**** BEGIN LOGGING AT Wed Feb 10 02:59:59 2016 Feb 10 05:32:21 sno: yes so I think it should be ok once 2.23 hits master Feb 10 09:03:04 good morning Feb 10 09:08:22 hi, did you built current kernel for rpi ? Feb 10 12:32:22 I added met-{oe,python,networking} form openembeded layer to yocto for install squid. In build process of poky-tiny util-linux fail to compile. Did yokto 2.0 compatible with this layer|can any body guide me to fix error Feb 10 12:37:35 -j #technotux Feb 10 13:26:23 thanks rburton Feb 10 13:49:06 I added met-{oe,python,networking} form openembeded layer to yocto for install squid. In build process of poky-tiny util-linux fail to compile. Did yokto 2.0 compatible with this layer|can any body guide me to fix error Feb 10 13:52:16 did you checkout jethro branch of it? Feb 10 13:58:29 JaMa, no i downloaded from layers.openembeded.org Feb 10 13:58:51 behrad: checkout the jethro branch :) Feb 10 13:59:53 CTtpollard, Thanks :) im newbie in yocto sorry :( Feb 10 14:00:14 behrad: aren't we all at some stage? :) Feb 10 14:01:51 CTtpollard, yes thanks Feb 10 14:05:34 Cheers! When aiming for (m) multiple architectures in (having distinct board support layers) and (n) images, what is the suggested amount of build environments for that? Guesses from my side are m*n,m,n,1 :-) Hints highly appreciated! Feb 10 14:06:10 1 Feb 10 14:07:45 I'd have an env for each arch, but not for each image Feb 10 14:10:02 Thnx! I'll be a little more specific: is it suggested/no problem to add e.g. meta-raspberrypi and and another architecture/board layer to bblayers.conf? This would make the difference for me between the JaMa and CTpollard answers. Feb 10 14:10:10 no problem Feb 10 14:10:30 any problems you have are bugs in the bsp Feb 10 14:11:01 (bsps should ensure that they don't modify anything unless its for the machines they target) Feb 10 14:11:08 okay, good to hear that, so I'll go that path and can consider problems due to this good questions/exercises :-) Feb 10 14:11:45 eg i have a single build directory and flip it between one of four architectures and distro configurations several times a day Feb 10 14:12:37 okay, that's similar to what I am aiming for! Thnx! Feb 10 14:12:48 I wish we had resources to test bsp's for good beavior Feb 10 14:12:54 behavior Feb 10 14:13:23 several distro configurations in the same build directory aren't officially supported, but should work in most cases Feb 10 14:13:44 but using the same bblayers.conf with multiple BSP should work fine with all "good" BSPs Feb 10 14:14:57 and building multiple images is best in the same bitbake call so that it can run more tasks in parallel Feb 10 14:15:52 JaMa: okay, good to know some background! Thnx Feb 10 14:19:09 can I create an spdx manifest for an image by simply put SPDX_MANIFEST_DIR ?= "/yocto/spdx/fossology/xxx“ Feb 10 14:19:10 INHERIT += "spdx“ in local conf? Feb 10 14:35:45 I'm hoping to generate a spdx I can feed to fossology Feb 10 14:55:05 CTtpollard, with jethro branch i have same issue Feb 10 14:55:33 utils-linux and perl failed Feb 10 14:56:04 i used this: git clone -b jethro --single-branch git://git.openembedded.org/meta-openembedded Feb 10 15:35:25 hmmm bitbake can't parse INHERIT += "spdx" when I add it to local.conf Feb 10 15:36:31 RROR: Unable to parse conf/bitbake.conf: ParseError at /home/ubuntu/builds/qemuspdx/gdp-src-build/conf/local.conf:12: unparsed line: 'INHERIT += "spdx“ Feb 10 15:39:33 the spdz.bbclass is in my poky checkout Feb 10 15:39:35 *spdx Feb 10 15:40:46 gah fixed it Feb 10 15:41:04 hi, i am struggling in building valgrind-nativesdk, anyone ? http://pastebin.com/4mgLU081 Feb 10 15:46:24 although it's ignoring my manifest dir, and trying to create it in /home/yocto ...... Feb 10 16:06:53 mastier_: target is nativesdk- Feb 10 16:07:00 you have it otherway around Feb 10 16:07:09 bitbake nativesdk-valgrind Feb 10 16:07:14 would be the one you want Feb 10 16:43:10 khem: thanks! that works Feb 10 16:43:52 Hi guys. Reading the docs I get a bit confused: Feb 10 16:44:20 PACKAGE_GROUP The PACKAGE_GROUP variable has been renamed to FEATURE_PACKAGES. See the variable description for FEATURE_PACKAGES for information. If if you use the PACKAGE_GROUP variable, the OpenEmbedded build system issues a warning message. Feb 10 16:44:51 FEATURE_PACKAGES Note: Packages installed by features defined through FEATURE_PACKAGES are often package groups. While similarly named, you should not confuse the FEATURE_PACKAGES variable with package groups, which are discussed elsewhere in the documentation. Feb 10 16:45:51 short answer.. if you see "PACKAGE_GROUP" (variable with that name) it's invalid Feb 10 16:46:16 FEATURE_PACKAGES is the new name.. and contains various key phrases to select certain things Feb 10 16:47:13 fray: am I allowed to use "FEATURE_PACKAGES" to create a package group (loosely speaking) Feb 10 16:47:20 ? Feb 10 16:47:46 no.. it's a selection criteria when you write an image recipe Feb 10 16:47:47 to create a package group, a recipe just inherits packagegroup Feb 10 16:47:50 Or, in other words, if I want to define a group of packages, what should I do? Feb 10 16:48:05 it is possible to create a spdx manifest without a fossology server? I've got a build currently hanging on do_spdx Feb 10 16:48:06 if you look at meta/classes/core-image.bbclass you'll find a list of defined 'FEATURE_PACKAGES" Feb 10 16:48:32 CTpollard not that I know of.. the SPDX stuff is fallen out of maintenance.. Feb 10 16:49:13 fray: ah :/ at least I've got bitbake at the stage where it's atleast trying to do it Feb 10 16:49:15 the code was designed originally to allow you to setup a fossology system w/ the SPDX extension.. it would then package up the code you were building and send it remote for scanning, returning to your system an SDPX document Feb 10 16:49:28 rburton: and the recipe name will be the name to put in the IMAGE_INSTALL var then? Feb 10 16:49:43 but that is pretty much all it ever did.. there was not much organization, and the quality of the SDPX documents was certainly suspect.. (not the codes fault, but due to fossology rules) Feb 10 16:50:03 diego_r: yeah Feb 10 16:50:17 fray: what exactly was it sending to the fossology server, I'f I could just have that, I could probably tarball it and pass it manually Feb 10 16:50:30 it's sending the patches sources to the thing you are building Feb 10 16:50:35 'er.. patched Feb 10 16:50:53 rburton: perfect, thanks. Feb 10 16:51:45 what I've wanted, and no time to do it.. a tie to partial SPDX data on a 'hash' basis.. you hash the files, it points to SPDX information for that file.. it collects the files for a given set of sources and produces an SDPX that matches the recipe.. Feb 10 16:51:53 any hashes it can't find are things "to be scanned" Feb 10 16:52:44 but up to this point, what most people do is usse the 'archiver' option in the system, they get the src archives or similar out in the end -- then scan the SPDX after the build is completd with the artifacts Feb 10 16:53:37 fray: yeh, I've been using archiver, maybe I should try and play with the fossology settings some more Feb 10 16:54:38 I was trying to scan the licences directory created by the archiver Feb 10 16:54:48 maybe I should try the actual source dirs instead Feb 10 16:55:41 the archiver licenses come from the recipe meta data.. they're only as good as the person who built the recipe. (If you find differences, it's a bug and we want to know..) Feb 10 16:55:57 the fossology / SPDX side was a way to actually scan the code and generate the SDPX dictionary of information.. Feb 10 16:57:36 fray: cool Feb 10 16:58:33 the idea being that the recipe author info is a "hint" at what they think it is.. but it's not authoritative.. the SPDX information is (eventually) programatically scanned and verified by YOUR legal organization to give your company the confidence that it's right Feb 10 16:58:48 then if the "hint" is wrong, you can file a bug and we can get it fixed.... Feb 10 16:59:04 but frankly, few people are going past just using the hint.. Feb 10 16:59:18 yeh, I suppose it can check the hash of the licence matches but you don't actually know if the license type defined in the meta data matches the content of the licence Feb 10 16:59:28 I've only seen a VERY small set of our customers care beyond the high level we think this is GPLv2.. vs actually scanning the code) Feb 10 16:59:49 SPDX is also more then just license info, it has copyright statements and other things in it.. Feb 10 17:00:07 this is why you want to hash the source code files themselves.... this way you can detect a file has "changed" Feb 10 17:00:21 and determine if the change has a license, copyright or other impact that you need to know about from an IP perspective Feb 10 17:00:38 if the hash didn't change, then any prior automatic or manual scanner you've done should still be valid.. Feb 10 17:01:01 if all of the hashes in a set of sources (recipe) stayed the same, then the license you decided on would still apply -- since this shows the soruces have not changed Feb 10 17:01:41 indeed, you can tell that the content of the licence hasn't changed, but you still have to trust the licence type defined in the recipe matches what it actually is? Feb 10 17:02:08 once you move to SDPX, the recipe info is just a hint to the build system.. Feb 10 17:02:37 your legal department needs to "bless" something as being correct in their eyes.. either specific SDPX info that matches specific sets of files, the recipe hint, or other things they've deemed are correct.. Feb 10 17:02:59 (usual case I see is an excel spreadsheet with a list of either sources or recipe versions and the license they believe it is..) Feb 10 17:03:35 You have to think of this problem as a lawyer and not a technical person.. we have technical answers (and I've got suggestions), but the lawyers work in completely different way to determine the responsibilities.. Feb 10 17:03:44 (or you have companies and uses that just ignore it and hope it goes away) Feb 10 17:03:58 (and those people are going to get burned) Feb 10 17:04:25 fray: this is all really useful information for me, thanks Feb 10 17:04:45 has the irc logger stopped btw, it looks to have not recorded anything in 2016 Feb 10 17:05:05 Analogy I often use with lawyers is that this is no different then your company buying an MSDN subscription, building a product and then releasing it.. did you actually read and understand MS terms? Did you use libraries (DLLs) that they allow you to redistribute? or do your cusotmers have to go back to MS for the files.. or..... Feb 10 17:05:22 you can't just assume since you have aquired something legally that you can turn around and use it.. you need to review the contract terms.. Feb 10 17:05:47 that usually gets them to understand the Open Source is no different then commercial -- only you have more potential contracts to "read" and understand.. Feb 10 17:06:03 ...and it becomes YOUR responsibility to "find" the contract (license) Feb 10 17:08:25 another question, it seems that I would like to have valgrind-native , but it doesn't work : http://pastebin.com/e2h6AWxN . Also tried bitbake glibc-dbg . nothing provides that, but it is required in recipe Feb 10 17:08:26 fray: I will digest all of this, thanks again :) Feb 10 17:08:45 glibc-dbg is a binary package, not a recipe Feb 10 17:09:10 CTtpollard: no problem Feb 10 17:09:33 I care a lot about license issues (and I don't mind dealing with the IP lawyers, for the most part, but there is a lot that should be done (still) that hasn't been.. Feb 10 17:11:19 I'd love to take 3-6 months and design a system for managing SPDX data and interacting with the Yocto Project build system Feb 10 17:11:58 (University of Omaha Nebraska was working on a partial system, but I've been out of the loop on that for a couple years now.. I don't know what progress they've made) Feb 10 17:12:01 kergoth: then what should I do ? i tried creating recipe glibc in my meta with BBCLASSEXTEND = "native", but doesn't work Feb 10 17:12:06 Missing or unbuildable dependency chain was: ['valgrind-native', 'glibc-dbg-native'] Feb 10 17:12:15 'er.. UNO - University of Nebraska Omaha :) Feb 10 23:34:26 khem, ping Feb 10 23:35:09 armpit: ping Feb 10 23:35:13 :) Feb 10 23:35:42 khem, I have patches for uclibc stable branches. it has a PR, do I bump it? Feb 10 23:36:56 ex: fido PR = "r9" Feb 10 23:37:08 should I make it r10? Feb 10 23:37:38 * armpit does not recall when prserve kicked in Feb 10 23:49:25 armpit: PR is quite old Feb 10 23:49:40 for fido you shuld be good without bumping PR Feb 10 23:50:01 k Feb 11 00:52:08 * armpit that sucks... uclibc image wont build Feb 11 00:55:10 * armpit trys jethro **** ENDING LOGGING AT Thu Feb 11 02:59:58 2016