**** BEGIN LOGGING AT Thu May 19 02:59:59 2016 May 19 04:03:37 bluelightning: I am just wondering if same can happen to OE where we dont get increase in young devs May 19 04:04:43 that is something we need to keep in mind, but from my reading it seems more like LEDE has come about because of a split in the OpenWrt community May 19 04:05:36 bluelightning: yeah, thats true, but one reason of the split is they wanted to do different things but were not enough devs May 19 04:06:43 the "not enough devs" thing is more of a consequence though right? because certain things that should have been changed to theoretically attract new developers weren't able to be changed May 19 04:09:19 for our project, I always feel like we're moving things forward in a reasonable direction, but I hope people feel like they can speak up if that's not the case May 19 04:10:22 yeah, i don't think we have same sorts of blocks on the progress. there are certainly things we could do to further improve the onboarding process, ease of use, etc, and that's ongoing.. May 19 04:10:39 indeed May 19 04:11:05 I do worry sometimes that we place a pretty high barrier to new contributors in terms of all the things you need to learn and get right May 19 04:11:38 but then, I'm not sure how to lower that barrier without just making more work for ourselves May 19 04:15:00 some of it is tough to address because of the tradeoffs. e.g. a more opinionated design would make certain things cleaner and more obvious to a new dev, but we'd sacrifice much of the flexibility that has helped drive the ability of the project to fit into so many areas May 19 04:26:57 more and more folks are getting into rpi kind of world where debian just fits in May 19 04:27:16 so in general embedded Linux is dead :) May 19 04:32:40 I disagree, I think with IoT and automotive it's taking off more and more, and plenty of those won't want to use existing commercial general purpose distros May 19 04:32:43 * kergoth shrugs May 19 04:47:43 I do think our biggest "threat" should one want to consider in that way is traditional distros May 19 04:48:17 we have to round off our sharp edges and sell our advantages (which aren't simply limited to just being able to target embedded hardware) May 19 06:02:05 kergoth: IoT is about sensors May 19 06:02:28 linux doesnt have to play much role in that may be uclinux for some of it May 19 06:02:57 rest can run on debian/raspian/rpi sort of things May 19 06:03:06 or gateways May 19 07:14:38 Hi, can somebody help me with an error in core-image-minimal.bb, do_rootfs (yocto fido)? It seems like Python is unable to execute any hash function (e.g. "ERROR:root:code for hash sha1 was not found." from python2.7/hashlib.py:127). Any ideas what to do? May 19 07:20:36 ...and libcrypto.so.1.0.0 is empty (0 bytes) May 19 10:08:29 What does this mean: ERROR: Taskhash mismatch c71a41325cf8beb9df90176526709810 verses 97e0d4020c4b04904073c487c1314452 for foo-image.bb.do_rootfs (with latest Poky)? May 19 10:25:31 that your do_rootfs task is most likely using DATETIME somewhere May 19 10:25:43 which causes the taskhash to be changed every second May 19 10:33:32 JaMa: Hmm, ok. Any simple way to tell where? I am not too knowledgeable about the hashes. I have used bitbake-dumpsig some. Can it help here? May 19 10:40:19 is anybody else facing issues with failing packaging because of BUILDSPEC and external toolchain being e.g. release 2016.02? May 19 10:40:47 Saur: yes bitbake-diffsigs and bitbake -S is the only way to debug this May 19 10:45:18 Saur: I just saw that today too (though with do_populate_sdk_ext) May 19 10:47:15 Ok, I am still getting a lot of errors "ERROR:root:code for hash sha1 was not found" (also for different hash functions) during Index creation in build_rootfs (poky fido). Any idea how to debug/fix this? May 19 10:49:14 (do_rootfs, not build_rootfs) May 19 10:51:37 Saur: did you just update to master or did you have older master-next? May 19 10:52:14 Saur: what was the previous working version? I'm trying to figure out whether this was a recent change or not May 19 10:52:16 RP: I updated from 9094ea9f to 7707cf65 May 19 10:54:28 Saur: You mean 7707cf65 (May 9th) to 9094ea9f (may 16th) ? May 19 10:54:29 if it makes any difference, I am at 52b3cc90667d + the pseudo workaround patch cherry-picked May 19 10:56:15 RP: 7707cf65 is current master of Poky... May 19 10:56:47 Sorry, gotta run to lunch before they close. Back in short while... May 19 10:56:59 Saur: its not, 2970800a09eb440c73ba1ff is May 19 10:57:32 Saur: there were changes last night and this morning since then May 19 10:58:35 Saur: sorry, I'm misreading dates in these logs. Its authored dates, not commited dates. Now makes more sense May 19 10:59:08 but master has moved on. We can at least rule out the recent changes which did cause taskhash issues but I thought we'd fixed May 19 11:01:12 Saur: I'm going to guess that we changed something which really needed a cache invalidation. "rm tmp/cache" would likely fix those taskhash warnings. The good news is if you upgrade again there is an explict cache invalidation for another change May 19 11:17:19 RP: Bah, I keep forgetting that we have a local mirror of Poky that is only updated once a day. :P May 19 11:46:51 If I already built my Yocto Image and its running... is there a way to build a SDK for QT or is it also a huge process to go over? May 19 12:42:05 If I already built my Yocto Image and its running... is there a way to build a SDK for QT or is it also a huge process to go over? May 19 13:28:39 Hi folks, looking for a little help. I'm building a .so to use as a dynamic crypto engine for openssl. I cross compile it with gcc 4.9.2 (arm-poky-linux-gnueabi) using the following flags: -O1 -march=armv7-a -marm -mthumb-interwork -mfloat-abi=hard -mfpu=neon -mtune=cortex-a9 --sysroot=... -fPIC, everything builds and links fine, but when attempting to load, I get unexpected reloc type 0x03. May 19 13:28:59 After some investigating, I find that my code is compiled with the instruction R_ARM_REL32 which causes this, but to my knowledge -fpic is supposed to prevent this. I'm not able to find a way around this, would anyone have any suggestions? May 19 13:33:18 zeddii: Hello May 19 13:34:09 zeddii: berton and I are looking at the docker version issue, with opkg. It seems it is not working because the = operator in opkg is considering the package revision. May 19 13:47:16 hello May 19 13:50:01 If I already built my Yocto Image and its running... is there a way to build a SDK for QT or is it also a huge process to go over or could I use my existing bitbake environment to create it? May 19 13:52:59 Or a better question. ANYBODY know of example/tutorials that I can follow to build a Targeted compiler and QT IDE files for a NXP IMX6 board ? May 19 18:20:35 so I am using yocto to build a cross compiler, I want to build a new c app. Is there a good sample for adding a custom app to the build? I assume the layers and recipes I've appended already would be a good start but they are mostly just tweaking install of new files. Nothing for specifying compiler or source. May 19 18:23:16 davis have you looked at the available documentation? or the wiki? May 19 18:24:13 Such as http://www.yoctoproject.org/docs/2.1/dev-manual/dev-manual.html#new-recipe-writing-a-new-recipe May 19 18:28:19 dvhart: I seldom look at the official yocto docs. I find them obscure and not very helpful. For instance that section you provided, does not show the contents of a single file, nor show many commands and their results. May 19 18:30:43 davis: I'm sorry to hear that, we have generally received positive feedback on the documentation. You would like to see the entire file instead of snippets? Can you provide an example of the commands or the section/paragraph where a command is missing? May 19 18:30:59 davis: you may find meta-skeleton useful for recipe examples May 19 18:32:22 personally, i believe the arch linux wiki is a good example of a wiki. May 19 18:39:53 the official yocto docs aren't a wiki, so i don't really see how that's useful May 19 18:41:03 hmm. actually in your doc, in section 5.3.18.1 there is a section on a single .c file package. May 19 18:42:36 so if there is any thing I could suggest it would be that at the top level header 5.3, include a toc which shows the topics in this section. May 19 18:43:31 If I go to this section that you provided and I scroll down 10 pages and don't see anything relative to what I'm trying to do, I seldom keep looking. May 19 20:11:37 Hi, I'm having issues with __aeabi_uidiv when building a .so, I have -lgcc in my link command, but when loading the .so I still get unresolved errors, is there some kind of a workaround for this? May 19 20:14:49 you need -lgcc_s May 19 20:15:23 and order of -lgcc also matters on linker cmdline May 19 20:15:33 specify it last for best results May 19 20:15:44 much appreciated, I'll try it now and see May 19 20:25:41 Thank you khem, that worked! is the use of libgcc_s documented anywhere? May 19 20:40:29 look into gcc docs **** ENDING LOGGING AT Fri May 20 02:59:58 2016