**** BEGIN LOGGING AT Fri Mar 20 02:59:59 2015 Mar 20 08:13:51 good morning Mar 20 08:31:19 is there some documentation regarding the signature that is generated for rpm packages? Mar 20 08:31:25 /cc bluelightning Mar 20 08:32:22 morning all Mar 20 08:32:26 jaeckel: signature? Mar 20 08:32:41 gpg signature of rpm packages Mar 20 08:34:10 because I can see that they're signed, but I have no clue with which key Mar 20 08:34:53 resp. where that key comes from Mar 20 08:41:16 hmm, that's news to me at least Mar 20 08:47:05 $ rpm --checksig tmp/deploy/rpm/armv7a_vfp_neon/bash-4.3-r0.0.armv7a_vfp_neon.rpm Mar 20 08:47:08 tmp/deploy/rpm/armv7a_vfp_neon/bash-4.3-r0.0.armv7a_vfp_neon.rpm: (sha1) dsa sha1 md5 OK Mar 20 08:47:44 'rpm -qip' gives "Signature : DSA/SHA1, Thu Mar 19 15:08:13 2015, Key ID b0b43a456bc4c8da" Mar 20 08:52:14 jaeckel: I can't find any code that would be adding that Mar 20 08:52:40 neither did I Mar 20 08:53:29 okay, I will have to check back with the guy who set up the yocto environment Mar 20 10:02:04 any yocto-autobuilder savvy folks in here? Mar 20 10:03:36 I have set SSTATE_PUBLISH_DIR in my autobuilder.conf but I'm not seeing sstate be published Mar 20 10:03:51 so I started looking at the code and that variable doesn't seem to be referenced anywhere in the source Mar 20 10:24:14 morning all. Mar 20 10:24:30 Hi everybody. I'm wondering what's the best way to work with protected repositories where the user needs credentials to access them. Mar 20 10:25:30 For example a git repository where the user needs a username and password. I don't want to include the credentials in the recipe as each user should use his own credentials to access the server Mar 20 10:25:55 Is there any best practice for this? Mar 20 10:26:49 im using the meta-intel layer and building as virtual/kernel the 3.14% linux-yocto-rt. Then I define a new layer, with a recipes-kernel/linux/linux-yocto-rt_3.14.bbappend rule and a local folder "files" with the .config fragments and the 0001-blabla.patch. Why i'm not applying my .bbappend patch and fragments? Where to see which patches are being applied? TIA Mar 20 10:32:12 bachp: use variables and have the user set them in their local.conf? Mar 20 10:33:06 I would like to avoid variables if possible, because who knows where they might end up (logs, ..) Mar 20 10:35:00 joshuagl: I was thinking about using something like "git config credential.helper" and make the user execute this before running bitbake and then just use the stored credential from the tool. There are similar commands for SVN and ClearCase Mar 20 11:07:43 bachp: use an ssh key and let ssh-agent do its magic Mar 20 11:09:45 my odroid bsp works fine for odroid-xu3, and i'm finalizing build for odroid-c1 Mar 20 11:11:57 there's only one problem i'm going to double check again: u-boot gets miscompiled for odroid-c1 Mar 20 11:12:59 it boots fine thought when the same u-boot source is built by ubuntu saucy's gcc-arm-linux-gnueabihf Mar 20 11:13:49 in any case, u-boot doesn't build without http://pastie.org/10041625 Mar 20 11:18:50 you can change that in your recipe instead of patching the .inc Mar 20 11:32:02 abelloni: thanks Mar 20 11:32:57 rburton: RP: (+others:) what do you think about the systemd minimum kernel version issue? is Net147's patch likely to be accepted for 1.8? Mar 20 11:33:35 hm i thought i picked that Mar 20 11:34:16 and now i can't find it Mar 20 11:34:37 ah there it is Mar 20 11:35:29 can the userspace stuff check the kernel version to determine if it needs to do anything? Mar 20 11:39:33 rburton: it depends on the kernel config Mar 20 11:40:31 bluelightning: good question. We could perhaps do this for 1.8 but I don't think its going to be sustainable longer term Mar 20 11:40:32 rburton: if CONFIG_FW_LOADER_USER_HELPER_FALLBACK=y, kernel will fallback to udev to load firmware if in-kernel loading fails Mar 20 11:41:15 RP: right, yes Mar 20 11:42:11 if it is not sustainable for newer systemd versions, could we have multiple systemd versions available for compatibility? Mar 20 11:42:56 Net147: the big question is where we draw the line for older versions Mar 20 11:52:21 RP: as long as someone wants/needs it and has time to maintain it? Mar 20 11:58:47 Net147: no, I really don't like this idea as it means we have to limit what we add to be compatible and the different maintainerships and priorities tend to conflict after a while :(. This is the kind of problem that tore OE-classic apart :/ Mar 20 12:00:22 bachp: the credential helper is also the solution I would've proposed Mar 20 12:00:55 Good to hear. Do you think it would make sense to add sections to the bitbake manual that would describe the recommended precedure for each fetcher? Mar 20 12:01:56 yeah, sounds good Mar 20 12:02:54 RP: if it gets to that point then the compatibility can be moved to and maintained in a separate layer outside of the scope of Yocto Mar 20 12:02:57 also I couldn't find a section about the url schema, dependency on the master branch and SRC_URI Mar 20 12:03:21 Net147: I think that may be the better way to do this to be honest Mar 20 12:05:08 RP: I was just surprised the the kernel version requirement changed so late in the release schedule that many people haven't noticed Mar 20 12:07:33 Net147: for 1.8 I agree and I may take the patch for that reason Mar 20 12:08:05 Net147: but it would be as an exception, not a rule and there wouldn't be a good reason to keep during 1.9 for example Mar 20 12:08:27 RP: I agree Mar 20 12:35:23 How can I view the sato screen over ethernet on my laptop? I've tried vnc, but it doesnot seem to work. All i want is to get the video output from a webcam connected to a pandaboaed es running yocto. Mar 20 12:40:09 sorry for the connection loss, has anyone streamed video from a webcam over ethernet? Mar 20 16:47:45 Happy Friday everyone! Quickie - does "FILES_{$PN} = {$bindir}/foo/" include all the files in the directory? Mar 20 16:48:07 Or do I have to be explicit and add a star at the end of the path? Mar 20 16:51:34 yes, a directory in FILES will be pulled in recursively with its contents Mar 20 16:53:13 kergoth: no need for /*? Mar 20 16:53:19 correct Mar 20 16:53:30 also, /* would miss any hidden files in that directory Mar 20 16:54:26 kergoth: hmm I see. There seems to be a bug in the php recipe. Doing populate_sdk fails when I include php-cgi in my image. Mar 20 16:54:49 kergoth: The error being, no package provide /usr/bin/php file. Mar 20 16:55:29 kergoth: I had to add the line FILES_{$PN} = ${bindir}/php Mar 20 16:55:40 kergoth: original line said FIELS_${PN} = ${bindir}/ Mar 20 16:56:27 as i said, including bindir will include everything under bindir, so changing that isn't going to do anything useful Mar 20 16:57:12 kergoth: ok thanks Mar 20 18:13:57 is there a way to configure kernel modules without using this command bitbake virtual/kernel -c menuconfig ? Mar 20 18:14:30 i have tried according with the documentation MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS but it didn't work :( Mar 20 18:30:07 Has anyone tried building Ceph through yocto/bitbake? Mar 20 18:32:19 I found a recipe for a python client API but not much else Mar 20 20:35:43 Should I ever see references to sysroot within the files from the image directory? Mar 20 20:36:17 I see a bunch in the php package Mar 20 20:36:52 * nerdboy not surprised it's php... Mar 20 20:36:57 me neither Mar 20 20:37:00 and no, you should not Mar 20 20:37:11 LOL Mar 20 20:37:19 I see a bunch and it's giving QA warning Mar 20 20:37:31 It's also failing populate_sdk Mar 20 21:00:49 hey guys...goofy issue...I have a -dev rpm package that one of my recipes produces which is dependent on the base package, but there is no base package produced because it's a staic link lib, so headers go in -dev and the lib goes in -staticdev Mar 20 21:01:34 not sure how to resolve this...should I just stick a dummy file in FILES_${PN} so that the base package is created? Mar 20 21:01:45 RDEPENDS_${PN}-dev_remove="${PN}" Mar 20 21:01:57 (assuming you're using a relatively recent release) Mar 20 21:02:05 that will remove the dependency on PN from PN-dev Mar 20 21:02:14 excellent...thank you much Mar 20 21:02:22 if you want to force a package to exist although its empty, use ALLOW_EMPTY_${PN}="1" Mar 20 21:02:44 that's the traditional solution to this problem but why bother creating empty packages when its just going to confuse everyone Mar 20 21:02:51 indeed Mar 20 21:25:20 Guys, any ideas on how to put yocto to automount pendrives or other usb devices? Mar 20 21:37:00 realBigfoot: udev rules should handle that Mar 20 21:37:26 nerdboy, i heard that you can manage to do this with systemd, do you know if it is true? Mar 20 21:37:58 systemd assimilated udev, so my guess is "probably/maybe/idk" Mar 20 21:38:14 sysvinit and udev works Mar 20 21:38:35 nerdboy, alright :) thanks Mar 20 21:39:27 * nerdboy should have fresh qt5 bbb image shortly Mar 20 21:40:33 * nerdboy building on laptop "build machine" while other machine is updating... Mar 20 21:42:28 yay, down to less than 400 packages... **** ENDING LOGGING AT Sat Mar 21 02:59:59 2015