**** BEGIN LOGGING AT Fri Jul 20 03:00:01 2018 Jul 20 06:42:55 Hi, is there a way I can run a bitbake build offline? I have all the required sources of the target in the sources directory, and I have built the target multiple times with internet connection, but when I build offline it fails because it needs internet access during parsing recipes Jul 20 06:44:20 Can I tell bitbake to skip running git ls-remote during parsing recipes so I can build offline? Jul 20 11:26:34 New news from stackoverflow: How do i port cgi module to embedded systems with help of yocto? [on hold] Jul 20 13:15:50 how do i determine which installable image items A, B, C, ... which a recipe provides that I can added to an image recipe via IMAGE_INSTALL += "A B C ..." ? Jul 20 13:17:28 i'm not certain of the terminology so I'm using the term "items" - don't know if they are called "packages" or what. Jul 20 13:17:48 namely i'm asking about meta-swupdate Jul 20 13:25:02 is it defined by the do_install() script in the recipe? e.g., https://paste.fedoraproject.org/paste/4meseX~3LefdBWt-W87H4A Jul 20 13:31:46 s/is it/are they/ Jul 20 13:36:00 yates: They are called packages Jul 20 13:36:32 And there are a lot of ways they can be specified, the main on being the PACKAGES variable in a recipe Jul 20 13:37:16 oe-core's bitbake.conf has a set of default packages that are usually correct for most recipes, which is why you son Jul 20 13:37:31 don't see that variable set explicitly often Jul 20 13:44:37 thanks for the clarifiaitons, JPEW Jul 20 13:44:43 clarifications Jul 20 13:44:57 dyslexic fingers.. Jul 20 13:52:40 other than PACKAGES, can you list the one or two other, most popular ways it is done? Jul 20 14:09:48 its just PACKAGES Jul 20 14:10:10 some recipes construct that in non-obvious ways (using do_split_packages) but that's still just writing to PACKAGES Jul 20 14:10:24 best way is build recipe then use oe-pkgdata to list the packages the recipe generated Jul 20 14:17:59 nice tip, thanks rburton Jul 20 14:26:11 rburton: did you mean oe-pkgdata-util ? Jul 20 14:26:21 yes Jul 20 14:29:08 i see the list-pkgs subcommand, but that emits all packages. how can i show just the packages from a specific recipe? Jul 20 14:29:34 nm Jul 20 14:29:43 got it: -p Jul 20 14:29:59 beautiful Jul 20 14:44:54 zeddii: the parallelism patches should help kernel-devsrc (the old large one) and kernel module packaging :) Jul 20 15:25:21 yates, rburton: I was thinking of RPROVIDES when meant it can be done other ways Jul 20 15:26:05 oops, just noticed newappend -e and my new edit command aren't mapping the virtual to real filename Jul 20 15:26:25 also it looks like a number of those commands might not be handling multiconfigs well, if at all, will have to test that.. Jul 20 15:26:30 ugh, need coffee Jul 20 15:28:39 hmm, need to review richard's series about parallelization of shlibs, etc, and that pending RFC patch series about hash equivalence Jul 20 15:30:20 I just got the u-boot build race condition (lib/efi_loader/helloworld.efi', needed by `__build'.), is there anything useful I can do to debug it before I try rebuilding? Jul 20 15:33:14 kergoth: We have too many ways of doing things in parallel with multiprocessing :/ Jul 20 15:33:22 kergoth: I'm trying to create one good way... Jul 20 15:33:56 yeah,i saw the series, looks good to me, i only reviewed the first patch of the series so far Jul 20 15:34:38 kergoth: I have a few more bits locally in testing, dropping utils.multiprocess_exec() and starting to get rid of the oe.utils.getstatusoutput() calls Jul 20 15:35:12 JPEW: archiving the entire build dir would be a start Jul 20 15:35:33 JPEW: then rebuild and diff the makesfiles to see what is different Jul 20 15:36:31 kergoth: also need to get rid of more of the shell=True bits Jul 20 15:36:43 we're just being lazy currently Jul 20 15:37:38 agreed. i think in some cases we were lazy and in other cases we dind't know about or didn't think about the fact that we can use shlex to convert fragments coming from the metadata Jul 20 15:39:08 kergoth: right. There are some cases where we really do use the shell and pipelines for example but not that often... Jul 20 15:40:49 rburton: musl succeeded Jul 20 15:47:05 Is there anyone interested in helping which what should be a straight forward cleanup? Jul 20 15:49:02 I'll post on the list I guess... Jul 20 15:52:13 RP: thanks, chopped up mut and will respin Jul 20 16:08:04 rburton, just the "mu" part ? Jul 20 16:09:25 Load on the system from a change package.bbclass and rebuild with a stampeding repackaging of everything is pretty satisfying :) Jul 20 17:24:05 Hey all, I'm trying to build an image provided by Digi for a iMX6ul dev kit and am running into failures during the do_rootfs stage of the build. Here is a link to my build output: https://paste.ofcode.org/dbi4SqiMHx3mfBPQnWvh8j Jul 20 17:24:52 I did a find tmp/deploy -name \*libcrypto\* which shows that the library does exist in the deploy directory: tmp/deploy/rpm/cortexa7hf_neon/libcrypto1.0.2-1.0.2o-r0.0.cortexa7hf_neon.rpm Jul 20 17:26:49 I am having a hard time tracking down what recipe is actually requiring libcrypto....from what I can tell the greengrass recipe doesn't list libcrypto as an RDEPENDS so I'm assuming it is something that it is inheriting the dependency, I just haven't figured out how to determine the recipe that is actually requiring that library...I'd love some insight into determining that. Jul 20 17:28:14 Here is a link to the Greengrass recipe that I'm using: https://paste.ofcode.org/rqwqxWHiNdeZ7JBjEtNepV Jul 20 17:30:18 Here are a few of my questions....1) what are some first steps in figuring how what recipe depends on libcrypto? 2) How do I figured out what recipe is providing the libcrypto in the tmp/deploy directory, and why isn't bitbake able to use that to satisfy the libcrypto dependency? Jul 20 17:57:31 huh, this is odd.. immediately after wiping tmp: https://gist.github.com/kergoth/b9b80511a684ac1eb4d042be4434b49f Jul 20 19:06:07 Does anyone know how testimage is reading from qemu's output? Jul 20 19:09:20 a little. I think it uses expect Jul 20 19:09:56 or do you mean the boot up sequence ? Jul 20 20:08:50 this makes no sense Jul 20 20:12:53 wow, something is doing a massive reordering of my PATH Jul 20 20:13:03 PATH outside of bitbake has a *completely* different order than in devshell Jul 20 20:13:10 * kergoth scratches head Jul 20 20:13:13 * kergoth checks bitbake -e Jul 20 20:22:59 PATH in BB_ORIGENV is almost completely different than PATH in the shell that's running bitbake. what the hell is going on with my build Jul 20 20:25:03 armpit: I meant when runnning the tests, it parses the output of qemu, but idk if its parsing a log or the actual serial terminal or what Jul 20 20:26:06 aehs29, oh.. have know idea Jul 20 20:27:03 armpit: haha its ok, thanks Jul 20 20:35:00 * kergoth bangs head against desk Jul 20 20:36:47 oh, shit, i see the problem.. damn.. Jul 20 20:36:50 * kergoth ponders Jul 20 20:46:55 FYI, use of pyenv to manage your python installs wreaks havoc on a bitbake build. even if 'python' in your PATH is python2 and 'python3' in your PATH is python 3, pyenv's wrappers add the bindir of the python install to the PATH, and the python3 install did put a 'python' binary in its install Jul 20 20:47:07 have to remove the 'python' link in the pyenv python3 install Jul 20 21:18:21 kergoth: Are you trying out pipenv? Jul 20 21:26:04 kergoth: I get the feeling "python" is going to be going away in fedora soon so we may need to fix things to python3 or python2 anyway Jul 20 21:26:12 (thanks arch) Jul 20 21:27:42 JPEW: not for this yet, no, just am using pyenv for my python installs on all the machines i work on, due to the differing distros with inconsistent python versions available Jul 20 21:28:10 also on the mac, as i'm not a fan of how the system pythons are handled Jul 20 21:29:12 kergoth: Ah. I played around with "pipenv for bitbake" for about an hour a while ago and ran into the same problem Jul 20 21:29:38 ah, interesting Jul 20 21:30:23 RP: I'm thinking we should just make the hosttools setup link python to python2 in the hosttools dir directly, rather than pulling 'python' from the host Jul 20 21:30:34 kergoth: I'd agree with that Jul 20 21:33:03 hmm, more selftest -j failures. Its full of races :( Jul 20 21:41:28 should we place our bets on who is going to win? **** ENDING LOGGING AT Sat Jul 21 03:00:02 2018