**** BEGIN LOGGING AT Fri Feb 10 03:00:02 2017 Feb 10 08:16:48 good morning Feb 10 08:18:05 Hi, how to get image in format for qemu ( for example raw in .img) instead of rootfs.ext3 or howto convert it? Feb 10 08:23:10 You can convert using qemu-img Feb 10 09:09:36 have I misunderstood something about linking order: how is "undefined reference to 'floor'" possible if the last thing in linker command is "-lm"? Feb 10 09:09:55 (this is http://errors.yoctoproject.org/Errors/Details/130587/ ) Feb 10 09:12:20 is there somewhere a method like in fs-uuid.bbclass for the repplacement of <> but replace it with the PARTUUID so it works without initrd as well? Feb 10 09:55:16 Hi, i'm getting "fatal: unable to access '': gnutls_handshake() failed Error in the pull function." Feb 10 09:56:06 I previously entered an invalid pw for the keyring to access a repo since i wanted to abort an accidental push. After fixing and pushing it again i re-entered my pw correctly this time. Might that be causing such issue? Feb 10 09:56:11 Any ideas how to fix it? Feb 10 10:17:06 jku: seems odd :/ Feb 10 10:53:33 * rburton1 looks at mut results and screams Feb 10 10:55:02 aah Feb 10 10:55:03 checking $CC variable... broken Feb 10 10:55:03 configure: WARNING: your environment misuses the CC variable to hold CFLAGS/CPPFLAGS options Feb 10 10:55:07 thanks, ncurses Feb 10 10:59:55 trimmed and re-fired a mut, should be quick this time Feb 10 11:31:27 hm, is there any "what's new in yocto 2.2" summary anywhere? and upgrade instructions? Feb 10 11:31:43 the release is not even mentioned on the blog. Feb 10 11:34:14 ionte: literally one click from yoctoproject.org https://www.yoctoproject.org/downloads/core/morty22 Feb 10 11:34:53 thanks Feb 10 11:35:07 i was looking for something like https://wiki.yoctoproject.org/wiki/What's_New_In_2.1 Feb 10 11:35:37 but that is better Feb 10 11:47:04 rburton1: only two issues left with dnf... so close! Feb 10 11:47:14 oh, did I say that yesterday already? Feb 10 13:48:25 rburton1: the dnf branch is ready for the autobuilder! Feb 10 13:48:34 rburton1: https://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=akanavin/dnf-rpm4 Feb 10 13:48:40 I expect nothing less than a total disaster Feb 10 13:48:49 RP: ^^^ Feb 10 13:51:54 ooooo Feb 10 13:52:33 rburton1: I will polish the patchset and write better commit messages on monday, then it will go on the oe-core list as well Feb 10 13:52:49 (regardless of the AB results - I'm so tired of the solo hackathon) Feb 10 13:53:05 https://autobuilder.yocto.io/builders/nightly-x86-64/builds/172 <— lets see what this does Feb 10 13:53:35 start with just x86-64 to check it passes that on the AB Feb 10 13:55:23 rburton1: will that be quick-ish? Feb 10 13:56:14 ish Feb 10 13:56:16 kanavin: 3-4 hours total I think Feb 10 13:57:47 rburton1: I did no checks with anything other than qemux86 (and a quick smoke test on aarch64 to see that rpm/dnf do not refuse to build an image for that) Feb 10 14:00:22 do we not have any AB builds that use "ld-is-gold"? Feb 10 14:01:35 no, we don't Feb 10 14:06:28 so many times I've looked at a failure in jamas build and scratched my head until finally "oh, lets try with gold" Feb 10 14:06:49 You'd think I'd learn but so far no Feb 10 14:23:44 kanavin: first image build worked \o/ Feb 10 14:27:32 rburton1: \0/ Feb 10 14:39:55 kanavin: nice to get to this stage, looking forward to seeing how it fares on the AB :) Feb 10 14:40:28 RP: yes, but it took a lot out of me :( Feb 10 14:40:35 RP: similar to you and rss Feb 10 14:40:41 I guess I need some quality sleep Feb 10 14:45:08 RP: if you can take a breath, what to do with klcc-cross crosscript? We tried to avoid sstage for it but now it is arch specific again. Wouldn't it be easier to inherit cross and treat it like a cross-compiler? Feb 10 14:51:46 * kanavin has fixed vnc access to the big build box, yay! Feb 10 14:51:55 was giving me a black screen before Feb 10 14:52:13 so no need to be restricted to ssh sessions Feb 10 14:52:58 if anyone's interested, it's specific to xfce, and here's the discussion https://bugs.launchpad.net/ubuntu/+source/light-locker/+bug/1287171 Feb 10 14:58:09 ant_work: I can't really give specific advice on it without trying to look at the code, setup a build and see what is going wrong with it, all of which does take a fair bit of time :( Feb 10 14:58:41 kanavin: Right, I know that problem. Sounds like you need a good break for the weekend (and some sleep)! Feb 10 15:01:02 RP: it's just the strange error puzzling me http://pastebin.com/zhmCtUK0 Feb 10 15:01:08 hope it is recipe specific Feb 10 15:01:10 | ERROR: configure failed Feb 10 15:01:10 | Unescaped left brace in regex is deprecated, passed through in regex; Feb 10 15:01:56 sounds like you used a bitbake variable which was undefined, and it got left unexpanded, but that's not looking at your code at all Feb 10 15:02:53 see, I only look at the last well-known one | configure: error: C compiler cannot create executables Feb 10 15:02:55 :) Feb 10 15:03:29 the poin tis, sysroot was once per arch, then for machine, now for recipe Feb 10 15:03:40 ant_work: I suspect that iust just a warning and not the real failure. The real failure looks like "| checking whether the C compiler works... no" Feb 10 15:03:44 these old recipes must be probably rewritten Feb 10 15:03:55 ant_work: and config.log would tell you more about why Feb 10 15:04:25 RP: I know 99% where the problem is, just wondering if instead the recipes are faulty Feb 10 15:04:44 some pointing to STAGING_DIR_HOST Feb 10 15:05:02 have to remember why Feb 10 15:14:22 RP: anyway, the fact that if you rebuild it it is then found by the building recipes can only mean it is installed right but wronlgly mangled for sysroot Feb 10 15:14:43 so sstate stages it incorrectly Feb 10 15:16:23 I'll verify that Feb 10 15:16:25 bbl Feb 10 15:16:27 thx Feb 10 15:27:49 kanavin: https://autobuilder.yocto.io/builders/nightly-x86-64/builds/172 Feb 10 15:29:03 hi, can someone explain me why qt4-embedded has a "E" at the end of all libs ? Is this yocto specific ? Feb 10 15:32:13 E for embedded? Feb 10 15:33:06 yes, that is clear but why? Feb 10 15:34:29 i have seen several versions of Qt/QtEmbedded since 3.X but i have never seen that "E" not on ptxdist, buildroot oder other systems... Feb 10 15:35:06 so, for me it looks a yocto specific thing, that "E" is appended, is there a way to remove ist ? Feb 10 15:35:17 s/ist/it/ Feb 10 15:35:29 hello Feb 10 15:35:33 grma: grepping meta-qt4 says the E is to mark the embedded version instead of the x11 version Feb 10 15:35:39 rburton1: and also tells you how to stop it Feb 10 15:35:45 QT_LIBINFIX Feb 10 15:36:05 ok, so i have to look how i overwrite variables in that bbclas, right ? Feb 10 15:36:18 in the recipe Feb 10 15:36:24 or, just ignore it Feb 10 15:36:38 I added a layer to my bb recipes and I want to rebuild my image. how do I let it rebuild without rebuilding the whole image Feb 10 15:36:46 with bitbake Feb 10 15:36:59 wesam: bitbake myimage, will only rebuild what needs to be rebuilt Feb 10 15:37:16 i added the layer and it didnt rebuild ant=ything Feb 10 15:38:55 then nothing changed Feb 10 15:38:56 thats because adding a layer just makes software available to you Feb 10 15:38:58 doesn't add it to the image Feb 10 15:39:05 because *what* image, and *what* packages Feb 10 15:41:28 ok so i am doing bitbake agl-demo-platform. I added another recipe to meta-webserver and when i just went to build it didnt rebuild anything Feb 10 15:41:56 rburton1: NOTE: recipe nativesdk-gnu-config-20150728+gitAUTOINC+b576fa87c1-r0: task do_package_write_deb: Failed :D Feb 10 15:42:44 wesam: add the packages you want to be installed to the image recipe Feb 10 15:42:44 I guess rewriting rpm creation breaks deb too Feb 10 15:43:05 will look into it next week Feb 10 15:43:15 wesam: you appear to be expecting "adding a layer -> packages in image" which for meta-webserver would mean you're image would have several web servers in Feb 10 15:43:21 obviously that's not what you want Feb 10 15:44:42 alright so how would typically add a recipe Feb 10 15:44:57 any good documentation i can look at for that Feb 10 15:46:54 wesam: the documentation is quite good: http://www.yoctoproject.org/docs/2.2/dev-manual/dev-manual.html#usingpoky-extend-customimage Feb 10 15:48:25 thank you Feb 10 15:52:56 rburton1: thx, so the correct way would be to make a recipes-qt4/recipes-qt4/qt4/qt4-embedded_%.bbappend file and override it there in my layer ? Feb 10 17:49:45 fray: do you know what rpm needs ossp-uuid for? Feb 10 17:49:56 you can turn it off, and its off by default upstream, but we turn it on Feb 10 17:51:10 At one point we had different uuid providing libraries.. The different libraries had slightly different behavior, so without the ossp-uuid we would get corruption and occasionally different output from different builds (without changes) Feb 10 17:51:26 The recommendation at the time from the RPM author was to mandate ossp-uuid Feb 10 17:52:09 so of course the default behaviour upstream is to have it off :) Feb 10 17:52:25 ossp-uuid is incredibly unmaintained Feb 10 17:52:30 like almost a decade since last release Feb 10 17:52:36 there is no real default behavior.. If you look in the devtool (community ones) those are the configurations that are tested. Feb 10 17:52:45 How often does it need to change? Feb 10 17:53:10 seriously, other then perhaps a minor bug, the generation of a uuid is a standards based item and shouldn't change often Feb 10 17:54:02 it's also possible that we no longer need it... Feb 10 18:08:09 yay, apparently we have an issue building morty toolchain on hardened Feb 10 18:09:35 switching from krogoth to morty wants to build 6.2 which is pickier about build config i guess... Feb 10 18:10:03 hasn't puked yet, so i might have a config fix Feb 10 18:13:56 as in adding "--enable-default-pie --enable-secureplt" to EXTRA_OECONF in gcc-cross.inc Feb 10 19:02:04 nope, not quite enough Feb 10 21:13:08 Hi, I'm wondering if this channel can give me some advice. I am porting an existing project that has several applications in the same git repo (with some shared headers, makefile boilerplate, etc). I'd like to write seperate recipes for each of them. Is there a way to share the git checkout (so the build instructions for each recipe would something like "cd to such-and-such directory in the shared git tree and Feb 10 21:13:10 run make"? Feb 10 21:17:59 sjburt: yes, but i doubt it'd buy you all that much. the clone will be shared regardless. there'd just be an individual local clone from DL_DIR for each recipe, and local git clones are either shared or hard linked by default Feb 10 21:27:52 Thanks! So this strategy in general is OK? (I realize that it's probably not ideal) Feb 10 22:36:04 halstead - hello Feb 10 22:38:04 dreyna_, The call just dropped. Feb 10 22:38:19 I restarted call Feb 10 22:39:01 I was having headset issues Feb 11 02:28:28 okay, i stopped trying to make it build a hardened toolchain (for now) but found a vapier patch that at least fixes the gcc-6.21 toolchain build Feb 11 02:28:37 *6.2 even **** ENDING LOGGING AT Sat Feb 11 03:00:01 2017