**** BEGIN LOGGING AT Tue Jul 26 02:59:58 2016 Jul 26 06:04:38 hi Jul 26 06:11:06 hello Jul 26 06:59:38 morning Jul 26 07:24:20 is there anything that can be set in local.conf to say 'use protocol=http for all git:// src uri"? Jul 26 07:24:42 or git global config maybe Jul 26 07:31:22 CTtpollard: i did use this trick in my local.conf: PREMIRRORS_append = "\n git://github.com/.* git://github.com/PATH;protocol=https" Jul 26 07:34:08 lzo: yeh I was thinking about that! This is probably good enough as I only need to override for one git server Jul 26 07:38:42 good morning Jul 26 08:05:03 Top 'o the mornin' Jul 26 08:38:16 I've got a recipe that is failing to fetch, does bitbake expand .'x' in branch names to anything? Jul 26 08:46:01 i.e it's specifying branch '2.11.x' which doesn't explicitly exist in the upstream Jul 26 08:46:44 i'd be suprised if it did... Jul 26 08:47:00 there is no common, well-known way of doing that sort of expansion Jul 26 08:47:28 boucman_work: so it's more likely that the branch has been removed from the upstream in this case? Jul 26 09:02:37 is kernel configuration deployed somewhere on the target? Jul 26 09:07:34 lzo: zcat /proc/config.gz in most qemu targets (when compiled with with CONFIG_IKCONFIG_PROC ) Jul 26 09:09:55 sed /with with/kernel with/ Jul 26 09:11:19 thanks, because i'm actually having troubles to figure out why a touchscreen driver is no longer working in 4.4.13 as it was working in 4.1.26. basically from jethro to krogoth. Jul 26 09:17:43 CTtpollard: no idea.. try "git ls-remote" on the remote to check... Jul 26 09:42:27 hi everyone Jul 26 09:42:46 I would like to add intltool in my sdk Jul 26 09:43:18 because it seems to be unavailable in the sdk Jul 26 09:43:53 I tried to add : SDKIMAGE_FEATURES += "intltool-native" in my image. but there was no effects Jul 26 09:49:48 intltool is not a sdk feature i guess. you probably should add it to your meta-toolchain recipe or in a packagegroup Jul 26 09:53:20 features != package names Jul 26 09:53:32 also you don't want a native package in a sdk, you want a nativesdk package Jul 26 09:53:40 how are you generating the sdk? Jul 26 10:17:04 Hi there ! I'm currently working on an RZ/A1 board (Hachiko from Architech) and I'm trying to implement bluetooth (I want bluetoothctl) on it. I added DISTRO_FEATURES_append = "bluez5", and MACHINE_FEATURES ?= "bluetooth", It compiles without error but I can't execute it on my board. I tried to add opkg to the board and it worked with the same copilation chain. Do you have an idea on what to make ? Jul 26 10:17:49 did you install the packages in your image? Jul 26 10:17:56 Kakounet: you should check your environment first. bitbake recipe -e Jul 26 10:18:11 MACHINE_FEATURES is probably overriden somewhere else Jul 26 10:18:14 the distro supporting bluez5 and the machine supporting bluetooth just mean that you could have built bluez5 Jul 26 10:24:55 lzo: what do I have to look for inside bitbake tiny-image -e ? Perhaps it's not the same thing but I found a really usefull command to check if bluetooth package is present : bitbake -s | grep bluez Jul 26 10:25:02 And it says : architech@architech:~/architech_sdk/architech/hachiko-tiny/yocto/build/build$ bitbake -s | grep bluez bluez5 :5.7-r0 Jul 26 10:26:51 -s is just show versions Jul 26 10:26:58 that's nothing to do with what is inside your image Jul 26 10:27:08 take image recipe, use IMAGE_INSTALL_append to put bluez5 in it Jul 26 10:28:09 rburton: Ok, I add IMAGE_INSTALL_append in my local.conf and I tell you Jul 26 10:31:29 write your own image recipe, basically Jul 26 10:32:30 Ok, it compiles now, thanks for helping me :) Jul 26 10:32:55 I will tell you as soon as I can test it Jul 26 10:51:53 rburton: you said "did you install the packages in your image?", that is the same to add IMAGE_INSTALL_append in local.conf (I'm sorry but it's a bit hard to understand for me :) )? Jul 26 11:18:32 Does anybody know if webRTC should work on an imx6 with chromium? Jul 26 11:20:15 when i try the browser tab crashes, but i maybe i screwed something up with the drivers Jul 26 11:25:17 rburton, i'm generating it using "bitbake -c populate_sdk image" Jul 26 11:40:16 how is it possible to get intltool in the sdk... Jul 26 11:45:59 Kakounet: image.bbclass usually creates a .manifest file together with your ./deploy/images/$MACHINE/ dir, it contains all the packages that are in the image. Jul 26 11:46:56 Nice thing is that it actually creates this at the beginning of the do_rootfs step, so you know relatively soon if it worked :) Jul 26 11:56:19 aurele: hm, try SDK_RDEPENDS += nativesdk-intltool Jul 26 11:56:23 in your image recipe Jul 26 12:00:34 rburton, nativesdk-intltool doesn't exists... only intltool-native i'm trying with this one Jul 26 12:01:04 best add nativesdk to BBCLASSEXTEND then Jul 26 12:43:45 jubr: I looked for the package.manifest that is in the good file. I can't see bluez bluetooth or bleutoothctl :( Jul 26 13:04:34 Hi, I'm trying to update my application on my development setup. The problem is that I did update several packages, including some dependencies of my package. Is there a way for bitbake to show all built dependencies? Jul 26 13:05:17 Kakounet: `grep bluez5 ./tmp/deploy/images/$machine/$yourimage-$versionstuff.rootfs.manifest` should give you some results. Jul 26 13:06:11 For us, it's `bluez5 cortexa9hf-vfp-neon 5.28-r0`, I checked, bluetoothctl is in that pkg. Jul 26 13:10:27 Maybe something went wrong with your RDEPENDS in $yourimage.bb, we typically use packagegroups as much as possible. For testing you could use IMAGE_INSTALL += "bluez5" in your local.conf. (browser reset wiped my channel history, it might have been mentioned earlier) Jul 26 13:11:43 Hmm.. I remember seeing something like MACHINE_FEATURES ?= "bluez5" - are you sure this var is not already set at that time? Try += Jul 26 13:12:17 jubr: Strange, I don't have such a file (.rootfs.manifest)... I will test using IMAGE_INSTALL += "bluez5", and no it has not been mentionned above. Jul 26 13:13:20 For MACHINE_FEATURES, I know that it has not been set in the same file (local.conf) Jul 26 13:13:27 `bitbake -e $yourimage | grep --color -A10 MACHINE_FEATURES` will give you some nice info about that var's history Jul 26 13:14:40 Hmm. maybe that file is only generated if you enable http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#enabling-and-disabling-build-history Jul 26 13:16:07 no Jul 26 13:16:15 buildhistory is unrelated to -e Jul 26 13:16:40 and you might want to use IMAGE_INSTALL_append = " bluez5" instead of += Jul 26 13:18:28 rburton: Yes, they are not related, just two different ways to keep track of what's going on. Jul 26 13:18:47 rburton: do you know if .manifest files are always created? Jul 26 13:18:48 rbuton: It's set like that i, my local.conf :MACHINE_FEATURES ?= "bluetooth usbgadget usbhost" DISTRO_FEATURES ?= "bluetooth bluez5 usbgadget usbhost" IMAGE_INSTALL_append = " tcf-agent bluez5" Jul 26 13:19:46 well those machine features and distro features are likely not being used Jul 26 13:19:56 luckily they're generally on by default Jul 26 13:20:03 Ok, I can remove them :) Jul 26 13:20:18 With my fancy bitbake-e+grep you can quickly see if those vars keep their values as you would expect. Jul 26 13:20:47 i'd just bitbake -e foo | less and use the search in there instead of hoping that -A10 covers enough context Jul 26 13:21:29 True. I usually `| tee bitbake.$recipe.out` in the middle. I like the colorized output too much :) Jul 26 13:21:36 also grep ^VARIABLE= -B10 will have less false positives Jul 26 13:21:44 jubr: I've got a lot of lines containnig bluetooth, I'm trying to understand a bit what's going on Jul 26 13:23:56 rburton: ^VAR -B10: nice one, tnx Jul 26 13:24:31 Kakounet: that's generally the case with OE/Bitbake :) Jul 26 13:25:23 for the question "how to i install bluez on my image" the answer is "ensure bluez5 is in your IMAGE_INSTALL Jul 26 13:28:39 jubr: :) I'm discovering, and there is a lot to discover ! Jul 26 13:30:36 Been OE-ing for going on 6 years now, still learning new things every day. It has evolved quite a bit since then. Yocto has brought a lot more docs + QA/sanity-checking to the mix! Jul 26 13:30:51 rburton: so how do I ensure bluez5 is in my IMAGE_INSTALL ? With the grep jubr gave me (replacing MACHINE_FEATURES ?) Jul 26 13:31:14 one way if IMAGE_INSTALL_append = " bluez5" Jul 26 13:31:20 the other is to write your own image recipe Jul 26 13:32:36 rburton: yes, IMAGE_INSTALL_append is well set, and for the fact to write a receipe I'm not sure to be ready for that :D Jul 26 13:33:58 Kakounet: it's not that hard, look for core-image-full-cmdline.bb in poky for a nice example Jul 26 13:34:47 Ok, I have a look at it. Jul 26 13:40:53 I had the idea that perhaps the compilation is ok but my board is toot short on memory. Bluetoothctl won't be set available because it costs too much memory... Wrong idea ? Jul 26 13:47:42 Hi all. Where can i file a bug / search for bugs before filing one? :) Jul 26 13:47:53 rburton: I'm having trouble with the SVN fetcher Jul 26 13:49:38 bugzilla.yoctoproject.org Jul 26 13:49:45 cheers Jul 26 14:22:40 Kakounet: disk-memory or random-access-memory? Jul 26 14:25:10 RAM, I've got only 1,7M free Jul 26 14:25:52 bluez5 is 2.4M here. If you want to keep your image size down, then truning on buildhistory will help you a lot. Exhibit A: a file called `installed-package-sizes.txt`. Exhibit B: a file called `depends.dot` :) Jul 26 14:26:18 Yikes. What HW are you running on then? Jul 26 14:27:28 http://architechboards-hachiko-tiny.readthedocs.io/en/latest/unboxing.html Jul 26 14:28:15 It's a Hachiko-tiny from Architech, 10M RAM Jul 26 14:28:53 Yikes (again) Jul 26 14:29:12 linux on something so tiny is non-trivial Jul 26 14:30:18 heh. Jul 26 14:30:32 pretty sure my first linux box had less memory than that :) Jul 26 14:30:32 i presume you've already switched to uclibc or musl Jul 26 14:30:55 fledermaus: things were smaller then, and that likely didn't expect to have a dbus-controlled bluetooth stack on :) Jul 26 14:31:17 You are not very reassuring guys :) Jul 26 14:31:20 musl+busybox is pretty tiny Jul 26 14:31:53 it ran X and a copy of the company webserver and compiled things and whatnot. Jul 26 14:32:08 I would have considered 10 MiB a luxury :) Jul 26 14:32:24 of course we didn't have MiB in those days, just dual-mode MB. Jul 26 14:32:29 * fledermaus reminisces. Jul 26 14:32:30 fledermaus: best stop talking now because this turns into a yorkshiremen sketch Jul 26 14:32:38 erm, /because/before/ Jul 26 14:33:01 too late. Jul 26 14:33:48 haha Jul 26 14:33:49 i remember trying to figure out how to make a broken gcc work in minx environment on my atari st Jul 26 14:34:31 I think my first computer had 2GB ram..... Jul 26 14:34:39 pah, upstarts Jul 26 14:35:10 48 delicious K. Jul 26 14:35:20 not like those 16 K peasants :) Jul 26 14:35:57 or those hoity-toity 64 K rich kids. Jul 26 14:36:19 Couldn't expect that I would start such a discussion :D. Do you think if I can had 512Mb of RAM it could help ? I bought the RAM component, just have to solder it Jul 26 14:37:04 Kakounet: with half a gig you'd not have any worry about ram usage at all. depends what you're doing really. if this is just a toy, then throw the ram on and stop worrying :) Jul 26 14:39:16 * jubr thinks you should heat up your soldering iron Jul 26 14:39:25 It's not for a big volume (100 pieces), mainly to make an access-point-switch between CAN, LAN and bluetooth Jul 26 14:40:08 * Kakounet is plugging in his soldering iron Jul 26 14:40:12 yeah i would :) Jul 26 14:40:56 shame connman doesn't handle CAN as otherwise meta-eca is what you want Jul 26 14:45:08 64K, makes me think of Farbrausch - The Product, old-skool demo-awesomeness in 64K (https://en.wikipedia.org/wiki/Farbrausch) Jul 26 14:46:17 Oh cool, the've even released the source code for their demo tools in 2012, didn't know that! Jul 26 18:25:55 Hi, I'm fairly new to yocto and I'm attempting to write a recipe but running into a bit of trouble. In the do_fetch stage I want to download a tarball but in order to know which tarball to download I have to read a manifest file (also downloaded). I'm trying right now to define a do_fetch_prepend() function. Right now it successfully downloads the manifest, determines the tarball to download and its MD5 and then sets the SRC_URI, + MD5 Jul 26 18:26:09 In the do_fetch stage though it doesn't seem that anything is being downloaded. Jul 26 18:26:27 Is something like this possible or am I going about this the wrong way? Jul 26 18:28:13 dmiess: you seem to need a special fetcher Jul 26 18:28:29 dmiess: you may be able to write a do_fetch of your own Jul 26 18:28:55 or let the fetcher get the manifest and then you get the tarballs after applying the logic Jul 26 18:29:22 in another function which you insert between do_fetch and do_unpack Jul 26 18:30:04 if this a one of known artifact management system then may be a full fledged fetcher is what you might be better of writing Jul 26 18:30:21 So I would write a special fetcher in the same way as the _preprend except I would keep going and actually retrieve the tarball myself? Jul 26 18:31:00 in bitbake Jul 26 18:31:19 like git fetcher or wget fetcher Jul 26 18:31:26 depends if its one off Jul 26 18:31:36 then just do it in recipe via additional functions Jul 26 18:32:56 I don't quite understand. I would probably be using wget but wouldn't I just keep my do_fetch_prepend, rename it to do_fetch and then instead of setting the SRC_URI just invoke wget myself? Jul 26 18:39:26 I'm looking to include a header file in a module I'm modifying. I see it in "/tmp/work-shared/nitrogen6x/kernel-source/include/linux/" how do I reference this location in my c file? Jul 26 18:43:48 I'm looking to #include to allow me to change the alpha level on my hardware. I'm also oppend to suggestions if there is a another way to get this Jul 26 18:45:14 dmiess: do_fetch is a pythonic function to _prepend/_append wont work Jul 26 18:45:25 you have to insert a function Jul 26 18:46:11 nillerbrun: couple of things. Find out if the header you want is a User API header defined by kernel Jul 26 18:47:11 khem: So when you say I have to insert a function you're talking about adding an extra do_something phase after do_fetch? Jul 26 18:47:16 if it is then it should already be installed via linux-libc-headers package and your app can simply expect it to be in staging sysroot and include it via include Jul 26 18:47:37 dmiess: yes, there are examples in oe-core of this kind of usage Jul 26 18:48:51 nillerbrun: my hunch is that this is vendor kernel only thing Jul 26 18:50:53 khem: Alright, thanks for the help. I'll take a look for some examples and try to do it this way. Jul 26 18:52:07 nillerbrun: so you probably want to install the extra header via patching linux-libc-headers via bbappend Jul 26 18:52:40 nillerbrun: another option is to stage it into usr/include when you build the kernel Jul 26 18:52:52 and then depend on kernel for your app Jul 26 18:54:14 Thanks khem. Both sound good to me, in your opinion which method is 'cleaner'? first? Jul 26 19:42:44 If I'm using BB_SRCREV_POLICY = "cache", what is the right way to force bitbake to re-determine the SRCREV values? Jul 26 20:02:26 nillerbrun: both of them are equally ugly :(. I would rather like vendors to upstream such changed Jul 26 20:02:42 jmesmon: delete your cache/ folder Jul 26 20:03:13 and change BB_SRCREV_POLICY = "clear" Jul 26 20:03:24 khem: agreed, thanks again Jul 26 20:04:02 khem: so even if I want to immediately re-cache them, I need to change the policy to clear them? Or is that just if I want them to not be re-cached? Jul 26 20:08:36 no then you dont chage it to 'clear' Jul 26 20:08:48 its for second part Jul 26 20:32:33 khem: I'm trying to implement this extra stage after do_fetch but I'm running into some trouble. If I just have my manifest hosted on svn in the SRC_URI it downloads fine. Once I add my extra do_get_tarballs and set it to be called with do_fetch_append I am no longer able to fetch and I recieve nothing. Jul 26 20:38:41 dmiess: is your do_fetch_append written in python or shell? Jul 26 20:39:20 (as do_fetch is python and just throwing some shell there will understandably break) Jul 26 20:39:21 In python. It's just this one line: bb.build.exec_func('do_get_tarballs', d) Jul 26 20:39:30 i'd just make a new task, or postfunc Jul 26 20:39:49 do_fetch[postfuncs] += "do_get_tarballs" Jul 26 20:40:03 or, addtask get_tarballs after do_fetch before do_patch Jul 26 20:40:04 Thanks, I'll give that a try Jul 26 20:40:19 depends on what the function does and what makes more sense Jul 26 20:42:08 I tried the second suggestion you mentioned and that worked great. Thanks! Jul 26 21:02:29 Do postfuncs fail the task if they fail? Because I had the impression they do not, and that can be unexpected at times. Jul 26 21:46:34 neverpanic: i believe they'll fail it Jul 26 21:46:52 OK, then it's just me not using bb.fatal where I should Jul 26 21:47:26 * rburton checks Jul 26 21:47:54 try: Jul 26 21:47:54 for func in (prefuncs or '').split(): Jul 26 21:47:54 exec_func(func, localdata) Jul 26 21:47:54 exec_func(task, localdata) Jul 26 21:47:54 for func in (postfuncs or '').split(): Jul 26 21:47:55 exec_func(func, localdata) Jul 26 21:48:09 literally no difference in handling Jul 26 23:07:27 hmm this gettext update has regressed couple of packages on musl Jul 26 23:07:59 I would suggest that we start doing a world build instead of image build on AB Jul 26 23:08:05 for 1 target for musl **** ENDING LOGGING AT Wed Jul 27 02:59:58 2016