**** BEGIN LOGGING AT Mon Aug 11 02:59:58 2014 Aug 11 05:29:34 Hi yocto. Aug 11 05:34:49 I compiled qt5 from meta-qt5 after compilation. if i try to compile qt application by passing qmake i am getting following error " Inconsistency detected by ld.so: get-dynamic-info.h: 134: elf_get_dynamic_info: Assertion `info[15] == ((void *)0)' failed! " please any one tell me what is the problem and to resolve this error. Aug 11 06:47:16 Hi please help me to resolve this issue. Because i am struggling a lot due to this error. Aug 11 06:49:51 what is in that header file? Aug 11 06:50:06 * nerdboy is no c++ guru... Aug 11 06:51:27 qt isn't really my thing either, but the linker apparently doesn't like what you're linking against Aug 11 06:52:09 googling that error message only gives very obscure results, so without sufficient detail information, i guess there's not much helping. Aug 11 06:56:19 all i got is "hi, capt obvious, something in your qmake env is different" Aug 11 06:56:39 not very helpful... Aug 11 06:56:55 morning Aug 11 06:56:57 *i'm Aug 11 06:57:24 nerdboy: yes, and: "i patched the haader" Aug 11 07:00:42 its always the same with this person... obscure error message, no details provided. Aug 11 07:54:58 Hi please help me to resolve this issue. Because i am struggling a lot due to this error. Aug 11 08:15:38 LetoThe2nd: count fenrig is the guy i'm thinking of... Aug 11 08:17:07 nerdboy: nah, fenring actually was capable of own effort Aug 11 08:17:37 not that, his job... Aug 11 08:18:20 nm, it was more of a humorroid anyway... Aug 11 08:18:26 you mean like "dirty work"? ;) Aug 11 08:19:41 i'm sure he would say "do what's necessary" or some such Aug 11 08:20:19 but still, fenring was capable of "doing what's necessary" ;) [spot the difference] Aug 11 08:20:58 i'm not arguing... Aug 11 08:21:07 embed: AGAIN: googling that error message only gives very obscure results, so without sufficient detail information, i guess there's not much helping. Aug 11 08:21:19 nerdboy: \o/ Aug 11 08:22:53 nerdboy: ok :) Aug 11 08:23:00 LetoThe2nd: nerdboy: agreed :) Aug 11 08:23:33 * LetoThe2nd hands out some 10gr of spice to nerdboy and JaMa Aug 11 08:23:54 * JaMa eyes glowing Aug 11 08:24:55 JaMa: nm the crazy the crazy driver, but make sure you give ka6sox your top 10 hit list Aug 11 08:25:10 more blue enlightened code today. hooray! Aug 11 08:25:49 however many are the most annoying i guess... Aug 11 08:26:37 morning all Aug 11 08:27:09 not sure how popular it will be, but we should have maybe 6 people and the same build box you got from him Aug 11 08:27:15 * LetoThe2nd also mourns, espcially his latest duncan Aug 11 08:27:35 for several hours at least Aug 11 08:28:23 note: get some dune clothing for düsseldirf ;) Aug 11 08:28:31 wait, is he psychic? Aug 11 08:28:58 me? certainly. Aug 11 08:29:23 can you channel eric raymond? Aug 11 08:29:42 hrhrhr nope Aug 11 08:33:08 hmm, sort of a freudian typo on her driving... Aug 11 08:37:19 nerdboy: I've ~ 35 lines of QA log and 3 failing recipes, you can pick from these :) Aug 11 08:37:59 i have bunch of QA warnings too Aug 11 08:38:10 fails are even better Aug 11 08:38:46 LetoThe2nd: are you Jodorowsky's son, or Leto from Lynch? :) Aug 11 08:39:33 this is still a next weekend thing for the group "sprint" Aug 11 08:40:12 be working on my stuff in between other stuff Aug 11 08:40:23 * nerdboy not lacking in stuff Aug 11 08:40:35 JaMa: neither.. http://en.wikipedia.org/wiki/Leto_Atreides_II Aug 11 08:40:38 except sleep maybe... Aug 11 08:40:48 JaMa: (short form: I appear later in the series) Aug 11 08:41:42 Occupation: God Emperor Aug 11 08:41:47 ah good, nice to meet you :) Aug 11 08:42:18 and he's psychic even Aug 11 08:43:15 JaMa: Tyrannic God Emperor, please. Aug 11 08:43:26 yes, my Lord Aug 11 08:44:01 just don't ask him to channel any famous open source developers... Aug 11 08:44:12 * LetoThe2nd nods forgivingly. Aug 11 08:44:34 nerdboy: are you messing up my powers with some fiction? ;) Aug 11 08:44:52 i was thinking stallman's toenails... Aug 11 08:45:00 nm... Aug 11 08:45:28 too far? Aug 11 08:46:01 past my bedtime anyway, so... Aug 11 08:46:18 ;) Aug 11 08:46:20 * nerdboy yawns prodigiously Aug 11 08:46:25 night Aug 11 08:46:54 g'night Aug 11 08:48:41 nite Aug 11 09:03:01 hi Aug 11 09:04:22 is there a way to instruct the git fetcher to pull into an existing repo, rather than re-cloning the whole thing when the SRCREV in a recipe changes? Aug 11 09:06:32 good morning Aug 11 09:08:39 LetoThe2nd:then please tell me any other possible ways to solve this. Aug 11 09:09:17 is there a way to tell the -dbg package not to generate the source only the unstripped binaries? Aug 11 09:09:39 (in case if you are wondering: use case is using gdbserver where the source files just bloat the very small flash) Aug 11 09:09:50 LetoThe2nd:I could not understand what is the problem weather the toolchain issue or qt qmake issue. Aug 11 09:09:50 embed: show source, recipe, full logs. you know. details. Aug 11 09:19:41 lpapp: PACKAGE_DEBUG_SPLIT_STYLE = "debug-without-src" in your recipe, i believe Aug 11 09:19:54 danielki: I thought it would just do a fetch, tbh Aug 11 09:20:16 rburton: hm weird Aug 11 09:20:48 lpapp: meta/conf/local.sample.extended has a 'section' about customizing the debug package, but this option doesn't seem to be mentioned there. but it should work. Aug 11 09:29:03 ndec: you are the man :-) Lemme try. Aug 11 09:29:13 (I will open a request to Scott to put it into the manual) Aug 11 09:29:41 ndec: OMG! you are the MAN! Aug 11 09:30:09 ndec: http://en.wikipedia.org/wiki/The_Man my condolences. can i do something to comfort you? Aug 11 09:31:16 LetoThe2nd: i think he really meant this variant: http://en.wikipedia.org/wiki/The_Man#Use_as_praise ... you da man! Aug 11 09:31:36 what would be the place for it if I always want it to behave like this because we tend to prefer gdbserver... in fact we do not have choice as our flash is small. Aug 11 09:31:41 the distro.conf? Aug 11 09:31:51 ndec: hey i'm a kid of the eighties, i'm unable to read beyond the first paragraph ;) Aug 11 09:33:54 hmm, probably the local.conf sample in the custom layer just like in Yocto: PACKAGE_DEBUG_SPLIT_STYLE = 'debug-file-directory' Aug 11 09:34:26 distro config Aug 11 09:35:49 rburton: it seems you are right; looks like I need to get back to the customer :) Aug 11 09:43:20 ndec: https://bugzilla.yoctoproject.org/show_bug.cgi?id=6621 Aug 11 09:43:21 Bug 6621: normal, Undecided, ---, scott.m.rifenbark, NEW , Document debug-without-src Aug 11 09:46:07 blitz00: ok, thanks. Aug 11 09:46:15 oops, I meant @bluelightning Aug 11 10:21:54 ndec: got a clue for this? http://paste.kde.org/pdajdhapy Aug 11 10:22:32 ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.14, not stripped -> this is what file returns on the host. Aug 11 10:23:00 <[Sno]> aloah Aug 11 10:23:15 hola Aug 11 10:23:19 <[Sno]> I created some recipes and have trouble with dependencies ... Aug 11 10:23:52 <[Sno]> eg. adding that https://github.com/rehsack/meta-cpan/blob/master/recipes-www/dancer2-perl/dancer2-perl_0.143000.bb to an image doesn't result in app-cmd-perl is included in final image, too Aug 11 10:24:00 show architecture Aug 11 10:24:00 The target architecture is set automatically (currently armv5t) Aug 11 10:24:05 this is with gdb on the host ... Aug 11 10:24:17 <[Sno]> I tried DEPENDS += and RDEPENDS_${PV} += Aug 11 10:25:15 <[Sno]> any idea what I make wrong? Aug 11 10:31:33 gdbserver can debug one, but not the other: file foo: foo: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.14, not stripped Aug 11 10:31:37 file foo-stripped: foo2-stripped: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.14, not stripped Aug 11 10:41:43 [Sno]: change RDEPENDS_${PV} to RDEPENDS_${PN} Aug 11 10:47:47 it is kind of strange because I manually copy the binary over, it just works... I wonder if Yocto makes something to the binary? Aug 11 10:47:51 (before packaging) Aug 11 10:47:56 if I* Aug 11 10:48:21 or why would gdbserver detect arch-error with the -dbg package compared to manual copying? Aug 11 10:57:13 <[Sno]> bluelightning: thanks - seen that typo while stepping through yocto-rev-doc :) Aug 11 10:58:03 <[Sno]> re-run generator and baking now :) Aug 11 11:27:57 gdb is horrible for debugging.... could Yocto not get cgdb into gdb-cross? Aug 11 11:28:12 RP: thanks! :) Aug 11 11:28:35 since it is just host-side thing, it would not bloat anything much either, but debugging would become a muuuuch smoother experience. Aug 11 11:30:02 or at least having an option to turn it on or something. Aug 11 11:30:04 Good evening all , am facing this issue for long time. Am building images for at91sam9x5ek , while compile kernel an getting following error "arm-poky-linux-gnueabi-ld: no machine record defined". And there is no board-dt.c file "mach-at91" directory. Here is the complete error "http://pastebin.com/NnGpk6k3" Aug 11 11:32:16 can any one help me to fix this Aug 11 12:00:04 I do not understand Yocto.... I got the checksum issue when creating a recipe based on another, and that is fine. I updated the checksum with the expected ones, and then bitbake -c cleanall foo, yet, I am getting the same complain for bitbake foo. :-( What am I missing? Aug 11 12:03:05 here you can see the error log, fwiw: http://paste.kde.org/pifcmrcpa Aug 11 12:05:11 ah, hmm, never mind, my bad! Aug 11 12:15:19 it is interesting to see that some packages specify -native for the dependency, too... why is that not auto-detected when a native package is built, it oughta to use the native dependency? Aug 11 12:25:32 lpapp: it is; sometimes the target package needs -native, that's when you see it specified explicitly Aug 11 12:28:45 Is this fixable ?? Aug 11 12:31:37 bluelightning: when is -native needed? Aug 11 12:32:08 lpapp: when some part needs to be run on the build host ... Aug 11 12:32:58 I am not sure I understand. That is what I meant by auto-detection: if it is build host, then it is native, naturally. Aug 11 12:33:21 you can't auto-detect it Aug 11 12:33:22 lpapp: some target packages need to run host binaries that they'd normally build directly. Aug 11 12:33:41 e.g. tcltk has this: DEPENDS = "tcl-native" Aug 11 12:33:42 eg wayland depends on wayland-native Aug 11 12:34:20 because wayland-native installs the wayland-scanner tool Aug 11 12:34:30 which wayland-target then uses to generate code Aug 11 12:34:57 yeah, but buildtime always uses native tools. Aug 11 12:35:05 so why not autodetect it? Aug 11 12:35:23 how? Aug 11 12:35:32 wayland-target will run and fail because wayland-scanner hasn't been built Aug 11 12:36:17 well, you specify wayland dependency for wayland, I would assume. Aug 11 12:36:24 rburton: I've just noticed that I still have "Revert "package.bbclass: add a stub implementation of package_name_hook"" in my branch, was debian.bbclass issue already resolved? Aug 11 12:36:25 without -native since it would be implicit. Aug 11 12:37:10 JaMa: oe-core 3aa52f85e774bf82e8470a8fd849bd6953930720 Aug 11 12:37:14 (hopefully) Aug 11 12:37:34 ok, will try to drop it in next build, tahnks Aug 11 12:37:36 thanks Aug 11 12:38:09 lpapp: i dont see what the point of changing an explicit depends=wayland-native to wayland depending on wayland and the self-dependency magically turning into a native dependency would be Aug 11 12:38:32 lpapp: especially as other things will need wayland-native *and* wayland so its not a general solution Aug 11 12:38:39 rburton: it is not so obvious for self-dependencies, but it is beneficial for non-self-dependencies. Aug 11 12:38:54 rburton: since you just specify what you need rather than adding another layer of abstraction. Aug 11 12:39:18 it is a general solution as far as I see so far. Why would it not be? Aug 11 12:43:36 is it possible that I made my autotools based recipe wrong? I am seeing this: --build=x86_64-linux --host=arm-foo-linux-gnueabi --target=arm-foo-linux-gnueabi Aug 11 12:43:59 whereas, it should be this: --build=x86_64-linux --host=x86_64-linux --target=arm-foo-linux-gnueabi Aug 11 12:44:49 and hence I see this issue: checking for /dev/ptmx... configure: error: cannot check for file existence when cross compiling Aug 11 12:45:21 here you can see my simple recipe... http://paste.kde.org/pqwkuse4j Aug 11 12:45:38 isn't Yocto supposed to pass the right host parameter to the autotools based build system? Aug 11 12:45:52 after I just say "inherit autotools" and be done with it? Aug 11 12:46:17 should be fine Aug 11 12:46:56 my config.log is 157 KB, so I would not like to paste unless it is really necessary ... Aug 11 12:47:27 pl help...i've poky tool chain ready and i've configured qt-creator for that and during cross compile ii'm getting error of make ...pl if you can see the post http://qt-project.org/forums/viewreply/189080/ and help me Aug 11 12:47:27 but cgdb seems to look for /dev/ptmx, and that should be available on the host machine (in the debian wheey chroot): lrwxrwxrwx 1 root root 8 Aug 11 09:38 /dev/ptmx -> pts/ptmx Aug 11 12:47:42 should I manually increment PR when updating git recipes to a newer SRCREV? Aug 11 12:47:50 rburton: debian.bbclass confirmed Aug 11 12:47:55 Yocto seems to be passing --host=arm-foo-linux-gnueabi, why not x86_64-linux, the actual host? Aug 11 12:48:07 Net147: wasn't PR obsolete? Aug 11 12:48:11 (a long ago) Aug 11 12:49:24 Guest8954: coincidentally, I came across such an issue today with a Yocto generated dbg package myself, too: *sh: /opt/poky/1.6.1/sysroots/cortexa9hf-vfp-neon-poky-linux-gnueabi/usr/share/qt5/examples/quick/demos/clocks/clocks: cannot execute binary file: Exec format erro Aug 11 12:49:43 I have no idea why Yocto generates a broken package for me, when manually copying the binary from the build folder works OK. Aug 11 12:52:08 Net147: if you set up the PR server you don't need to do that Aug 11 12:52:48 lpapp: can you pl tellhow did you do that Aug 11 12:52:58 bluelightning: okay, thanks. I just wanted to make sure it's not going to think it's an older version when the SRVREV becomes lexicographically less after updating. Aug 11 12:56:29 Guest8954: sorry, I am not much into Qt and not a pro about Yocto either. You need to wait for someone else. Aug 11 12:57:57 lpapp: ok .....but i was just asking that how did you manually copied Aug 11 12:58:13 scp :) Aug 11 12:59:08 rburton: do you see anything wrong about my short recipe above then? Aug 11 12:59:18 lpapp: actually its ok for demo purpose...but what when we have to develop a custome program in qt-creator? Aug 11 12:59:36 since it is just the usual ./configure --prefix=/usr && make && make install, I have not done anything else than "inherit autotools", hoping that it would be OK. Aug 11 13:00:09 Guest8954: I do not know, my ide IDE is vim :) Aug 11 13:00:17 my IDE* Aug 11 13:00:42 lpapp: that ok....anyways thanks for your suggestions Aug 11 13:00:55 Guest8954: I am using Qt Creator fine with Yocto generated Qt 5 SDK. cross-compiling, deploy to target, remote commands, run on target and debug (C++ and QML) on target working. Aug 11 13:01:51 Net147: can you please help me for this problem posted here http://qt-project.org/forums/viewreply/189080/ Aug 11 13:02:40 Guest8954: btw, I used to be active on Stack Overflow, you may want to submit Qt questions there, too. There were a few chaps helping with Qt. Aug 11 13:03:03 Guest8954: there, you can even tag your question as qt + openembedded-core. Aug 11 13:03:23 openembedded* Aug 11 13:03:56 lpapp: ok....i'll try there too...thanks Aug 11 13:04:25 Guest8954: are you running . /path/to/environment-setup-script in terminal before running qtcreator? Aug 11 13:04:49 Net147:yes Aug 11 13:06:55 Guest8954: compiler path should point to arm-*-g++, debugger should be arm-*-gdb Aug 11 13:08:21 Net147: compiler pathy has been pointed there where there is the location of cross-compiler ...i'll do that for debugger too Aug 11 13:08:56 rburton: my gdb config.log shows this, too: --host=arm-foo-linux-gnueabi Aug 11 13:09:06 apparently, gdb did not get upset about it like cgdb. Aug 11 13:09:13 Guest8954: Qt mkspec should be linux-oe-g++ Aug 11 13:09:19 but this is wrong.... Yocto oughta pass the correct host. Aug 11 13:09:36 so why is Yocto not passing the correct --host information to an autotools based software? Aug 11 13:10:44 Net147:referring to http://wiki.wandboard.org/index.php/Building_Qt5_using_yocto_on_Wandboard i've given the mkspec....is that wrong? Aug 11 13:11:34 Guest8954: not sure. I don't build for wandboard. Aug 11 13:12:23 Guest8954: it should be the same as what you get when you do qmake -query after sourcing the environment script Aug 11 13:12:38 Net147: but the real error lies on the path of make which is creating mischiefs....can you just look at the [8] url Aug 11 13:13:21 Guest8954: do you have make installed on the host? Aug 11 13:14:05 Guest8954: after you source the environment script. do "which make" Aug 11 13:14:29 Net147: it gives the make file exc error...and if i delete the make path the build error will go away...also the oup file is also not generated Aug 11 13:15:13 Guest8954: /opt/poky/1.6.1/sysroots/cortexah9f-vfp-neon-poky-linux-gnueabi/usr/bin/make exists? Aug 11 13:15:27 Net147: yes Aug 11 13:15:51 Guest8954: what is the output of "file /opt/poky/1.6.1/sysroots/cortexah9f-vfp-neon-poky-linux-gnueabi/usr/bin/make"? Aug 11 13:16:09 Net147: which make ...output:/usr/bin/make Aug 11 13:17:06 Guest8954: in my SDK, there is no make. it should use the host make. Aug 11 13:18:30 Net147:file /opt/poky/1.6.1/sysroots/cortexa9hf-vfp-neon-poky-linux-gnueabi/usr/bin/make /opt/poky/1.6.1/sysroots/cortexa9hf-vfp-neon-poky-linux-gnueabi/usr/bin/make: ELF 32-bit LSB executable, ARM, EABI5 version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.16, BuildID[sha1]=e015ff08b4459868a9b9ca3c65f81e74970a78c9, stripped Aug 11 13:19:11 Guest8954: if you rename /opt/poky/1.6.1/sysroots/cortexah9f-vfp-neon-poky-linux-gnueabi/usr/bin/make to /opt/poky/1.6.1/sysroots/cortexah9f-vfp-neon-poky-linux-gnueabi/usr/bin/make.disabled, does it help? Aug 11 13:20:29 Net147: let me try ...this Aug 11 13:21:50 Guest8954: also, run qmake again Aug 11 13:22:04 Guest8954: from Qt Creatort Aug 11 13:23:11 rburton: most autotools based projects ignore everything, but --build Aug 11 13:23:25 rburton: even my strace and gdb config.logs have the wrong --host parameter supplied by Yocto. Aug 11 13:23:46 so it seems to be an overall issue in Yocto which was not caught by any error yet. Aug 11 13:25:16 Net147:no ...the same problem Aug 11 13:27:34 Guest8954: after running qmake too? Aug 11 13:28:23 Net147:running qmake means....i've just provided the path in qt-creator Aug 11 13:28:51 Guest8954: Build > Run qmake Aug 11 13:30:06 Net147: yep...the same error Aug 11 13:30:56 Guest8954: not sure then.. Aug 11 13:31:52 bluelightning: rburton: try to reproduce it, bitbake strace or bitbake gdb, and check log.do_configure, it shows the target for --host, too. Aug 11 13:31:54 Net147:ok......for which board you have configured qt Aug 11 13:32:01 (I will file a bugreport) Aug 11 13:32:25 Guest8954: I am using meta-sunxi layer for board support Aug 11 13:33:10 Net147:ok...anyways ...thanks for your suggestions...... Aug 11 13:33:35 Guest8954: my SDK is generated separately from my image Aug 11 13:34:20 Net147:o...i see Aug 11 13:35:42 Guest8954: I have a meta-toolchain-myplatform.bb recipe that has "inherit populate_sdk populate_sdk_qt5", and I add various package-dev, etc. packages to TOOLCHAIN_TARGET_TASK Aug 11 13:35:59 Guest8954: then I bitbake meta-toolchain-myplatform to generate the SDK Aug 11 13:36:21 rburton: bluelightning https://bugzilla.yoctoproject.org/show_bug.cgi?id=6623 Aug 11 13:36:24 Bug 6623: normal, Undecided, ---, richard.purdie, NEW , Autotools: --host contains the target for the ./configure script Aug 11 13:36:43 Guest8954: I only add the bare minimum I need for development as needed into TOOLCHAIN_TARGET_TASK Aug 11 13:37:13 Net147: gr8.. Aug 11 13:38:23 Net147:wru located Aug 11 13:38:44 Guest8954: Australia Aug 11 13:39:04 Net147:ok...i'm in India Aug 11 13:48:14 k Aug 11 13:51:32 is it possible with Yocto to instruct the buildsystem not to rebuild anything just what I want? Aug 11 14:15:41 Im trying to build the kvaser and canlib drivers via bitbake recipe and have realized I have no idea what Im doing... Aug 11 14:16:07 haha Aug 11 14:16:57 can someone help me create a recipe for http://www.kvaser.com/support/downloads/ (Kvaser linux Driver and SDK) .... im willing to paypal for it if it works ... or walk me through creating a recipe Aug 11 14:17:02 for drivers Aug 11 14:17:09 this is so frusterating :( Aug 11 14:17:16 i literally spent all weekend on it Aug 11 14:17:27 is this just a binary blob? Aug 11 14:17:48 naw it builds Aug 11 14:17:53 using make and make install Aug 11 14:18:23 so I read about writing a recipe for stuff with make files. .. but i think that just confused me more Aug 11 14:19:38 joranb: try to customize this, http://paste.kde.org/pxfhls0dy Aug 11 14:19:48 replace the meta data with your stuff, and it should just work. Aug 11 14:19:56 (if the software only requires make and make install) Aug 11 14:20:47 hah ill try it ... that looks really simillar to what I tried Aug 11 14:20:53 but thanks alot :) Aug 11 14:21:04 so what was the error? Aug 11 14:21:09 can you show your current recipe? Aug 11 14:22:04 yeah sure let me goto my dev box Aug 11 14:22:13 gimme 3 mins or so Aug 11 14:25:12 oh thats part of the problem it doesnt create the main file just [kvaser-dbg-0.0-r0.pepper.rpm kvaser-dev-0.0-r0.pepper.rpm] Aug 11 14:27:25 here is my (albeit crappy) recipe Aug 11 14:27:26 https://gist.github.com/anonymous/2d6370da5a6f2fa03de3 Aug 11 14:29:11 and this is what I get when I run my bitbake Aug 11 14:29:12 https://gist.github.com/anonymous/0606b9f449adf7473219 Aug 11 14:40:43 @lpapp I have to head to work ... but i would be forever in your debt if you could take a look at my recipe and the source my email is joranbeasley*at*gmail.com ... If you can help me get this working I would gladly send you some paypal funds :P Aug 11 14:41:24 if you dont have time or whatever of coarse I also understand and thank you for that example ... I might be able to get that working Aug 11 14:49:09 joranb: sorry, I was away. I will have a look later if I do not forget it. I do not need any paypal funds. :p Aug 11 14:51:49 joranb: looks like it has already been built to me, so it's not doing anything the second time Aug 11 14:52:07 joranb: I'm not sure if you have seen it but we have a new manual section on writing recipes here: http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#new-recipe-writing-a-new-recipe Aug 11 14:52:43 looking at your recipe, I suspect you need to add some extra pieces to tell the system how the software should be built / packaged Aug 11 14:54:46 joranb: what exactly is the issue? The packages are empty? Aug 11 14:55:26 joranb: is this some socketCAN thing, btw? Aug 11 14:56:28 it is Aug 11 14:56:29 joranb: have you checked if the driver is built in the build folder? Aug 11 14:57:13 joranb: do you have a complete URL to the tar.bz2 file? Aug 11 14:57:14 all I could find was the dev and dbg files ... neither of which worked Aug 11 14:57:34 http://www.kvaser.com/software/7330130980754/V5_3_0/linuxcan.tar.gz Aug 11 14:57:53 but its gz ... but doubt that matters... Aug 11 14:58:14 im going to feel really stupid if its just cause I had to put the url insteado f my local tar.bz2 Aug 11 14:58:58 joranb: I suspect at minimum you need to define a do_install Aug 11 14:59:30 oh I thought if make had an install rule it would use that automagically Aug 11 14:59:51 joranb: you are missing the enumeration of the files for the package. Aug 11 15:00:02 anyway Im really late for work now ... I'll get back on in 20 mins Aug 11 15:00:19 joranb: I would expect that to happen to be honest, if not I would argue that it is a shortcoming in Yocto. Aug 11 15:00:22 joranb: if the software uses autotools and you inherit autotools that will work, yes; if not, you do need to define your own do_install Aug 11 15:00:24 I thought i just had to point it to the bz file ... heh you can tell im totally confused Aug 11 15:00:44 oh Its becoming marginally clearer Aug 11 15:01:05 I will be back on in 20 minutes ... I hope you guys are still around because this is helpful pointers Aug 11 15:01:11 yep I'll be here Aug 11 15:01:14 :) Aug 11 15:03:00 lpapp: bitbake -b will skip dependencies and just do what you tell it to do Aug 11 15:03:41 rburton: I changed autotools.bbclass for a quick test against a package, but everything got rebuilt. Aug 11 15:04:10 yes Aug 11 15:04:18 thanks. Aug 11 15:04:26 the hash changed, so it has to rebuild everything Aug 11 15:04:38 thats why when testing changes in automake, pick something with minimal other depends :) Aug 11 15:04:42 i tend to use m4 Aug 11 15:07:10 rburton: I see.... Aug 11 15:08:21 bluelightning: why cannot Yocto deal with proper make install targets? Aug 11 15:09:21 proper: make install already has the rules where to install what. Aug 11 15:10:38 joranb: try do_install and/or FILES_${PN}* Aug 11 15:11:34 rburton: I have another question if you do not mind ... do you know why my cgdb binary does not get renamed to the usual long name like other toolchain file names? Aug 11 15:12:11 no idea, never looked at what renames the toolchain Aug 11 15:12:23 I am referring to ./tmp/sysroots/x86_64-linux/usr/bin/armv5te-foo-linux-gnueabi/arm-foo-linux-gnueabi-* Aug 11 15:12:36 my cgdb is outstanding there because it is just ./tmp/sysroots/x86_64-linux/usr/bin/armv5te-foo-linux-gnueabi/cgdb Aug 11 15:12:47 all the rest got the arm-foo-linux-gnueabi- prefix. Aug 11 15:13:13 first I thought "inherit cross" should take care of that, but since it does not, I am not sure anymore. Aug 11 15:29:52 lpapp: probably because it can't be sure whether or not a "make install" target exists Aug 11 15:30:45 or whether it respects a destdir variable and if it does, what it's called. Aug 11 15:30:46 bluelightning: why is that an issue? If it does it succeeds, otherwise it fails, yeah? Aug 11 15:42:11 lpapp: fails because it's not available or because it just failed for another reason? Aug 11 15:42:39 bluelightning: what do you mean? You have the build log, you can check what the buildsystem says, right? Aug 11 15:42:48 (as an end user) Aug 11 15:48:34 im back .... Aug 11 15:49:09 so I probably do need more than just metadata? (IE a do_install() routine?) Aug 11 15:49:14 yes Aug 11 15:49:17 (sadly IMHO) Aug 11 15:49:50 ok ... well thats probably fine ... basically I can try and convert the install rule to bitbakiness Aug 11 15:50:20 but the do_compile will automatically run make so I dont need to explicitly define that? Aug 11 15:50:21 joranb: would you like one single package to "catch all" or modularized small packages? Aug 11 15:50:37 one singe package is fine for my needs ... Aug 11 15:51:01 or whichever is easier (I was assuming that one) Aug 11 15:51:22 and the memory difference is trivial Aug 11 15:52:50 do you guys just figure this stuff out through trial and error usually? Its hard to find documentation with complete examples that are relevant (they are all relevant and building blocks...) and/or tutorials ... and stackoverflow hasnt quite caught up with yocto it seems Aug 11 15:56:44 I feel like an idiot though ... it seems like all the info is out there I just havent put it together in my head right yet Aug 11 15:57:24 joranb: stackoverflow has an openembedded tag Aug 11 15:57:27 even bitbake. Aug 11 15:57:40 there's no standard for how to specify where do_install will put its files Aug 11 15:57:45 for ordinary makefile based buildsystems Aug 11 15:57:51 they're whatever people happen to choose Aug 11 15:58:03 so there's no way to provide a default do_install for make based buildsystems in a way that would actually work Aug 11 15:58:15 s/do_install will/make install will/ Aug 11 15:58:44 autotools.bbclass is able to provide a default do_install because automake has a set of conventions for such things, including DESTDIR, which is the important one Aug 11 16:00:33 yeah Ive seen the flags ... I also notice most questions have no answers ... (a few do) ... to be fair its hard to correctly frame yocto questions in a way that people are able to help Aug 11 16:00:54 (in the snippet that you have to do it in with stack overflow) Aug 11 16:01:53 @kergoth thanks for the explanation on why you need to provide do_install with make file :) Aug 11 16:02:40 eh actually lots do have answers on stackoverflow I guess ... for bitbake tag at least Aug 11 16:03:21 joranb_: if you haven't already I'd really recommend having a look at the manual section I linked above; part of the reason I pushed for creating that was to try to make it easier for new users to write their first recipe, so I hope that it's helpful for that :) Aug 11 16:03:37 hmm, cgdb still does not work :( Aug 11 16:03:48 is it ok to specify gdb for the dependency or it oughta be gdb-cross? Aug 11 16:03:49 if it could be improved, we'd definitely like feedback Aug 11 16:04:00 I am lost by this -native -cross -foo complication :( Aug 11 16:04:34 (and still cannot see why it cannot be made transparent and underlying details only) Aug 11 16:04:35 meeting brb Aug 11 16:04:41 one minute Aug 11 16:06:03 rburton: do you know if a cross-host tool depends on a cross-host tool I should specify dep-native or dep-cross or just simply dep? Aug 11 16:11:46 @BlueLightning ... I didnt see the link can you repost it? Aug 11 16:11:53 sure one sec Aug 11 16:11:55 chances are high Ive already looked at it Aug 11 16:12:02 http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#new-recipe-writing-a-new-recipe Aug 11 16:12:16 @lpapp is that cgdb question re that canlib thing? Aug 11 16:13:06 joranb_: nah, that is my own little isuse :p Aug 11 16:13:14 issue*, btw, @foo is too SO'ish :) Aug 11 16:13:27 moin Aug 11 16:15:36 lol how do I whisper or whatever(im using the webclient)? or if i just put "username: msg" it will highlight it for the other user? Aug 11 16:15:46 nerdboy: hey, was it you with whom I discussed the gdb thing last week? Aug 11 16:16:06 you're asking me? Aug 11 16:16:11 joranb_: just use the name without prefix/suffix, but I am fine with @foo. Aug 11 16:16:18 nerdboy: you cannot have worse memory than mine :) Aug 11 16:16:27 i yell a lot of stuff your way... Aug 11 16:16:59 probably, since my kids are probably your age Aug 11 16:17:14 at least closer than i am... Aug 11 16:17:33 nerdboy: hmm, yeah, you helped me about INHIBIT_PACKAGE_DEBUG_SPLIT = 1 Aug 11 16:17:47 yup Aug 11 16:17:53 do you happen to know if I need to depend on gdb or gdb-cross when building my own little cgdb? Aug 11 16:18:08 currently I depend on gdb, but I cannot communicate with the gdbserver. Aug 11 16:18:16 (whereas gdb-cross can) Aug 11 16:18:55 gdb-cross would make sense Aug 11 16:19:11 yes, I was just trying that. Aug 11 16:19:16 should be what goes in the adt Aug 11 16:19:19 but it did not seem to work either. :/ Aug 11 16:19:34 doesn't work how? Aug 11 16:19:38 I can connect to the gdbserver no problem, but when I press continue, it exists the gdbserver. Aug 11 16:19:48 exits* Aug 11 16:20:19 these are my current depends: DEPENDS += "readline ncurses gdb" Aug 11 16:20:30 I wonder if I should modify readline and ncurses there, too. Aug 11 16:21:27 lpapp: heh http://stackoverflow.com/questions/21284008/bitbake-fails-on-compiling-gcc-cross I doubt that helps ... but just saw it Aug 11 16:22:59 yeah, it is a bit different thing, but thanks :) Aug 11 16:23:22 nerdboy: (gdb) c Aug 11 16:23:22 Continuing. Aug 11 16:23:23 [Inferior 1 (Remote target) exited with code 01] Aug 11 16:23:28 (after "b main") Aug 11 16:23:28 maybe try strace gdb? where are your debug symbols installed? Aug 11 16:23:49 currently on the host as well as the target (just in case) Aug 11 16:23:55 but the host should be enough. Aug 11 16:24:09 because I just copied the binary over because Yocto breaks the -dbg package for me :( Aug 11 16:24:17 -because Aug 11 16:24:21 you should be able to crank up the debug output on the debug... Aug 11 16:24:26 (but that is another issue for another day ... ) Aug 11 16:24:36 just not in a public place Aug 11 16:24:37 if I use the same binaries, gdb works. Aug 11 16:24:55 so I only replace cgdb with gdb in the flow, and then I can debug. Aug 11 16:25:18 so it is something to do with the cgdb build... my suspicion is that it is not compatible with the gdbserver, i.e. not using the same gdb that I am running from the command line. Aug 11 16:25:30 I do not know how to make sure ... Aug 11 16:25:50 have you built the adt and enabled debug stuff in your build? Aug 11 16:26:20 this does not work: ./tmp/sysroots/x86_64-linux/usr/bin/armv5te-foo-linux-gnueabi/cgdb -q ./tmp/work/armv5te-foo-linux-gnueabi/foo/$myver/git/foo -ex "tar rem 192.168.0.1:2345" Aug 11 16:26:43 this does work: ./tmp/sysroots/x86_64-linux/usr/bin/armv5te-foo-linux-gnueabi/arm-polatis-linux-gnueabi-gdb -q ./tmp/work/armv5te-foo-linux-gnueabi/foo/$myver/git/foo -ex "tar rem 192.168.0.1:2345" Aug 11 16:27:30 both are ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26 Aug 11 16:27:47 other than getting more data (strace, more -DDDs, whatever) i don't have much else Aug 11 16:28:14 strace on the target (gdbserver) and/or the host (c/gdb)? Aug 11 16:28:51 easier to try host first Aug 11 16:29:39 nerdboy: btw, this is the issue with the -dbg package: http://paste.kde.org/pdajdhapy Aug 11 16:29:51 even though the format is the same like for the stripped. Aug 11 16:29:59 and if I copy the file over manually, it works like a charm. Aug 11 16:30:06 have not checked md5 for both to compare... Aug 11 16:30:15 but it seems Yocto gets it wrong in the package. Aug 11 16:30:21 or some rule in my recipe, etc. Aug 11 16:30:29 the package installation is definitely broken. Aug 11 16:35:58 nerdboy: seems the md5sums are different for the binary built and what is available in the debug package even though files say the same about both Aug 11 16:36:05 what is Yocto doing with the packged binary? Aug 11 16:36:10 packaged* Aug 11 16:36:32 I would expect the binary having the same md5 like the one in the package after decompression. Aug 11 16:37:24 one is 128 K (original), the other is 72K (package) Aug 11 16:38:08 the default is split packages where the build setup strips, etc Aug 11 16:38:42 is your recipe/makefile stripping binaries or something? Aug 11 16:39:24 and which yocto package? what image is it part of? Aug 11 16:39:36 it is part of our proprietary stuff. Aug 11 16:39:47 I am comparing the binary after the buildsystem run. Aug 11 16:39:59 your own target image? Aug 11 16:40:02 so it probably does not matter what my system does, but it does not strip anything. Aug 11 16:40:05 yeah Aug 11 16:40:09 well, it is just a package, but yeah. Aug 11 16:40:50 both binaries are unstripped and for the same arch Aug 11 16:41:06 the only different is that one is the end result of the build system and the other is extracted from the package generated by Yocto Aug 11 16:41:22 huge size difference and the one generated by Yocto has format/exec error. Aug 11 16:41:57 have you done do_populate_sdk for your custom image? Aug 11 16:42:22 nope, just bitbake myrecipe Aug 11 16:42:28 and have you built the adt installer with your own stuff? Aug 11 16:42:28 the rest is core-image-minimal for now. Aug 11 16:42:44 I am not using sdk for platform development. Aug 11 16:42:50 well, I do, inside Yocto itself. Aug 11 16:43:23 do_populate_sdk is specific to your own image recipe Aug 11 16:44:11 anyway, my guess is building a recipe is probably not enough Aug 11 16:44:50 Can I get more information what Yocto does with my binary after make? Aug 11 16:45:04 I am sure it does something with it, but I do not yet get what. Aug 11 16:45:06 i would probably build a full image and the adt installer/sdk and go from there Aug 11 16:45:28 the reference manual and other docs is all i got Aug 11 16:45:58 the latter targets are there to pull in the proper deps and whatnot Aug 11 16:46:16 we are doing install -m 4755 Aug 11 16:46:44 that doesn't seem like the issue Aug 11 16:47:58 well, the file in the package is significantly smaller than its origin. :D Aug 11 16:48:07 something must do something with it. Aug 11 16:48:24 yeah, but it's a little deeper than what the install command does Aug 11 16:49:08 yeah, I agree. I am not sure... I guess log.do_package is my friend. Aug 11 16:49:48 just humor me and build your image (including all package recipes), the sdk, and the adt_installer package Aug 11 16:50:27 the second one would be "bitbake -c do_populate_sdk Aug 11 16:50:43 after bitbake Aug 11 16:50:57 this package/recipe is not in the image since it is debug stuff. Aug 11 16:51:17 maybe shlib removes some needless stuff? Aug 11 16:51:19 what exactly is "debug stuff"? Aug 11 16:51:34 -dbg Aug 11 16:52:06 you don't let the build system create -dbg packages? Aug 11 16:52:14 * nerdboy is confused Aug 11 16:52:38 local.conf tells it where/how to build/install "debug stuff" Aug 11 16:53:11 I do, but it is not included in the image. Aug 11 16:53:47 your recipe should build something sane, either a cross(target) package, a native(sysroot) package, or nativesdk package Aug 11 16:54:10 * nerdboy needs another espresso Aug 11 16:54:32 let the system work for you... Aug 11 16:54:44 -dbg is just for manual hacking: bitbake foo; scp ../../../foo-dbg ...; ssh; etc. Aug 11 16:55:03 it is not supposed to be part of the production flow. Aug 11 16:57:25 is it a test version of something normally installed, or purely a test/developer tool? Aug 11 16:58:23 and -dbg has a specific meaning in the package tree Aug 11 16:58:51 if it's just your binary with symbols that's a bit confusing Aug 11 16:59:36 it is a development version of the normally released product. Aug 11 16:59:59 I am trying to improve by fixing bugs on the target, but currently I am even struggling with getting the debugging properly done. Aug 11 17:00:12 without proper debugging environment, it is impossible to improve the software on the target :) Aug 11 17:00:20 that's what those config settings are for Aug 11 17:00:44 test version would just be a different/modified recipe Aug 11 17:01:01 you could use preferred_version or a git recipe for testing Aug 11 17:01:11 that is what I am already doing! :) Aug 11 17:01:19 and trying to use foo-git-dbg Aug 11 17:01:32 then using the local.conf options and sdk targets should get you what you want Aug 11 17:01:49 but it is platform development, not application. Aug 11 17:02:08 a -dbg should just work without SDK, too, yeah? Aug 11 17:02:13 the tools are there, i would use them Aug 11 17:02:34 well, cgdb is not there... I am trying to make it being there, but that is the second issue here. Aug 11 17:02:39 except those targets and config options invoke the right build/install deps Aug 11 17:02:47 raw gdb is there, but that is awfully painful IMHO Aug 11 17:03:13 exactly my point Aug 11 17:03:14 it is very hard to step through a program with gdb; you have to list the code all the time, and it is black and white, etc... it is horrid. Aug 11 17:03:35 cgdb was meant to fix that, but it is not there... I am trying to be the innovator here to be there! :) Aug 11 17:03:49 the adt supports both gdb and the eclipse target framework Aug 11 17:03:58 cgdb is none of those. Aug 11 17:04:16 i would still innovate on top of the right setup Aug 11 17:04:30 there's a lot of integration there Aug 11 17:04:30 Yocto is the right setup to write recipes. Aug 11 17:04:48 and the sdk/adt is the right setup to test/debug on the target Aug 11 17:05:08 no, because they do not provide cgdb, so that is not OK for me, personally. Aug 11 17:05:20 once I have cgdb, life will be soooo much simpler. Aug 11 17:06:16 (and of course, cgdb will be part of our SDK offering, too, afterwards) Aug 11 17:07:09 you add your cgdb to the proper setup, otherwise this happens... Aug 11 17:08:19 Yocto should be OK to add a recipe without external SDK, I believe. Aug 11 17:09:55 well, the normal build with dbg-tweaks enabled will build/split the packages Aug 11 17:10:21 everything is built, the problem is not there :) Aug 11 17:10:21 beyond that, you need a bit more to get the tools, etc Aug 11 17:11:25 just try what i said, humor me, then come back later and tell me why it's wrong... Aug 11 17:15:22 If I want to discuss/send a patch to add a new variable on the Yocto core, where should I do it? Aug 11 17:17:32 Anarky: http://lists.openembedded.org/mailman/listinfo/openembedded-core Aug 11 17:17:47 Anarky: the README file has a small section on where to send patches Aug 11 17:23:31 lpapp, bluelightning: Thanks. And I'm wondering, when a new version of Yocto is released, they synchronize the core with Openembedded? Aug 11 17:24:13 Anarky: they are completely synchronised; changes go into BitBake / OE-Core first and then immediately appear in the poky repository Aug 11 17:25:09 bluelightning: so they don't "freeze" the core? Aug 11 17:28:26 ref man has info also Aug 11 17:28:48 Anarky: as far as I know, the release cycle is usually half a year and the last one took place in April. Aug 11 17:28:58 Anarky: that's for day-to-day development; there are release branches for each release and the branches + release tagging occur at the same time in both places Aug 11 17:29:51 master and master-next would be latest/bleeding edge Aug 11 17:30:13 ok, thanks :) Aug 11 17:30:39 clone poky and look at the branches Aug 11 17:31:51 nerdboy: ps aux seems to be my frined Aug 11 17:31:53 friend Aug 11 17:32:34 gdb --nw --annotate=2 -x /home/lpapp/.tgdb/a2_gdb_init -q ... Aug 11 17:32:44 it seems cgdb is running the _system-wide_ gdb. Aug 11 17:32:53 no wonder that cannot really work with gdbserver Aug 11 17:33:24 although I am surprised that it can even connect to the server to be honest. Aug 11 17:33:51 oops Aug 11 17:34:40 * nerdboy wanders off to file an upstream rpi issue Aug 11 17:35:24 rpi is still alive? Aug 11 17:35:34 you betcha Aug 11 17:35:45 new model B+ just came out Aug 11 17:36:01 * nerdboy looks for his lake trout Aug 11 17:36:23 you're about => <= close... Aug 11 17:36:43 yes, exactly, I thought everyone would use the new stuff. Aug 11 17:45:53 yeah, but still no RTC Aug 11 17:46:26 heh, RTC is difficult Aug 11 17:46:32 but at least they produce much quicker as the beaglebone team Aug 11 17:46:52 lpapp: is it? really? $5 and you can add it to the rpi GPIO Aug 11 17:47:24 volker: RTC is relatively expensive which is more appearing on a cheap board like rpi. Aug 11 17:47:49 beaglebone is a little bit more but with RTC and MMC on board Aug 11 17:48:04 yes, but rpi was meant for poor people. Aug 11 17:48:12 with only the very basic needs. Aug 11 17:48:21 it is poor men's gadget IMHO :) Aug 11 17:48:27 I think the rpi is a little bit too hyped compared to the beaglebone (which has availability issues) Aug 11 17:48:35 bbb has no analog audio/video Aug 11 17:48:46 true Aug 11 17:48:47 different requirements Aug 11 17:49:17 I had a lot of issues with the rpi. power, sdcard, camera module not working on the A model... Aug 11 17:49:23 rpi has some limitations but still good for $35 US Aug 11 17:49:44 never tried camera, and usb is well, rpi usb Aug 11 17:49:53 it is what it is Aug 11 17:50:12 sdcard works great, with minimal optimazation Aug 11 17:50:16 yeah, $35 is a good price. nothing against it. I still is neat. But comparing it to the beaglebone doesn't show too much benefits (if any at all, depends on requirements) :) Aug 11 17:50:36 been running the same card in my car since last year Aug 11 17:50:39 nerdboy: sdcard crashed a lot here, killed the FS so frequent that I was shocked. Aug 11 17:50:45 not here Aug 11 17:50:58 but i never tried yocto RO-rootfs on it Aug 11 17:51:01 you have some issue i haven't seen Aug 11 17:51:10 i run gentoo on it too Aug 11 17:51:27 i just don't usually *build* gentoo on it... Aug 11 17:51:29 ok, I ran the weezy version on it, but didn't play with it for a year Aug 11 17:51:36 just smallish stuff Aug 11 17:52:04 lsat year I added a USB camera to it and put it in the car. CPU was slightly to slow for live encoding (the GPU API is...), but was impressed by the result Aug 11 17:52:14 some raspbian and raspbmc, but mostly just yocto/gentoo Aug 11 17:52:37 yocto in the car, the other one is for testing Aug 11 17:52:54 nerdboy: what do you do with yocto in the car? Did you hook it up to the OBD? Aug 11 17:52:56 * nerdboy about to test rpi-sized TFT display Aug 11 17:53:41 just remote-controlled audio so far, since i have no video wiring harness for rear-view camera (in car) Aug 11 17:53:50 stupid toyota... Aug 11 17:54:14 nerdboy: yay, -d is my friend for cgdb, so I have a working cgdb, awesome. Aug 11 17:54:34 so the only remaining issue is why -dbg does not contain a proper binary.... Aug 11 17:54:50 volker: mpd audio build is here: https://github.com/sarnold/meta-alt-desktop-extras/wiki Aug 11 17:56:06 it has x desktop but i never see it in the car... Aug 11 17:56:36 ic. How do you control it then without a screen? Aug 11 17:56:41 hooked to your wheel controlers? Aug 11 17:56:46 lirc and IR remote Aug 11 17:56:59 ic. Aug 11 17:57:04 i added the $1.99 IR sensor Aug 11 17:57:24 I saw something similar on hackaday but without an IR Aug 11 17:57:42 that is an interesting issue with gdb... http://paste.kde.org/p6cqwh3e4 May that be because I have not installed dbg libraries underneath, only for my software? Aug 11 17:58:29 I thought it would be possible to debug my software with symbols, as I am sure the issue is not in the underlying software stack? Aug 11 17:58:45 http://www.gentoogeek.org/image_archive/home_hardware/rpi-dashboard.jpg/view Aug 11 17:58:49 does it mean that I have to install the debug version of each dependency? Aug 11 17:59:29 now it's velcro'ed on the glove box next to the power Aug 11 17:59:47 nerdboy: quite fancy webdesign with firework all around and all that Aug 11 18:00:27 eh? Aug 11 18:01:25 nerdboy: gg.org :-) Aug 11 18:01:40 volker: the IR sensor thing was one of the older projects on rpi.org Aug 11 18:02:36 lpapp: i will add some 20th century glamour to it when i migrate to openstack infrstaructure Aug 11 18:03:14 the fancier stuff was a long time ago and for specific courses and whatnot Aug 11 18:03:44 * nerdboy made students use squishdot for homework discussion Aug 11 18:04:25 nerdboy: noooooooo! Leave it like that; do not sell your soul to the evil. Aug 11 18:04:33 must be too early to type stuff like "infrastructure" Aug 11 18:04:56 I do not want more lynx incompatible pages :) Aug 11 18:05:29 latest one I instigated/deployed is here: http://www.vctlabs.com/ Aug 11 18:05:40 lets try it with links... Aug 11 18:06:21 meh, not bad i guess Aug 11 18:06:51 * nerdboy is suitably impressed Aug 11 18:07:02 does that work for ya? Aug 11 18:11:41 and yes, it *is* cool working for someone who pushes kernel patches... Aug 11 18:12:19 :) Aug 11 18:15:18 I do not know, but it is very hard to contribute to the Linux kernel IMHO, at least in a small company when you do not have time to address nitpicks. Aug 11 18:15:38 * nerdboy recruited and mentored her as gentoo dev many years ago... Aug 11 18:16:15 she knows a guy who knows ingo, but it gets done... Aug 11 18:16:26 Ingo is also hungarian... :-) Aug 11 18:17:24 i'd visit him there... Aug 11 18:17:56 latest article is her patch Aug 11 18:18:09 who is "her"? Aug 11 18:18:11 some geekette? Aug 11 18:18:12 *latest patch Aug 11 18:18:23 my boss, president of vct Aug 11 18:18:40 http://www.vctlabs.com/posts/2014/Jul/15/liblockdep_fix/ Aug 11 18:19:46 kindof a dream job, just without enough current customers with money... Aug 11 18:20:54 so we fill time with open source stuff and business development... Aug 11 18:22:06 * nerdboy not a business guy, but i know how to extract customer requirements Aug 11 18:23:39 seen sailfish yet? Aug 11 18:23:50 I am lacto-vegetarian Aug 11 18:23:58 * nerdboy installed demo on nexus phone Aug 11 18:24:06 pretty sweet Aug 11 18:24:27 have to go to europe to buy a jolla phone... Aug 11 18:24:53 lpapp: http://www.vctlabs.com/posts/2014/Jun/30/sailfish_prerelease/ Aug 11 18:25:40 nerdboy: I wanted a Nokia X2, but they killed it again :( Aug 11 18:25:58 heh, i know someone who is lacto-starch-vegan or some such Aug 11 18:26:09 very limited Aug 11 18:27:04 relatively rare, like my PKD coworker from aerospace... Aug 11 18:27:34 who is on his second (kidney) transplant Aug 11 18:27:44 well, I have never understood SailFishOS to be honest Aug 11 18:27:51 and the whole mer thing... why don't they just use Yocto? Aug 11 18:28:05 they seem to be building something similar over there with OBS. Aug 11 18:28:15 it's just a different/better(?) interface Aug 11 18:28:37 well, rpm is mandatory for one AFAIK. Aug 11 18:28:39 not familiar with building it yet... Aug 11 18:29:07 but i like the whole swipe-based thing in that release Aug 11 18:29:27 already use cyanogen on both phones anyway... Aug 11 18:30:09 no reason to be stuck with iOS interface (the most annoying one for me) or stock android... Aug 11 18:30:39 * nerdboy 's favorite is probably webos, then sailfish, then cyanogen11, then... Aug 11 18:31:34 as soon as upstream (samsung) 4.x started taking over my phone i had to nuke it Aug 11 18:32:23 google has a lot more work to do, especially in selinux implementation and app permissions/privacy... Aug 11 18:32:46 * nerdboy looks for his rant box Aug 11 18:32:52 don't get me started... Aug 11 18:33:48 at least samsung is one release ahead of google in enabling selinux, but otherwise it sucks balls... Aug 11 18:34:04 samsung/sprint Aug 11 18:34:19 i imagine verizon is even worse... Aug 11 18:58:46 lpapp: how horrible is this? http://www.meetup.com/Central-Coast-Open-Source-Solutions-Exchange/ Aug 11 18:59:08 banner graphic and theme are new... Aug 11 19:38:57 nerdboy: erm, I figured out the yocto package generation issue Aug 11 19:40:28 and? Aug 11 19:41:27 nerdboy: I do not know how it is even possible, but Yocto enchant some dependencies, even the dynamic loader, so no wonder it is format error Aug 11 19:41:38 but _why_ Yocto does that remains to be seen tomorrow. Aug 11 19:42:02 * nerdboy still confused Aug 11 19:42:31 that is alright. Aug 11 19:42:47 enchants* Aug 11 19:42:47 are you really saying it needed those targets built and/or local.conf settings? Aug 11 19:43:28 cough it up, dude... exactly what works and what doesn't... Aug 11 19:43:46 no, I am saying "LD_TRACE_LOADED_OBJECTS='1' /usr/bin/.debug/foo" lists 3 dependencies instead of the expected 6-7 Aug 11 19:43:55 even the ld dependency is missing from that output. Aug 11 19:44:08 pretty bizarre, I know, and have no clue why Yocto generates such a package for me. Aug 11 19:44:41 without ld, etc, no wonder if it is exec error... that is the very first thing needed. Aug 11 19:45:16 sounds like you should file a bug or at least post it to the mailing list Aug 11 19:46:58 nerdboy: I will investigate tomorrow. Aug 11 19:56:45 * nerdboy kicks perl upgrade in the cajones Aug 11 19:57:08 you will bend to my will, oh yes you will... Aug 11 21:22:19 scheduled Aug 11 21:22:44 http://www.meetup.com/Central-Coast-Open-Source-Solutions-Exchange/events/200359452/ **** ENDING LOGGING AT Tue Aug 12 02:59:58 2014