08:20:58 cu all Jul 07 08:21:07 hrw|work: cu Jul 07 08:21:17 hrw|gone: cu Jul 07 08:21:17 proti: parse conf/local.conf and for each bbfile from BBFILES we say get me file backed data instance using make.cfg as basis Jul 07 08:21:42 proti: now the file gets parsed and operators like := get immediately evaluated Jul 07 08:22:15 proti: if we later rerun bitbake, we will use the cached version of the file backed bbfile Jul 07 08:22:22 proti: BUT inject a new make.cfg Jul 07 08:22:34 proti: and this has to fail and is not a good idea at all Jul 07 08:22:52 proti: things can really get inconsistent Jul 07 08:23:21 * mickeyl needs to do his weekly housecleaning :/ Jul 07 08:23:27 proti: for example if someone does CVSDATE_NEW := ${CVSDATE}+1 Jul 07 08:23:47 proti: and later he assumes \delta CVSDATE_NEW to CVSDATE is one Jul 07 08:24:05 proti: and he rebuilds later this assumption turned wrong Jul 07 08:24:25 the first objection : CVSDATE = CVSDATE + 1 gets you the correct behaviour. It is updated correctly when you change the make.cfg. Jul 07 08:24:50 proti: how? Jul 07 08:25:05 proti: setVar('CVSDATE'...) will unlink make.cfg right? Jul 07 08:25:13 proti: and it will stay unlinked? Jul 07 08:25:30 proti: or do we want to save how many times something got unlinked? Jul 07 08:25:39 zecke: No immediate assignement. The content of CVSDATE is ${CVSDATE} + 1 which is reevaluated when doing a update_data. Jul 07 08:25:52 proti: := == immediate assignment Jul 07 08:25:57 zecke: So, using this gdbserver and a cross-gdb on my laptop, is the cross-gdb able to read debugging information from the files on the laptop? Jul 07 08:26:03 zecke: As source code line and so on? Jul 07 08:26:16 Sirfred: if you have the unstripped binaries Jul 07 08:26:33 zecke: Sounds very interesting. Jul 07 08:27:15 Sirfred: old version http://handhelds.org/cgi-bin/cvsweb.cgi/opie/development/cross-debug/ Jul 07 08:28:16 zecke: I know. That is why I ask : do you really need this immediate assignement ? CVSDATE := CVSDATE + 1 is no different than CVSDATE = CVSDATE + 1 in the case you mention. Jul 07 08:28:33 And CVSDATE = CVSDATE + 1 is safe. Jul 07 08:30:09 proti: I don't think we can do it in a secure way Jul 07 08:34:34 zecke: give me on example where the := is mandatory and is not safe. Jul 07 08:36:28 zecke: Anyway, it was the following I didn't get: "Can we gurantee that replacing the 'base' can work on all data backend" Jul 07 08:40:15 Humm, I wonder how the rootfs is generated, are the files from the ipk used?, the files in the staging dir? Jul 07 08:40:51 I'm seeing that the SIGSEGV is in QTransformedScreen::mapToDevice, I had this before while changing the virtual functions in QScreen. Jul 07 08:41:02 Sirfred: Files from ipk. Have a look at the log_do_image. Jul 07 08:41:08 So I suspect that it's a fault of the image I've built. Jul 07 08:43:57 Sirfred: However, if the Index is not updated, it will keep installing the previous ipk packages listed in this file, even if there is new ones available. Jul 07 08:44:00 proti: we had data_dict.py around Jul 07 08:44:25 proti: and there we were not able to gurantee we could exchange the base again Jul 07 08:44:40 proti: we've to ask ourselves if that can happen again Jul 07 08:44:48 proti: I think I'm going to cleanup everything and start again. Jul 07 08:45:13 I have more than 1.000 stamps in the stamp directory, this is starting to be difficult to follow. Jul 07 08:48:07 zecke: So define the semantic you want. I built the data smart (or cow) using the examples given in the function documentation to discover some were wrong. Jul 07 08:48:09 Sirfred: from ipk Jul 07 08:48:43 rwhitby-away: ping Jul 07 08:48:56 I've one monotone db from two servers Jul 07 08:49:10 Sirfred: Remove the index in deploy ipk or do a refresh by hand (the correct command is in the command that build the image). Jul 07 08:49:13 I've propagated some branch from one server to another branch from the other server Jul 07 08:49:20 now I want to sync the branch back... Jul 07 08:50:05 proti: I've erased the stamps for the suspecting outdated ipk, and run again bitbake opie-image. I expect that this will be enough. Jul 07 08:52:04 Sirfred: It is in the root_fs.bbclass. Jul 07 08:52:05 ipkg-make-index -r ${DEPLOY_DIR_IPK}/Packages -p ${DEPLOY_DIR_IPK}/Packages -l ${DEPLOY_DIR_IPK}/Packages.filelist Jul 07 08:52:09 -m ${DEPLOY_DIR_IPK} Jul 07 08:52:42 ipkg-make-index is found in the staging dir. Jul 07 08:53:09 This will refresh the Packages index. Jul 07 08:53:37 proti: It also get updated with 'bitbake opie-image' ? Jul 07 08:55:22 Sirfred: I used to have some problem with it. The update was not always done. Jul 07 08:55:31 proti: OK Jul 07 08:56:14 Sirfred: Check your log_do_image and have a look at the installation messages. It will tell you which packages is installed. Jul 07 09:11:54 Well, I've found the problem, it was my fault. Jul 07 09:12:27 In my tests, I changed a class virtual table and modified the .h in my staging directory to compile valid libraries for that change. Jul 07 09:13:14 It seems that header on the staging directory didn't get restored because the package that owns it didn't need to be recompiled. Jul 07 09:28:59 hm, anyone experience with mplayer in OE? Jul 07 09:47:47 * chouimat is away: lunch Jul 07 09:56:46 *phrw* Jul 07 09:56:58 exhausting? ;) Jul 07 09:57:02 yea Jul 07 09:57:17 why did i marry when i have to do cleaning? Jul 07 09:57:19 (just kidding) Jul 07 09:57:20 * mickeyl hides Jul 07 09:57:47 hahaha Jul 07 10:02:19 hah! Jul 07 10:02:19 * mickeyl dances a jig Jul 07 10:02:33 adding a sleep 5 fixes BT initializing on my husky Jul 07 10:02:36 shepherd, even Jul 07 10:12:50 http://dashjr.org/~luke-jr/tmp/wombat.jpg Jul 07 10:12:56 03mickeyl 07org.openembedded.dev * r1e69705d... 10/packages/bluez-utils/ (bluez-utils-common_2.17.inc files/base.patch): Jul 07 10:12:56 bluez-utils: Sleep a couple of seconds in /etc/pcmcia/bluetooth to give the Jul 07 10:12:56 hardware enough time to complete initialization (tested on Sharp C7x0) Jul 07 10:22:09 mickeyl: Have you cleaned behind the doors? Jul 07 10:22:26 mickeyl: under the bed? Jul 07 10:22:30 Mm Jul 07 10:23:35 Is it normal that in my shining just created opie-image filesystem the /etc/apm/scripts.d is empty? Jul 07 10:24:09 I remember to find there two little cute scripts, one for the ifaces and another one for usb, or something so. Jul 07 10:27:33 good afternoon all Jul 07 10:28:24 * chouimat is away: nap Jul 07 10:29:26 * koen|ewi just received a mail confirming the successfull end of a 4 month course Jul 07 10:32:08 hey koen|ewi Jul 07 10:32:11 congrats Jul 07 10:32:34 Sirfred: actually, I've cleaned the bathroom and my computer room Jul 07 10:32:57 mickeyl: That's not too much. Jul 07 10:33:17 Sirfred: yes, /etc/apm/scripts.d is empty. this directory is deprecated Jul 07 10:33:19 mickeyl: Your wife should be harder with you. Jul 07 10:33:30 mickeyl: So, where is that stuff now? Jul 07 10:33:30 we nowadays use /etc/apm/suspend.d and /etc/apm/resume.d (or so) Jul 07 10:33:37 hah, she's hard enough with me Jul 07 10:33:41 mickeyl: :) Jul 07 10:33:50 mickeyl: Sure you deserve it. Jul 07 10:34:18 mickeyl: As far as i remember, there were symbolic links in suspend.d and resume.d pointing to the actual files in scripts.d Jul 07 10:34:27 mickeyl: What was bad with that ? Jul 07 10:34:31 mickeyl: Just curious. Jul 07 10:34:55 mickeyl: did you subscribe to angstrom-dev yet? **** BEGIN LOGGING AT Thu Jul 07 10:38:39 2005 Jul 07 10:40:18 koen|ewi: not yet Jul 07 10:40:51 Sirfred: ah i mixed that up. the scripts are properly still supposed to be in there. it's event.d that has been deprecated in favour of the new three directory scheme Jul 07 10:41:17 mickeyl: So, I should have broken something, because my scripts.d is empty Jul 07 10:41:26 mickeyl: But I don't know how could I have done it. Jul 07 10:42:52 mickeyl: Also, I'm not able to make rotation using the screen hinge work. Jul 07 10:43:00 mickeyl: Perhaps these messages have something to do: Jul 07 10:43:12 Zaurus::rotation() Jul 07 10:43:13 Zaurus::readHingeSensor() - couldn't compute hinge status! Jul 07 10:43:13 Zaurus::rotation() - hinge sensor = 1 Jul 07 10:43:28 mickeyl: What do you think? Jul 07 10:50:21 mickeyl: Also, in event.d I have an script: 01opiealarm Jul 07 10:51:41 looks about normal Jul 07 10:51:51 2.6.12 broke the input subsystem rotation Jul 07 10:51:55 it's not back yet Jul 07 10:52:08 waiting for RP and the input guys to decide on an interface Jul 07 10:53:25 * mickeyl subscribes angstrom-dev Jul 07 10:53:27 re Jul 07 10:53:32 bbl, playing a bit Jul 07 10:54:39 Hmm. Jul 07 10:56:50 hi koen Jul 07 10:56:58 hey woglinde Jul 07 10:57:36 france_: good afternoon! Jul 07 10:58:34 koen: hi Jul 07 11:01:25 * chouimat is back. Jul 07 11:02:22 * koen wanders of for lunch Jul 07 11:43:53 mickey|sports, sirfred|tv: Not waiting for me. Just waiting for the input people. Its causing me no end of problems... Jul 07 12:00:01 ~hail pb_ Jul 07 12:00:02 * ibot bows down to pb_ and chants, "I'M NOT WORTHY!!" Jul 07 12:00:17 zecke: o coremaster Jul 07 12:00:31 http://handhelds.org/~pb/collections-test.tar is my testcase Jul 07 12:00:47 er, no it isn't Jul 07 12:00:54 http://handhelds.org/~pb/collection-test.tar is Jul 07 12:01:53 zecke: it contains two files, test_100.bb and test_200.bb. if you do "bitbake test", it should build version 100 due to the COLLECTIONS. Jul 07 12:04:04 pb_: just do BBFILE_PRIORITY? or by PV as well? Jul 07 12:04:40 zecke: I don't understand what you're asking Jul 07 12:05:03 sorry, I'm lagging to my screen... Jul 07 12:05:21 I've simply broken collection, never used it before Jul 07 12:05:40 is tes_100 only picked due the BBFILE_PRIORITY hint in bitbake.conf? Jul 07 12:05:44 yes Jul 07 12:06:14 this comment: Jul 07 12:06:15 doesn't bitbake complain because it can't parse conf/local.conf? Jul 07 12:06:18 # If there is a PREFERRED_VERSION, find the highest-priority bbfile providing that Jul 07 12:06:18 # version. If not, find the latest version provided by an bbfile in the Jul 07 12:06:18 # highest-priority set. Jul 07 12:06:23 describes what's meant to happen Jul 07 12:06:26 zecke: no. Jul 07 12:06:48 local.conf is normally included by bitbake.conf, but the bitbake.conf I gave you doesn't do that. Jul 07 12:06:59 ah right ;) Jul 07 12:07:06 and, in any case, bitbake doesn't (afaik) produce a diagnostic if you try to include a file that doesn't exist Jul 07 12:07:20 my IQ is anti proportional to the speed of my internet connection Jul 07 12:07:34 for local.conf it does ;) Jul 07 12:07:51 ah Jul 07 12:08:10 anyway that will be easy to fix (if any error comes along) Jul 07 12:09:31 great Jul 07 12:15:29 hm, is OE moved to monotone 0.20 yet? Jul 07 12:15:52 keturn: no Jul 07 12:16:00 keturn: mickeyl checks his scripts Jul 07 13:30:47 hail mickeyl Jul 07 13:36:42 evening Jul 07 13:37:01 * mickeyl just sent mail to oe regarding 0.20 Jul 07 13:37:10 no time for some weeks to upgrade. Jul 07 13:38:24 hello Jul 07 13:48:02 sirfred|tv: ping Jul 07 13:48:21 sirfred|tv: you sent me two patches, however the 2nd patch is included in the first one as well Jul 07 13:49:06 hm, can anyone help me to find out why my bootstrap-image doesn't work? Jul 07 13:49:30 it seems like init doesn't do anything.. Jul 07 13:50:39 mickeyl: yah, it's a shame that monotone is still changing in incompatible ways. Jul 07 13:51:13 I kind of wish those guys would at least maintain backwards compatibility for netsync. Jul 07 13:52:02 mickeyl: ¿? Jul 07 13:52:08 I don't have to give something like init=/linuxrc to the kernel, do I? Jul 07 13:52:12 sirfred|tv: just sent you a mail. Jul 07 13:52:19 mickeyl: OK, I'm gonna see Jul 07 13:52:46 pb_: the good sign is that in 0.20 we finally have that trigger on push which I've requested Jul 07 13:53:02 pb_: so we'll soon have a push model instead of a pull model for the CIAnots Jul 07 13:53:55 mickeyl: ah, excellent Jul 07 13:54:20 no sense having the CIAnots spinning their wheels all the time. Jul 07 13:54:32 mickeyl: You're right, I've messed it up going forward and back with patcher. Jul 07 13:54:47 mickeyl: The qapplication_qws.cpp part is duplicated. Jul 07 13:54:52 mickeyl: Sorry, I'm going to fix it. Jul 07 13:55:06 sirfred|tv: np. Jul 07 13:55:46 pb_: Interesting problem. findBestProvider sort according priorities and then don't care. Jul 07 13:56:09 proti: yeah, I think the current code in findBestProvider is just broken. Jul 07 13:56:35 pb_: I did it. Just copied from the existing place. Jul 07 13:56:37 as far as I can tell, the stuff that it does with tmp_pn cannot be right. Jul 07 13:57:21 I'm not sure when that was introduced. Jul 07 13:57:47 (I didn't update my bitbake for a long time, then suddenly I upgraded to the latest one and everything stopped working.) Jul 07 13:58:41 priority calculation seems ok. After it doesn't make sens. What version was the last one working for you ? Jul 07 13:59:43 I'm not sure. I can try to find out, I guess. Jul 07 14:00:07 probably from about 8-10 weeks ago. Jul 07 14:01:18 I was looking some oe thing. And the code is already there. Jul 07 14:04:35 pb_: OT: If a pcmcia card doesn't resume on resume until you manually call cardctl resume, would that classify as a bug in the driver? Jul 07 14:04:40 or could it be on purpose? Jul 07 14:06:44 pb_: I don't understand. In the code there is a comment # get highest priority file set. Jul 07 14:07:04 And nothing after this use the priority of the package. Jul 07 14:07:32 Just the preferred_v or preferred_r and the vercmp. Jul 07 14:09:04 pb_: r235 and r263 seem to be significat revisions. in 236 and 264 findBestProvider changes were introduced. Perhaps the collections broke in one of those two Jul 07 14:09:49 mickeyl: I think it would be a bug in the driver. Jul 07 14:09:59 mickeyl: yeah, r235 would be my prime suspect Jul 07 14:10:29 yep. 236 also broke the PN-PV Jul 07 14:10:35 which was repaired in 264 Jul 07 14:10:52 proti: that's correct. after that point, it is meant to be operating only on the highest-priority set. but, because the code does that funky appending thing with tmp_pn, it is actually operating on the union of all sets. Jul 07 14:11:02 which, clearly, is no good Jul 07 14:11:03 pb_: ok, thanks. it's my serial CF bluetooth card malfunctioning that way. I'll take a look. Jul 07 14:11:19 * mickeyl needs to wander to bed now Jul 07 14:11:22 g'night Jul 07 14:11:27 mickeyl: sweet dreams Jul 07 14:11:33 thanks :) Jul 07 14:11:40 'night mickey|zzZZzz Jul 07 14:11:57 mickey|zzZZzz: just one thing. my monotone failed the link. undef __malloc. Jul 07 14:12:14 proti: no problem, disable the three DEFINES Jul 07 14:12:18 it'll build then Jul 07 14:12:48 ok thx. night Jul 07 14:12:57 main.cc: 457ff Jul 07 14:13:00 cu Jul 07 14:13:56 pb_: That code was already there in the oe time. Jul 07 14:14:15 pkg_pn[pn] = [] Jul 07 14:14:15 for p in p_list: Jul 07 14:14:15 pkg_pn[pn] = [ priorities[p] ] + pkg_pn[pn] Jul 07 14:14:21 from oemake. Jul 07 14:14:42 Ah. Jul 07 14:14:48 pb_: I see the problem. Jul 07 14:15:34 I did remove the bracket for a very good reason. Jul 07 14:26:10 pb_: ping Jul 07 14:28:57 ibot: botmail for: pb check the patch http://www.frankengul.org/~seb/fix_collection.patch Jul 07 14:44:20 ibot: ping Jul 07 14:44:21 pong Jul 07 14:44:33 ibot: botmail for pb: check the patch http://www.frankengul.org/~seb/fix_collection.patch Jul 07 14:46:03 proti: have you confirmed that collections work properly with that patch? Jul 07 15:09:38 hi jacques Jul 07 15:10:02 hi pb_ :-) had a little power outage here Jul 07 15:11:02 doh Jul 07 15:12:25 yeah power company said it would be three hours so I turned everything off Jul 07 15:12:33 but luckily it wasn'ty Jul 07 22:33:18 03mickeyl 07org.openembedded.dev * r6eb433c9... 10/packages/qte/ (3 files in 2 dirs): Jul 07 22:33:18 Qt-Embedded 2.3.10 patches courtesy Manuel Teira: Jul 07 22:33:18 - add suspend-resume hook for qws server Jul 07 22:33:18 - implement suspend-resume helpers in W100 driver Jul 07 22:55:29 good morning all! Jul 07 23:37:56 what constitutes a distribution ; just output of oe (filesystem and kernel) or something extra ? Jul 07 23:38:51 suppose i want to give out a linux distribution for my target apart from the filesystem what else is expected ? Jul 07 23:40:53 Good morning **** ENDING LOGGING AT Thu Jul 07 23:59:56 2005