**** BEGIN LOGGING AT Wed Aug 28 02:59:58 2013 Aug 28 03:14:27 hi challinan Aug 28 03:21:34 is there a yocto developer day this year at LinuxCon/ELCE? Aug 28 06:58:11 zibri: yes, installing libc6-dev-i386 fixes the problem, without any other changes Aug 28 07:55:44 morning all Aug 28 07:59:38 bluelightning: good morning Aug 28 08:10:13 tf: https://bugzilla.yoctoproject.org/show_bug.cgi?id=5054 Aug 28 08:10:14 Bug 5054: normal, Medium, 1.5 M5, dvhart, NEW , syslinux-native depends on i386 libc headers on amd64 Aug 28 08:32:29 morning all Aug 28 08:38:56 morning RP, bluelightning, all Aug 28 08:39:54 morning Aug 28 08:42:07 mihai: JaMa: I'm going to send a set of patches to remove the recipes added to OE-Core from meta-oe / meta-gnome FYI Aug 28 08:43:06 mihai: oops, wrong mihai, sorry ;) Aug 28 08:43:07 bluelightning: ok, I'll let Mihai Prica know ;-) Aug 28 08:43:15 * bluelightning has red face Aug 28 08:43:33 bluelightning: np Aug 28 08:46:53 RP: do you think this issue can happen? What about hardcoding usr? Aug 28 08:46:54 http://cgit.openembedded.org/meta-openembedded/tree/meta-initramfs/recipes-devtools/klibc/klibc-2.0.2/use-env-for-perl.patch Aug 28 08:47:17 RP: I've never hit that and I think same for JaMa's autobuilder Aug 28 08:51:56 bluelightning: thanks Aug 28 08:55:07 hi, I am writing a makefile-based recipe, it is very simple without do_compile or extra_oemake defined, but I noticed that the CFLAGS defined in the Makefile itself are not honored, instead only the CFLAGS environment variable setup by bitbake is used. Do you know what causes this and how to solve it? Thanks Aug 28 08:56:56 joseppc: -e in EXTRA_OEMAKE Aug 28 08:57:42 you should be able to set EXTRA_OEMAKE to "" in your recipe to get rid of that behavior, if that's what you want Aug 28 09:00:33 zibri: thanks. I believe I still need the cflags defined in the environment by bitbake, but I definitely want the ones in the Makefile too... Aug 28 09:03:07 EXTRA_OEMAKE = "" did the trick Aug 28 09:09:16 ant_work: its a problem. Don't we have the perlnative bbclass to assist with this? Aug 28 09:09:37 ant_work: You might end up having to use a wrapper :/ Aug 28 09:12:46 RP: seems that perl-native recipe creates the wrapper Aug 28 09:21:05 RP: so I'm already safe being DEPENDS = "linux-libc-headers perl-native" Aug 28 09:50:07 ant_work: I think if I remember rightly there is a perl-native binary you can convert it to use Aug 28 09:50:14 then we will get the right thing out of PATH Aug 28 09:54:04 RP: for a perl expert, the script is trivial and the patch just seems to touch a print. As for perl-native the recipe explicitely says it fixes such cases Aug 28 09:54:15 http://git.kernel.org/cgit/libs/klibc/klibc.git/tree/klcc/makeklcc.pl Aug 28 09:56:08 hi, how do i find out which package builds a specific programm? something like a search function for "package provides" Aug 28 09:58:55 RP: but I don't know what will end up with a collapsed /usr in / Aug 28 09:59:18 RP: guess what? I ask pb_ Aug 28 10:51:01 hmm, hi. Aug 28 10:51:08 it might be off-topic, but let me give it a try: | configure: WARNING: If you wanted to set the --build type, don't use --host. Aug 28 10:51:11 | If a cross compiler is detected then cross compile mode will be used. Aug 28 10:51:11 what does it mean? Aug 28 10:51:27 it was complaining about the host mode before. Aug 28 10:51:32 that is why I passed --host in there. Aug 28 10:53:45 I was using this from a Makefile, ./configure --host=`gcc -dumpmachine`; make Aug 28 10:57:08 bluelightning: should I use unset for CC too? Aug 28 10:57:10 like CPPFLAGS? Aug 28 10:57:15 or CC will get overriden? Aug 28 11:06:37 ./configure --build=`gcc -dumpmachine` --host='${CC} -dumpmachine`; $(MAKE) Aug 28 11:06:41 * lpapp is using that now Aug 28 11:17:26 where is Yocto getting the HOST_SYS variable set? Aug 28 11:20:58 ../meta/classes/staging.bbclass:32: if [ "${BUILD_SYS}" = "${HOST_SYS}" ]; then Aug 28 11:21:07 hmm, is that bug? Note, '=' instead of '==' Aug 28 11:22:16 hmm, apparently no, a right bash syntax... Aug 28 11:25:13 lpapp: Actually, == is bash syntax, = is posix Aug 28 11:25:43 yes, == is bash syntax Aug 28 11:25:49 although = works, too Aug 28 11:25:58 not sure what you mean by posix, though. Aug 28 11:27:14 lpapp: posix is the official standards behind most of unix. If you want stuff to be compatible, it should follow posix. Aug 28 11:27:22 lpapp: http://en.wikipedia.org/wiki/POSIX Aug 28 11:27:38 I still have no idea what you mean by posix in here. Aug 28 11:28:38 Portable Operating System Interface Aug 28 11:28:49 lpapp: What I mean is that using = is posix compatible and will work even if you use dash instead of bash. Using == will not. Aug 28 11:29:15 only if the shell in question wanna be compliant. Aug 28 11:29:27 i.e. cmd.exe will not work. Aug 28 11:30:15 I was writing a shell interpreter, I would not accept it either because it is silly... and would probably the root cause of subtle bugs. Aug 28 11:30:18 if* Aug 28 11:32:05 lpapp: Well, all I know is that after debian switched to using dash as its default shell rather than bash, we had to fix a lot of scripts that used bashisms (bash specific code). Aug 28 11:33:20 I consider that breaking. ;) Aug 28 11:33:36 I mean, I do not know any other programming language or even scripting, having this syntax. Aug 28 11:43:00 lpapp: vhdl uses "=" for comparing too, see http://www.csee.umbc.edu/portal/help/VHDL/operator.html Aug 28 11:43:06 fyi ;) Aug 28 11:43:22 thesignal: I said programming and scripting languages. Aug 28 11:43:26 erlang as well :) Aug 28 11:43:33 VHDL as far as I know is a hardware description language. Aug 28 11:44:33 http://en.wikipedia.org/wiki/VHDL : "VHDL can also be used as a general purpose parallel programming language" Aug 28 11:45:13 but yes you're right, 99% it is used for hardware description ;) Aug 28 11:45:53 thesignal: perhaps someone it can be hacked around, but the very beginning of the wikipedia link is: Aug 28 11:45:57 "VHDL (VHSIC Hardware Description Language) is a hardware description language..." Aug 28 11:46:20 can you show a programming language use case in /query (as it is getting off-topic)? Aug 28 11:46:30 anyway, it is not a bug; good to know. Aug 28 11:50:29 VHDL is also used for programming FPGAs. I know someone that is doing so. Aug 28 11:51:58 I wonder if you mean by programming hardware description, i.e. gates, etc. Aug 28 11:52:20 lpapp: programming gates yes, to build functionality Aug 28 11:52:41 they call it hardware description Aug 28 11:54:02 bluelightning, do you have any thoughts on resolving the dropbear/openssh issue while builds sdk's? Aug 28 11:54:06 I duped the problem Aug 28 11:54:35 Net147: and many also call the flashing programming. :) Aug 28 12:03:04 JaMa: do you think it would be acceptable to put openflow into meta-networking? Aug 28 12:03:19 the one in meta-virtualism is outdated, and it is a bit weird why it is there. Aug 28 12:15:56 anything special I need to do to switch to 3.10 kernel? I set the preferrred version and I get NOTE: preferred version 3.10% of linux-yocto not available (for item virtual/kernel), NOTE: versions of linux-yocto available: 3.8.13+gitAUTOINC+e7f42a416a_f20047520a Aug 28 12:16:26 Crofton|work: somehow the dependency chain has to be broken, there's no other way to fix it Aug 28 12:25:37 Crofton|work, did you happen to get a chance to look into the issue I was having yesterday? Aug 28 12:25:50 I made it happen again Aug 28 12:26:03 I need to look at the depndency chains Aug 28 12:26:07 I'll try and get to it Aug 28 12:26:16 but I am going on holiday tmoorrow Aug 28 12:26:32 Crofton|work, thanks. I appreciate your efforts. Have a good holiday. Aug 28 12:26:45 it is an annoying thing that will bite others Aug 28 12:27:39 RP: there is ever a bashism... $(shell bash -c 'type -p $(PERL)') \ Aug 28 12:27:40 Crofton|work, most everything I've run into has been an "annoying thing that will bite others". My plan is to eventually document my experiences. Aug 28 12:28:10 hopefully it will help some other newb like me. Aug 28 12:28:13 ant_work: I haven't had time to look into that in detail, sorry :( Aug 28 12:28:14 heh Aug 28 12:28:18 I want to get a fix in Aug 28 12:28:24 also looking for hack though Aug 28 12:28:41 cfo215: such docs would be much appreciated Aug 28 12:28:42 Either works for me at this point. I really need to get that SDK going. Aug 28 12:28:57 cfo215: we do want to try and remove such obstacles... Aug 28 12:29:12 RP: thx. I'll grep for recipes using perl-native.and report any interesting findings Aug 28 12:29:50 ant_work: I think the best way to fix this would be to ensure there is a symlink in the native sysroot bin directory for "nativeperl" or "perlnative" to the perl in the perl-native subdirectory Aug 28 12:30:09 ant_work: you'd then change the start of the file to call "nativeperl" or "perlnative" Aug 28 12:30:19 I could have sworn we had this symlink but I can't see it Aug 28 12:30:29 equally I'm trying to do 10 things at once Aug 28 12:30:37 does anyone know anything about the author of the meta-networking module? Is he on vacation, sick, etc? Aug 28 12:30:51 RP: imagine I'm afk here in office ;) Aug 28 12:31:47 nevermind, forgot to rename bbappend for kernel Aug 28 12:32:30 RP: I'm willing to help however I can. I'm not new to open source... just to ARM/OE/Yocto/Angstrom/etc. Aug 28 12:33:13 When I ran bitbake in a new build directory using an existing sstate cache, I got some 20 occurrences simliar to "WARNING: Setscene task 245 (/home/pkj/yocto/poky-dylan/meta/recipes-connectivity/connman/connman_1.12.bb, do_package_setscene) failed with exit code '1' - real task will be run instead". Aug 28 12:33:18 Is this normal? Aug 28 12:33:39 In the log I found "pseudo: pseudo: can't run useradd: Permission denied" which seems to be the cause... Aug 28 12:35:14 cfo215, TASK_BASIC_SSHDAEMON = "openssh openssh-sftp openssh-sftp-server" Aug 28 12:35:27 add that to conf/local.conf and rebuld Aug 28 12:35:50 Crofton|work, will do. It shouldn't take long. Aug 28 12:36:03 I am trying also, but I think I screwed it up :) Aug 28 12:36:18 :) Aug 28 12:36:32 bluelightning, do you know much about qt-embedded and touchscreens? Aug 28 12:36:45 Crofton|work: a little, what's the question? Aug 28 12:37:02 cfo215, is trying to use it on a bone+some lcd and has an issue Aug 28 12:37:43 cfo215: if you could expand on your issue that would be nice :) Aug 28 12:39:19 he's probably trying to fix hios sdk issue :) Aug 28 12:39:57 Crofton|work, bluelightning: in a nutshell, there is no (afaicd) touchscreen support in the console-image. I need that for my Qt application to work properly. It may be a recipe I can append, like I'm doing to get my Qt support and Phidgit library included in the build. Aug 28 12:40:29 so you are adding qt-embeeded via IMAGE_EXTRA_INSTALL? Aug 28 12:40:39 yes Aug 28 12:40:59 cfo215: this is qt4 right? Aug 28 12:41:34 bluelightning, Crofton|work, my conf has IMAGE_INSTALL_append += "qt4-embedded" in it. Aug 28 12:42:06 cfo215: ok, FYI that will pull in a load of things that you don't need, but you can trim that down later of course Aug 28 12:42:24 cfo215: by default, qt-e uses tslib to enable touchscreen support, and that provides a number of backends for different touchscreen devices Aug 28 12:42:48 bluelightning, yeah, I seen it add all the demos, etc. That's an easy fix though. Aug 28 12:43:35 cfo215: one thing to do might be to install the tslib-tests and tslib-calibrate packages and then run those outside of the qt-e application to see whether they work Aug 28 12:44:45 bluelightning, that's good. As long as I can calibrate it somehow I'll be happy. Do you know if ts_calibrate is included in the qt-e package? Aug 28 12:44:58 cfo215: it's not, no Aug 28 12:45:19 bluelightning, np. I can probably add it as a recipe to the build. Aug 28 12:45:30 cfo215: ts_calibrate is super bare bones though, I suspect you'd want a proper qt-e application for that purpose ultimately Aug 28 12:46:02 cfo215, I ahve an sdk Aug 28 12:46:43 bluelightning, do you know if one exists already for Qt? I know how to rotate the screen via the -qws params. Aug 28 12:47:49 Crofton|work, I'm on 0: console-image-1.0-r0 do_populate_sdk (pid 332) :-) Aug 28 12:48:25 cfo215: for the latter I'm not sure if there is one provided out of the box with qt, but I did just come across this: http://qt-project.org/doc/qt-4.8/qws-mousecalibration.html Aug 28 12:48:29 cfo215: not sure how relevant it is Aug 28 12:48:47 bluelightning, I'll have a look. thanks Aug 28 12:49:15 I'm sure if I bought a license for Qt they'll be able to tell me how to do it... Aug 28 12:50:09 cfo215: this may be of interest as well: http://qt-project.org/doc/qt-4.8/qt-embedded-pointer.html Aug 28 12:50:25 bluelightning, thanks again! Aug 28 12:51:03 cfo215: np Aug 28 12:52:47 cfo215, if you get this to owrk, go to ELCE and buy us some beer Aug 28 12:53:16 I'll dig into the depends hell later Aug 28 12:53:29 hopefully openssh over dropbear works for you for now Aug 28 12:57:17 Crofton|work, that sounds like a plan. When/where is ELCE? Aug 28 12:57:29 Edinburgh Aug 28 12:57:35 late October Aug 28 12:57:49 * Crofton|work needs to sort out planes Aug 28 12:58:25 Oh, that really sounds like a plan... I'll see if the missus would like to travel to Scottland! Aug 28 13:00:37 Crofton|work, bluelightning, thanks for all your help. I really do appreciate it! Aug 28 13:03:18 http://events.linuxfoundation.org/events/embedded-linux-conference-europe Aug 28 13:17:48 has anybody experience building images with xbmc? i get a lot of errors. fixing them becomes whack a mole. now i have: "Java not found, it will be used if found after configure. This is not necessarily an error.", but this is followed by a fatal make error: make: *** [java-not-found] Error 1 Aug 28 13:25:07 zibri: I think it's fair to say the xbmc recipe in meta-multimedia has been very problematic Aug 28 13:26:08 i'm not surprised. i've had similar issues last time i tried. Aug 28 13:26:54 I think patches would certainly be welcome, but AFAICT it's been a few different people spending ages trying to get the thing working since it's existed and only getting about 90% of the way each time Aug 28 13:27:27 yeah, i had a look at the xbmc build system last time i tried, but didn't make much progress. Aug 28 13:31:11 i've been able to solve some minor issues, but i'm not even close to be able to build it successfully. :( Aug 28 13:38:15 Crofton|work, I have a toolchain. Installing it now... thanks again. Aug 28 14:00:55 hmm, what can I do when there is no error message, but I get a do_compile error issue? Aug 28 14:01:35 lpapp: you check the do_compile log Aug 28 14:01:55 ERROR: Task 6 (/home/lpapp/Projects/Yocto/poky-dylan-9.0.1/meta-foo/recipes-foo/bar/bar_0.1.bb, do_compile) failed with exit code '1' Aug 28 14:02:05 mihai: as I said, it has nothing inside that writes error Aug 28 14:02:20 except that very line Aug 28 14:03:09 it should point you to a log.do_compile* file when bitbake ends Aug 28 14:03:43 yeah, so? Aug 28 14:04:05 only error result is: warning: variable 'error' set but not used [-Wunused-but-set-variable] Aug 28 14:04:09 but that is actually only a warning. Aug 28 14:04:59 * lpapp will try MAKE = "make VERBOSE=1" in do_compile Aug 28 14:05:28 will that be overriden by oe_runmake ? Aug 28 14:07:13 ok, VERBOSE=1 helped Aug 28 14:07:17 it showed a missing dependency... Aug 28 14:07:28 I wonder how to improve this? I should have gotten that right away... Aug 28 14:15:55 bluelightning: oh, I need to unset the LDFLAGS, too? Aug 28 14:16:57 i.e.: do_compile() { unset CPPFLAGS LDFLAGS oe_runmake -C Aug 28 14:17:01 foo } Aug 28 14:17:16 not sure, depends; but I see we do have some recipes that do that Aug 28 14:17:31 lots of undefined reference error Aug 28 14:17:38 even though the makefile is correct Aug 28 14:17:43 and the -lpthread, etc are not passed. Aug 28 14:17:48 yeah, that was it. Aug 28 14:18:28 lpapp: it looks like you are abusing / perusing the buildsystem Aug 28 14:18:31 ;) Aug 28 14:18:48 ? Aug 28 14:19:12 lpapp: the heap of issues you're mentioning is not normal Aug 28 14:20:12 lopapp: if the sources are written well you should not generally touch the flags Aug 28 14:20:30 lpapp: sorry for mistyping Aug 28 14:21:26 ant_work: ? Aug 28 14:21:36 how would you link against the symbol without a library providing that? Aug 28 14:22:00 lpapp: try starting with a single .c file and play with the flags Aug 28 14:22:19 lpapp: then pass to Makefile then to autofoo Aug 28 14:22:43 ant_work: why would I start playing? Aug 28 14:22:55 I have a working system. Aug 28 14:23:08 you think it is abusing to pass -lpthread when you wanna have a symbol from libpthread.so.X Aug 28 14:23:33 which is incorrect. Aug 28 14:23:41 lpapp: the author knows the dependencies Aug 28 14:23:50 I *am* the author. Aug 28 14:24:14 and I set up stuff properly. Aug 28 14:24:22 it is only that yocto overrides my wish by default. Aug 28 14:24:25 that is why you use unset first. Aug 28 14:24:41 lpapp: I know very well what happens resetting the flags and pointing to some other include dir Aug 28 14:24:58 but I have a very specific reason to do that (klcc) Aug 28 14:26:45 ant_work: I will find you the bugreport, sec, hold on... Aug 28 14:27:09 https://bugzilla.yoctoproject.org/show_bug.cgi?id=5046 Aug 28 14:27:10 Bug 5046: normal, Undecided, ---, scott.m.rifenbark, NEEDINFO , Document that CPPFLAGS override behavior and possible workaround Aug 28 14:29:03 rburton: you use Debian right? Aug 28 14:29:56 or does anyone know; i have someone trying to build my oe-derived system on Debian, and they're getting: `mkdir: cannot create directory '/usr/include/gnu': Permission denied` in eglibc-initial_2.17.bb Aug 28 14:30:10 that directory appears to exist on RHEL/Fedora hosts by default, but not on Debian Aug 28 14:30:39 walters: he does, yes but is away atm Aug 28 14:30:48 walters: i don't have that issue, but i have that directory. Aug 28 14:31:06 walters: sounds like a bust makefile somewhere :/ Aug 28 14:31:15 yocto shouldn't install stuff in hosts /usr/include/gnu Aug 28 14:31:47 It won't, it will just be a bust makefile where the existence check looks without the DESTDIR prefix :/ Aug 28 14:46:24 hmm, TypeError in gtk+ do_package, odd Aug 28 14:46:28 * kergoth digsd Aug 28 14:46:31 s/d$// Aug 28 14:52:38 what is the relation of oe-lite to openembedded? Aug 28 14:55:20 https://gist.github.com/kergoth/6366936 this is odd, particularly the TypeError with no context/message/arguments Aug 28 15:02:18 kergoth: Looks like it happened somewhere in setVar :/ Aug 28 15:17:04 aha. ' '.join(None) -> TypeError with no message. do_split_packages must be returning None for me for some reason Aug 28 15:17:12 :) Aug 28 15:17:15 * kergoth kicks python Aug 28 15:19:23 having issues with built SDKs: executables seem to be linking to the wrong libraries: http://pastebin.com/AuVAWmtD for my attempts at debugging this. Aug 28 15:23:24 added some links... new pastebin http://pastebin.com/em5yRhN4 Aug 28 16:04:16 odd. both the x11 and wayland backends are enabled, yet no immodules were installed Aug 28 16:07:35 ah, right, i see Aug 28 16:15:30 i hate when recipes include a .inc with the non-prefixed path, so you can't override the .inc in a high priority layer (though usually a temporary measure, can still be handy as a workaround) Aug 28 16:19:03 JaMa: QT5 layer for openembedded -> you might want to fix that to Qt5 or Qt 5 Aug 28 16:19:10 QT is Quick Time. ;-) Aug 28 16:19:15 this is the description on github. Aug 28 16:19:53 JaMa: Also, I have a patch for 5.1.1 Aug 28 16:20:02 at least, I think I have. :) Aug 28 17:14:03 what is this issue about when running "bitbake core-image-minimal"? http://paste.kde.org/~lpapp/pb3801594/ Aug 28 17:15:24 lpapp: looks like incorrectly set up /dev/shm Aug 28 17:15:26 lpapp: didn't you just comment on bug 4925? Aug 28 17:15:27 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=4925 enhancement, Low, ---, scott.m.rifenbark, RESOLVED WONTFIX, Hard to set up shard memory IPC in chroot for supported distributions Aug 28 17:16:00 bluelightning: I did, but how is that relevant? Aug 28 17:16:42 surely this is the situation that prompted you to file that bug in the first place.. Aug 28 17:18:37 no Aug 28 17:18:40 it was before that. Aug 28 17:24:32 well, it's the same problem AFAICT Aug 28 17:25:40 no, it is not. Aug 28 17:25:50 one is the documentation issue Aug 28 17:25:55 this is now a runtime issue at hand. Aug 28 17:26:04 or you mean it is all shm topic? Yes, different subparts. Aug 28 17:27:41 the runtime issue is due to you not setting up /dev/shm in your chroot properly Aug 28 17:28:15 bluelightning: http://paste.kde.org/~lpapp/p51aa7906/ -> what is this Aug 28 17:28:34 bluelightning: you are welcome to tell me how that is set properly ... (still) ;-) Aug 28 17:28:51 we've stated many times we do not support chroot setups Aug 28 17:29:04 if you want to use one, fine, and it can be made to work Aug 28 17:29:08 but it's up to you to do that Aug 28 17:29:09 how? Aug 28 17:29:11 lol Aug 28 17:29:15 anyway, what is the issue above? Aug 28 17:30:18 not sure, there's no mention of "root-ssh-authorized" anywhere in poky Aug 28 17:30:26 this also really isn't at all yocto specific, not setting up a chroot properly will break all sorts of software Aug 28 17:30:32 bluelightning: denzil Aug 28 17:30:47 it has not broken any softwares the last 10-15 years Aug 28 17:30:55 lpapp: or in denzil Aug 28 17:31:04 bluelightning: well, I apparently have... Aug 28 17:31:21 lpapp: then apparently this is another of the changes yourself or your colleagues have made to poky Aug 28 17:31:36 it is upstream denzil. Aug 28 17:31:46 I just grepped upstream denzil Aug 28 17:31:47 maybe with some layer. Aug 28 17:31:55 anyway, I am going to clone poky now. Aug 28 17:32:25 also, why would core-image-minimal include any custom stuff? Aug 28 17:33:57 assuming it's unmodified, you can still set CORE_IMAGE_EXTRA_INSTALL in your config to add additional packages to it Aug 28 17:34:15 wow, really? Aug 28 17:34:28 I am not sure that is a good idea. Aug 28 17:34:37 because that is not a core minimal image anymore then Aug 28 17:34:40 but some custom image. Aug 28 17:34:56 it's about enabling users to get started adding things easily Aug 28 17:35:24 well, what is wrong about extending an image in an own image? Aug 28 17:35:31 it's more effort? Aug 28 17:35:34 like including that + stuff Aug 28 17:35:37 more effort how? Aug 28 17:36:12 IMO it is misleading to call an image core-image-minimal if it can be full of customized stuff Aug 28 17:36:18 it just is Aug 28 17:36:56 the user could go and edit the file as well, that would change its contents without changing its name... so how would not allowing this solve that? Aug 28 17:40:27 lpapp: looking at the error more closely, it's task-core-boot that contains the dependency Aug 28 17:40:56 lpapp: meaning it could only be a setting of VIRTUAL-RUNTIME_* or MACHINE_ESSENTIAL_EXTRA_RDEPENDS or MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS that is causing this Aug 28 17:41:11 bluelightning: as I said, it would solve the issue Aug 28 17:41:17 no, it wouldn't Aug 28 17:41:17 because they would need to create an additional image. Aug 28 17:41:24 sure, it would. Aug 28 17:41:30 you could not inject from your layer anymore. Aug 28 17:41:32 no, they could just edit the existing image -> same "problem" Aug 28 17:41:35 no Aug 28 17:41:44 modifying a meta image is a m00t point Aug 28 17:41:49 meta layer* Aug 28 17:41:55 because that is something you have to trust Aug 28 17:41:58 checksum etc Aug 28 17:42:25 it is like saying you do not trust the linux kernel because anyone can inject a module. Aug 28 17:42:41 without security frameworks, you cannot do anything really secure right up from the kernel. Aug 28 17:42:56 however external layers injecting stuff... that can be a security concern easily. Aug 28 17:43:34 it is dangerous and confusing IMHO. Aug 28 17:43:51 if you wanna have your image on top of core-image-minimal, fine, include it and extend. Aug 28 17:44:00 that is what you do with distros as well after all, or layers. Aug 28 17:45:00 this is my opinion. :) Aug 28 17:47:33 btw, getting this for wheezy which should be a supported distribution? WARNING: Host distribution "Debian-GNU-Linux-7" has not been validated with this version of the build system; you may possibly experience unexpected failures. It is recommended that you use a tested distribution. Aug 28 17:53:33 if cfo pops in again, can someone tell him my sdk runs on Fedora 19 Aug 28 17:53:54 in the sense I did not get the error he got running gcc Aug 28 17:54:57 Crofton|work: sure Aug 28 18:10:03 bluelightning: http://paste.kde.org/~lpapp/pd8b697a2/ -> getting three errors after 1400+ successful jobs. Aug 28 18:11:07 python --version Aug 28 18:11:07 Python 2.7.3 Aug 28 18:18:13 lpapp: it looks like you have run out of space on your disk Aug 28 18:33:12 bluelightning: got a clue for that one? Aug 28 18:34:13 lpapp: the clue is what sgw_ said i.e. "No space left on device" Aug 28 18:34:46 I have not seen that as I had to leave. Aug 28 18:35:46 sgw_: I see, thanks. Aug 28 19:29:17 Hi all, what size on HD can Yocto build take? Aug 28 19:29:54 I am thinking about taking some VPS plan Aug 28 19:30:10 to have a Yocto build dedicated machine Aug 28 19:30:21 An I am looking at the pricing here: https://www.digitalocean.com/pricing Aug 28 19:30:55 What I am wondering is would 20GB SSD Disk be enough Aug 28 19:31:07 unlikely, unless you're not building very much Aug 28 19:31:57 kergoth, so, it is enough? Aug 28 19:33:14 ah... You said actually that it is not enough? Aug 28 19:33:15 drasko: 100 GB is the recommeded Aug 28 19:33:25 I see... Aug 28 19:33:42 Anyone aware of some web based build service that I can use? Aug 28 19:33:46 "The build process using Sato currently consumes about 50GB of disk space. To allow for variations in the build process and for future package expansion, we recommend having at least 100GB of free disk space." Aug 28 19:33:54 drasko: OBS Aug 28 19:34:02 (I do not have powerful PC nor lot of free space) Aug 28 19:34:21 drasko: use rm_work Aug 28 19:34:33 sadly, there is no rm_download, but there is a bugreport for that Aug 28 19:34:54 it depends on what you're trying to build Aug 28 19:35:13 which image, etc. there's a big difference between the space requirements of a minimal console image and a substantial gui image, for example Aug 28 19:35:37 lpapp, what is OBS? Aug 28 19:35:48 drasko: it is not related to Yocto, so forget about it. Aug 28 19:36:00 OK Aug 28 20:39:41 lpapp: if your still around, did you change any of the BB_DISKMON_DIRS numbers or comment it out? Aug 28 20:43:46 sgw_: nope Aug 28 20:43:51 cfo215: hey Aug 28 20:44:08 lpapp, howdy Aug 28 20:44:48 cfo215: https://www.yoctoproject.org/irc/latest.log.html#t2013-08-28T17:53:33 Aug 28 20:45:50 lpapp, OK? Aug 28 20:46:21 cfo215: you are welcome. Aug 28 20:46:30 lpapp: Ok, seems the disk monitoring did not do it's job then, how much disk space was available at the time? Aug 28 20:46:42 sgw_: not sure Aug 28 20:48:00 lpapp, still not following... a little brain dead today. Aug 28 20:48:13 lpapp: Ok, well just wanted to try and figure out what really happened. Aug 28 20:49:23 sgw_: you are right that the monitor should have stopped stuff. Aug 28 20:49:55 cfo215: well, I was just trying to hand a message over.. Aug 28 20:50:14 lpapp, ic. Aug 28 20:53:47 lpapp, I got my console-image to build finally. I'm still having issues with the Angstrom SDK in the Angstrom-2013.06-yocto1.4 git branch. koen is aware of it. Has to do with LD_LIBRARY_PATH http://pastebin.com/L9Kspun1 Aug 28 20:55:22 lpapp, I'm really trying my best to break everything! LOL Aug 28 21:01:04 cfo215, so gcc from the sdk ran for me on Fedora 19 Aug 28 21:01:10 hello Aug 28 21:01:15 what OS are you on? Aug 28 21:02:00 well, what's happened to stat recipe? ERROR: Function failed: package_get_auto_pr Aug 28 21:02:26 Crofton|work, Who? I'm using Ubuntu 12.04.2 LTS for my development. Aug 28 21:03:02 hmmm, anyone here running an sdk built on Ubuntu on the same machine? Aug 28 21:04:33 cfo215, the headache is I can't dupe your issue Aug 28 21:04:34 now Aug 28 21:04:56 btw can you run *-gcc --version and make sure the problem is there Aug 28 21:05:25 Crofton, http://pastebin.com/L9Kspun1 has what I've done to troubleshoot. Aug 28 21:05:37 Crofton|work, http://pastebin.com/L9Kspun1 has what I've done to troubleshoot. Aug 28 21:05:39 ant_home: is that the exception / timeout issue? there's a yocto bugzilla bug for that, and a patch hit the list for it Aug 28 21:05:44 I am trying to simplify Aug 28 21:06:23 line 4 runs? Aug 28 21:06:46 kergoth, thx, disappeared on second run Aug 28 21:06:56 yes Aug 28 21:07:04 and did you try the "recommended" path Aug 28 21:07:13 ant_home: yeah, its a local network connectivity timeout talkin to the pr server, apparently we can exceed the timeout on loaded systems Aug 28 21:07:18 i've hit it a few times today too Aug 28 21:07:42 yep ERROR: Can NOT get PRAUTO, exception timed out Aug 28 21:08:02 big task is building linux-yocto Aug 28 21:08:37 Crofton|work, I know why it works... :( '$ which make' results in /usr/bin/make Aug 28 21:09:01 arm-angstrom-linux-gnueabi-gcc -v Aug 28 21:09:05 does that work? Aug 28 21:10:46 no. only with the hack in the pastebin. I get "error while loading shared libraries: __vdso_time: invalid mode for dlopen(): Invalid argument" otherwise. Aug 28 21:11:32 ok Aug 28 21:12:05 now, we need to find someone else with Ubuntu to try the same thing from an sdk from straight oe-core Aug 28 21:12:12 I suspect the problem is there also Aug 28 21:12:36 I jsut have no Ubuntu stuff around to try with Aug 28 21:12:40 hello guys where can i read something about sanity.conf ? Aug 28 21:13:03 i imagine so... the meta-qt-sdk or whatever it's called does the same. Aug 28 21:13:03 i get a complaint about my version of make ( i have 3.81 , 3.80 is needed ) Aug 28 21:16:26 cfo215, the quick way to get some attention would be try an qemu of oe-core only on your machine and see if that toolchain works Aug 28 21:16:40 if it fails we can but a bug in the yp bug tracker Aug 28 21:17:13 my suspicion is the problem is in there somehow Aug 28 21:17:28 I would do it, but I have no Ubuntu to try Aug 28 21:17:34 and am going afk for a few days :) Aug 28 21:17:51 Crofton|work, I'll try that. I also discovered this when I miss typed 'ls' in the same shell I was trying to build main.c in. http://pastebin.com/H2uNSzGk Aug 28 21:19:44 basically, I am hoping the bug can be fixed in oe-core :) Aug 28 21:19:57 Crofton|work, source envi...; MACHINE=qemuarm bitbake core-image... sorry still newb. Aug 28 21:20:29 basically, clone a seperate copy of oe-core only Aug 28 21:21:10 http://www.openembedded.org/wiki/OE-Core_Standalone_Setup Aug 28 21:21:44 will do... Aug 28 21:21:45 build core-image-minmal, then -c populate_sdk core-image-minimal Aug 28 21:21:53 that should go fairly quick Aug 28 21:21:59 brb Aug 28 21:27:51 Do you want me to build qemuarm or qemux86 Aug 28 21:30:33 Hello bluelightning! Aug 28 21:30:46 hi cfo215 Aug 28 21:30:52 Hello. Anyone know about any Ruby on Rails recipes/layers? Aug 28 21:31:46 kspr_: what is the issue at hand? Aug 28 21:32:48 lpapp_, no issues, really.. just searching for existing layers or recipes, before I start creating my own :) Aug 28 21:33:11 cfo215, I hope it doesn't matter :) Aug 28 21:33:40 bluelightning, shouldn't you be in the pub? Aug 28 21:34:35 Crofton|work: who told you he is not in a pub ;-) Aug 28 21:34:41 cfo215: I feel like I ought to be Aug 28 21:34:45 maybe tomorrow evening Aug 28 21:35:07 kspr_: meta-ruby? Aug 28 21:35:09 kspr_: there's only ruby itself that I know of (in meta-ruby) Aug 28 21:35:31 Yes, thanks, found the meta-ruby :) Aug 28 21:35:55 I think I'll be in Edinburg for the ELC. I'll buy you all beers then. Aug 28 21:36:11 \o/ Aug 28 21:36:27 The missus and I war going to do a working holiday. I'll work, she'll holiday ;) Aug 28 21:37:54 How far to the Glenlivit factory from Edinburg? Aug 28 21:38:15 damn, our local expert is not in channel atm Aug 28 21:38:20 I need to go mow Aug 28 21:38:34 hopefully, we can dupe the asdk issue in an oe-core environment Aug 28 21:39:16 cfo215, where are you based? Aug 28 21:39:27 Crofton|work, I'll let you know later how it goes. USA, North Carolina Aug 28 21:39:35 Asheville area. Aug 28 21:39:40 rofl Aug 28 21:39:49 Blacksburg, VA Aug 28 21:40:00 Asheville is full of Hippies Aug 28 21:40:06 I was just in DC over the weekend. Aug 28 21:40:09 * mranostay pokes in Aug 28 21:40:11 lovely place Aug 28 21:40:12 not Aug 28 21:40:14 someone say hippies Aug 28 21:40:16 I travel there about every four weeks. Aug 28 21:40:25 I've spent years avoiding moving there foe work Aug 28 21:40:44 no doubt Aug 28 21:41:17 Crofton|work: what are North Carolina hippies like? Aug 28 21:41:38 smelly, much like the ones everywhere Aug 28 21:41:42 :) Aug 28 21:41:54 * Crofton|work rode his bike up Mt Mitchell once Aug 28 21:42:01 Crofton|work, mranostay ... that's very true. Aug 28 21:42:02 maybe twice Aug 28 21:42:15 nice Aug 28 21:42:20 hello everybody! hmmm but newby question: I want to use yocto for distributing a binary as a *.deb package (this is done, working great) but I want to give the source as a tarball, is there a way to generate a tarball containing the source code, something like: bitbake perl -c tarball_create so it gets the source code, tars it, and put the tar file at the deploy directory Aug 28 21:42:23 I have nothing against them :) Aug 28 21:42:35 sorry for interrupt the conversation =P Aug 28 21:42:39 rofl Aug 28 21:42:55 carlos001_: see archiver.bbclass Aug 28 21:43:07 heh Aug 28 21:43:17 thank you! going to check it Aug 28 21:43:54 * lpapp_ always wondered when _end_ users have to check the class implementation details rather than having docs for the func Aug 28 21:43:55 this may be a while... 398/1636 tasks... Aug 28 21:44:00 why* Aug 28 21:44:11 quick is relative :) Aug 28 21:44:24 very true. Aug 28 21:44:49 I make wine and beer so I'm used to waiting on the end product... ;-) Aug 28 21:44:57 cool Aug 28 21:45:28 you'll have to come to our winery when we get it going. About five years out LOL. Aug 28 21:46:29 davest, would be interested in that Aug 28 21:48:14 I bet you know some West Virgina jokes... Aug 28 21:48:24 I've lived there also Aug 28 21:48:25 cfo215: a winery in Ashville? Aug 28 21:48:32 I'm sorry.. lmao Aug 28 21:48:55 davest, winery will be in Cashiers NC. Aug 28 21:49:13 davest, SW of Asheville about 1 hour. Aug 28 21:49:55 ah, small town NC Aug 28 21:50:10 * davest remembers the small NC towns that Mom and Dad grew up in Aug 28 21:50:47 yeah... small town surrounded by 28 gated communities... and 15 miles from Highlands... since you seem familiar with NC Aug 28 21:51:22 davest, what part of NC, mountains, peidmont or coast? Aug 28 21:52:09 cfo215: It's been a while: Baden for Mom and Dad grew up outside Greensboro on a farm Aug 28 21:52:30 cfo215: what varietal ? Aug 28 21:53:55 btw, I'll be in deepest darkest West Va this weekend starting tomorrow, may lose the tubes Aug 28 21:56:31 We've done Pinot Grigio, Merlot, Sangeovese, Cab Sav, Syrah, Reisling, some other that are escaping me at the moment. Aug 28 21:57:33 cfo215: Interesting mix. You might have heard we do a pretty mean Pinot Noir here in Oregon. :-) Aug 28 21:57:41 cfo215 Glenlivet is in the highlands, bout 3 hrs drive away from Edinburgh Aug 28 21:58:12 davest, Yes, I have Aug 28 21:58:50 Pacific Rim has a reisling we like. I think they are from there. Aug 28 21:58:51 cfo215: good, best to stay away from it. It will break your heart Aug 28 21:59:51 * davest afk to see a demo of webhob Aug 28 21:59:53 davest, stay away from what the Glenlivit or the Pinot Noir? Aug 28 22:00:16 Jefro, thank you for that. Aug 28 22:00:38 wine talk? Aug 28 22:01:00 everybody's wining about something around here Aug 28 22:01:06 heh that too Aug 28 22:01:47 yes we're changing #yocto to #winetalk... didn't you get the memo? LOL Aug 28 22:02:25 this is why RP worries about build times Aug 28 22:02:49 Robert Plant? LOL Aug 28 22:03:46 Crofton|work, normally my builds go pretty quick. But I haven't set up a shared source directory on my machine yet. Aug 28 22:04:01 .. so I'm having to download everything again. Aug 28 22:04:05 :( Aug 28 22:04:31 I need to RTFM to sort that out. Aug 28 22:05:08 Crofton|work: I do continue to worry about build times, yes Aug 28 22:07:40 RP, stupid question... anyone ever think of refactoring Bitbake to be an executable instead of a python script? Aug 28 22:08:29 you mean rewrite it in a compiled language? Aug 28 22:08:37 yes Aug 28 22:08:47 it's been discussed from time to time Aug 28 22:08:57 I figured it had. Aug 28 22:09:16 Its been more than discussed, its been attempted. Aug 28 22:09:31 It turns out most of the time is spent in the finalise() python code Aug 28 22:09:49 The memory resident bitbake is an attempt to massively cut down on various latencies Aug 28 22:10:20 Or, we could all get Crays... ;-) Aug 28 22:10:37 they're usually slower then a nice dual 8-core + HT box these days.. ;) Aug 28 22:10:43 So whilst rewriting it might seem like a nice idea, it is unlikely to solve the problem. There are other ways to speed it up Aug 28 22:11:01 We are slowly getting faster over time Aug 28 22:11:08 sure, more procs, solid state drives, etc. Aug 28 22:11:35 We've cut a lot of the disk IO, we filter out the fsync calls and so on Aug 28 22:11:45 configure is now our single biggest problem Aug 28 22:11:48 cfo edit DL_DIR to point at the same sir, just do not run two builds at once Aug 28 22:12:08 god configure scripts are slow. try running them in qemu sometimes, its horrifying Aug 28 22:12:25 kergoth: its horrific watching it in strace :( Aug 28 22:12:41 Its the shear number of exec()s :( Aug 28 22:12:44 * kergoth nods Aug 28 22:12:53 Crofton|work, thanks. I'll implement that. Save me some time in the long run. Aug 28 22:12:57 not helped by needing to use shell lanugage that is as portable as possible Aug 28 22:13:09 I've wondered if we should create our own autotools macro set Aug 28 22:13:21 i.e. massively slim them down Aug 28 22:14:07 RP, I'm attempting to register an assigned port for bitbake prserv. I noticed our protocol should support versioning. Aug 28 22:14:17 RP, Should I mail bitbake-devel? Aug 28 22:14:31 halstead: sounds good, yes please Aug 28 22:14:44 Incidentally I should have some fixes for prserver shortly Aug 28 22:14:52 its currently single threaded and causing bottlenecks Aug 28 22:14:57 RP, cfo215 is trying to dupe an sdk issue he has with an sdk from Angstrom in oe-core Aug 28 22:15:05 some weird thing Aug 28 22:15:17 obviously, i am hoping it is an oe-core issue Aug 28 22:16:07 Crofton|work: I look forward to your fixes? ;-) Aug 28 22:16:12 rofl Aug 28 22:16:28 getting closer... 1311 of 1636 Aug 28 22:16:41 I am just trying to help him get a manageble test case :) Aug 28 22:17:19 if we can't use popualte_sdk to build and run sdk on ubuntu lts, I think that is an isse :) Aug 28 22:19:54 uname -a yields "Linux my-pc-name-here 3.5.0-39-generic #60~precise1-Ubuntu SMP Wed Aug 14 15:38:41 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux" if that helps. Aug 28 22:42:53 Crofton|work, running bitbake -c populate_sdk core-image-minimal right now. Aug 28 22:56:26 RP: this evening I get arm-oe-linux-gnueabi-ld: cannot find libgcc.a: No such file or directory Aug 28 22:56:36 any recent change ? Aug 28 22:57:18 ant_home: no, probably a build from sstate which isn't quite working properly? :/ Aug 28 22:57:24 argh Aug 28 23:05:04 drum roll Aug 28 23:14:31 pidge: I assume you saw this? Aug 29 01:07:30 RP: yes, that was me. I killed a bunch of builds **** ENDING LOGGING AT Thu Aug 29 02:59:58 2013