**** BEGIN LOGGING AT Mon Feb 01 02:59:58 2016 Feb 01 03:02:52 Hi guys! I have a possibly noobish question: I am attempting to build an image for the Raspberry Pi and want to include apcupsd. In my googling I discovered apcupsd is built by meta-oe, but i can't just include meta-oe in the layers file, as it depends on openembedded-core, which has it's own base... what's the correct way to go about including meta-oe? Feb 01 03:14:36 So you already tried adding meta-openembedded/meta-oe to your bblayers.conf? You're concerned about conflicts? Feb 01 03:34:00 @jynik yes Feb 01 03:34:24 It failed Feb 01 03:35:12 Had a parse error saying it was unable to inherit file classes/autotools_stage.bbclass Feb 01 03:35:37 the recipe is there under meta-oe Feb 01 03:36:04 can you pastebin your bblayers.conf? Feb 01 03:36:15 sure, hang on Feb 01 03:37:12 ScottyPcGuy_03: Probably good to include the Yocto and meta-openembedded git hashes in the paste.. Feb 01 03:38:36 * nerdboy is looking at jethro branch Feb 01 03:38:44 both files are there Feb 01 03:42:11 http://pastebin.com/N6RLhCJh Feb 01 03:44:04 Hrm... just tried it and found that it's currently blacklisted. Marked broken because it doesn't build with B != S Feb 01 03:44:19 (Yocto and meta-openembedded on jethro branches) Feb 01 03:44:25 Ohhh Feb 01 03:44:32 Jethro is latest? Feb 01 03:46:46 ScottyPcGuy_03: Yes, per https://www.yoctoproject.org/downloads Feb 01 03:47:50 Ah ok Feb 01 03:48:26 I guess I could try cloning fido instead Feb 01 03:48:38 http://cgit.openembedded.org/meta-openembedded/commit/meta-oe/recipes-support/apcupsd/apcupsd_3.14.10.bb?id=96a92761c0a1bb2317fa4ea422c32e4473405103 Feb 01 03:48:53 Looks like you can unblacklist it from your local.conf Feb 01 03:49:51 Oh ok Feb 01 03:50:47 ScottyPcGuy_03: How's the raspberry pi build? Last time I tried it, it took a ridiculously long time to fetch the broadcom binaries :( Feb 01 03:51:09 Took about 2 hours on my laptop Feb 01 03:51:45 I'm using my desktop for this attempt so I've yet to know how fast it'll go Feb 01 03:51:50 Not too shabby then. I was seeing some crazy git fetcher times on that one recipe, often resulting in timeouts. Feb 01 03:52:04 Yeah it did timeout once Feb 01 03:54:15 So is there a way to tell the age of the releases on the website? Feb 01 03:55:52 Oh derp Feb 01 03:55:58 Click on them. DUH Feb 01 03:56:28 Don't forget the cgit page: http://git.yoctoproject.org/cgit/cgit.cgi/poky Feb 01 04:05:49 I ma gtting followign error on iMX6 Feb 01 04:05:54 mx-ipuv3 imx-ipuv3.0: ERR:[0xb5a31400]-no:0x21 "wait_for_comp_timeout" ret:0,l3 imx-ipuv3 imx-ipuv3.0: warning: disable ipu dma channel 9 during its busy state imx-ipuv3 imx-ipuv3.0: warning: disable ipu dma channel 9 during its busy state imx-ipuv3 imx-ipuv3.0: warning: disable ipu dma channel 9 during its busy state imx Feb 01 04:10:08 jynik: looks like cloning dora instead worked (was just before the blacklist was commited). Feb 01 04:11:18 ScottyPcGuy_03: You really want to be on dora though? :/ Feb 01 04:11:43 Did you give Jethro plus unblacklisting it local.conf a shot? Feb 01 04:11:49 Not really... I was just shooting stuff at the wall to see what sticks Feb 01 04:11:56 Ah gotcha. Feb 01 04:12:46 No, since this is my first time using Yocto I'm not really sure how to do that. Been using a tutorial ;) Feb 01 04:12:55 * ScottyPcGuy_03 googles Feb 01 04:22:27 Hmm Feb 01 04:22:45 So I can't find any info on what to actually put _in_ the conf file Feb 01 04:26:19 jynik: I don't see any documentation on unblacklisting... can you give an example (or if I'm missing the docs, point me to them)? Feb 01 04:28:42 ScottyPcGuy_03: I haven't ever done it before, so I'm taking a guess here. But it looks like placing the following in your local.conf should do the trick: Feb 01 04:28:48 PNBLACKLIST[apcupsd] = "" Feb 01 04:28:52 Ah Feb 01 04:29:08 Ok I'll give it a shot. Just finished cloning jethro Feb 01 04:29:27 Just a guess that whatever parses PNBLACKLIST checks for an empty string...but I did not confirm this beyond seeing that I don't get the error. Feb 01 04:29:30 http://www.yoctoproject.org/docs/2.0/mega-manual/mega-manual.html#ref-classes-blacklist Feb 01 04:29:43 Shows how to blacklist... kinda just guessed for unblacklisting Feb 01 04:30:17 Yeah, I was looking at that Feb 01 04:30:27 I'm honestly not sure if unblacklisting it is going to yield some sort of build or QA issue later though. Feb 01 04:30:57 Try and see I guess. You'll at least have more info to ask folks who know more than I do. Feb 01 04:31:05 Right Feb 01 04:32:03 Nope Feb 01 04:32:17 It's erroring out on scsirastools Feb 01 04:32:37 Same thing as before Feb 01 04:33:50 parse error at .../meta-oe/recipes-extended/scsirastools/scsirastools_1.6.4.bb:17 could not inherit file classes/autotools_stage.bbclass Feb 01 04:35:47 Strange... that bbclass file definitely exists in Jethro @ meta/classes/autotools_stage.bbclass, and I see you have meta in your bblayers.conf Feb 01 04:35:57 yeah... Feb 01 04:37:03 Hmm Feb 01 04:37:15 I don't see it in any directories Feb 01 04:38:50 http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/classes?h=jethro Feb 01 04:39:14 Should be in there... Feb 01 04:39:51 Huh Feb 01 04:39:57 wait a second Feb 01 04:40:17 Are you still back on Dora? I'm looking at Jethro. Feb 01 04:40:26 No, I'm on jethro Feb 01 04:41:06 Ok, so I cloned poky and meta-raspberrypi from yoctoproject.org. I cloned meta-openembedded from github Feb 01 04:41:26 And for all of these, you did 'git clone -b jethro ' ? Feb 01 04:41:56 Nope, I did master. That's probably what did it Feb 01 04:42:07 * ScottyPcGuy_03 tries again Feb 01 04:47:30 Hmm Feb 01 04:48:20 I'm getting pretty tired here, sorry. If I continue any longer, I'll just give you worse help ;) Feb 01 04:48:23 good luck! Feb 01 04:48:47 It's ok, I appreciate the help you've given! Feb 01 04:49:20 Bitbake is now erroring on libsdl being assumed provided. That I think I can fix Feb 01 04:56:15 ScottyPcGuy_03: np. Just a heads up, the do_configure step failed on apcupsd for me... so it's blacklisted for a good reason I suppose. gnight Feb 01 06:40:14 Hello fine folks! Feb 01 08:22:23 rburton: ping Feb 01 08:33:11 good morning Feb 01 08:45:27 mckoan: good morning! Feb 01 11:15:46 We'll that wasn't fun. My PC just had a major wobbly. Feb 01 12:28:26 rburton: any info on http://lists.openembedded.org/pipermail/openembedded-core/2016-January/116483.html ? Feb 01 12:49:13 not yet, haven't looked again Feb 01 12:57:44 hi, does anyone ever used meta-debain , n successfully created an image with feature "packagemanagment" Feb 01 15:22:22 I want to use yocto to build my system for deploying python apps to beaglebone. Does it means that I need to rebuild the full image everytime I have a hotfix of my application? Feb 01 15:38:04 AxonetBE: you can have a package feed and use a package manager to update Feb 01 15:38:37 nerdboy: did not know about that - found a yanncook.com - or are you refering to "yan can cook" ? :) Feb 01 15:39:22 rburton: you mean just creating a linux package? Feb 01 15:43:15 AxonetBE: packages will be created as part of your build. you can expose the set of packages online and use rpm/opkg to upgrade the images Feb 01 17:26:17 yann|work: the latter Feb 01 17:26:29 it was popular many years ago... Feb 01 17:31:23 kergoth: around? Feb 01 17:31:48 RP: somewhat. whats up Feb 01 17:32:16 kergoth: I have a bit a of a philosophical dilemma with bitbake and python functions :) Feb 01 17:32:34 kergoth: Right now, some of the code expands the python before running it, some of the other code runs it unexpanded Feb 01 17:32:52 kergoth: I'm wondering if we should always not expand strings in the python? Feb 01 17:34:17 kergoth: FWIW, anonymous python isn't expanded at run time but is for variable dependencies, python task functions are expanded, general python functions probably aren't and inline python isn't Feb 01 17:35:36 kergoth: Most of the changes I see are things like: Feb 01 17:35:38 WARNING: func relocatable_binaries_preprocess changed from rpath_replace(d.expand('${SYSROOT_DESTDIR}'), d) Feb 01 17:35:38 to rpath_replace(d.expand('/media/build1/poky/build/tmp6/work/x86_64-linux/pseudo-native/1.7.4-r0/sysroot-destdir/'), d Feb 01 17:36:08 so some general python functions do get expanded at least Feb 01 17:45:24 RP: yeah, i've thought for a while that things would get a lot simpler and cleaner to stop expansion of python. we have direct access to the api to do expansion when we want already. We should probably keep it for inline python as a convenience, but otherwise.. Feb 01 17:46:30 sometimes you'd see an expansion error just because it was forced to be expanded early for code parsing.. Feb 01 17:48:43 kergoth: it would also be a lot less confusing for people to find that d.expand() happens at runtime, rather than before Feb 01 17:49:19 kergoth: I'll run some tests and see how much carnage changing this would let loose. Right now its looking like surprisingly little! Feb 01 17:49:45 that'd be a pleasant surprise Feb 01 17:50:05 seems like any time we change stuff like this we find a bunch of people relying on the existing behavior whether it was good or bad to rely on it :) Feb 01 17:52:56 kergoth: I'm still wondering if I really did make it do what I think I did ;-) Feb 01 17:53:28 kergoth: Equally, it might be that forcing everything through the codeparser code already cleaned up those kinds of references. Feb 01 17:54:00 that's a good point Feb 01 17:55:11 kergoth: top 7 commits of http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=rpurdie/wip are my wip on this, running builds atm. Obviously needs cleaup Feb 01 17:56:47 kergoth: I think my favourite trick here is putting #${@bb.fatal("eek!")} into functions :) Feb 01 17:57:25 hah. nice Feb 01 17:57:33 the commits look reasonable thus far Feb 01 17:59:27 kergoth: perl do_package threw "ERROR: QA Issue: perl-ptest is listed in PACKAGES multiple times, this leads to packaging errors. [packages-list]" Feb 01 17:59:33 kergoth: so something breaks :/ Feb 01 18:00:01 out of time for today, we'll see how bad it is tomorrow... Feb 01 18:05:03 but that's perl... Feb 01 18:10:08 RP: I assume the problems would be internal use of wrong variables etc. may not be evident right away Feb 01 18:11:12 * nerdboy has a stubborn perl-doesn't-like-gentoo-arm problem Feb 01 18:11:54 crazy upstream ifdefs... Feb 01 18:12:50 by definition the preprocessor is evil Feb 01 18:13:03 but some people are just plain stupid Feb 01 18:13:49 same perl code builds/barfs 3 different ways on 3 different machines Feb 01 18:14:48 and two of them are (technically) the same architecture! Feb 01 18:15:10 * kergoth chuckles Feb 01 18:15:14 stupidity is a choice Feb 01 18:16:02 khem, you remember me bitching nearly 15 years ago how aweful perl is/was and likely will be? ;) Feb 01 18:16:05 abusing the preprocessor like that should get your fingers broken Feb 01 18:17:11 yes thankfully after working on icc I had to never use it Feb 01 18:17:25 lol Feb 01 18:17:25 "here, you get the giant old-fart keys... use your elbows..." Feb 01 18:17:27 icc testsuite was all perl Feb 01 18:17:49 bypassed to using python and js now Feb 01 18:18:05 and other functional languages Feb 01 18:18:11 and other functional languages Feb 01 18:18:30 even C++ is inferior now Feb 01 18:19:00 * nerdboy has 3 versions of gcc, 4 of python, 2 of llvm/clang, 3 of ruby... Feb 01 18:19:11 perl just wets itself Feb 01 18:38:13 clang good boy Feb 01 18:38:43 * armpit saved by the bell ? Feb 01 18:40:35 http://errors.yoctoproject.org/Errors/Details/31531/ Feb 01 18:40:43 what does that error mean Feb 01 20:13:36 I'm puzzled by something. to start playing with a new platform, I only bitbaked gcc-cross-aarch64 to start with (and with rm_work in effect). This, to my surprise, deploys a toolchain without libgcc.a, and without even a populate_sysroot.tgz in the sstate. Feb 01 20:15:55 realizing that, I bitbake libgcc, which now has to recompile gcc-cross-aarch64 from scratch - sounds like a genuine bug, doesn't it ? Feb 01 20:38:46 khem: its a good question. It could quietly break, yes :/ Feb 01 20:39:23 busybox, perl and lixcb failing to package in the test build I let running Feb 01 20:39:40 and another failure which I think is likely some other breakage in my test tree :( Feb 01 20:39:56 Exception: IOError: [Errno 2] No such file or directory: '/media/build1/poky/build/tmp/work/core2-64-poky-linux/busybox/1.24.1-r0/image${sysconfdir}/busybox.links.nosuid' Feb 01 20:40:04 obvious at least :) Feb 01 20:59:06 and strangely using "ld --sysroot=..." as is done when compiling a kernel under bitbake does not seem to do its job: strace reveals libgcc.a to be only looked for in current directory (and symlinking libgcc.a in the current dir does allow to proceed...) Feb 01 21:05:01 yann|work: shouldn't that be -ld ? Feb 01 21:05:45 yann|work: libgcc would be in the libgcc target recipe, not in gcc-cross-xxx Feb 01 21:06:43 khem: are you planning a patch for that ltp issue? If not, I think I might have one just to remove that test Feb 01 22:11:38 RP: yes, I omitted the ld prefix for brevity :) Feb 01 22:13:14 RP: you're right, it has also built libgcc for the target - but then, I'm wondering why it thought it had to rebuild the croxx-compiler too! Feb 01 22:16:34 any easy way to understand after the fact what change caused the rebuild ? Feb 01 22:18:51 yann|work: knowing a bit about the way gcc source is shared by all these pieces, I'm not that surprised :/ Feb 01 22:54:25 Anyone built with meta-mingw recently? Feb 01 22:54:42 I do regularly.. Feb 01 22:54:51 last time I looked the main meta-mingw was still out of date.. Feb 01 22:55:02 but I posted all of my changes to a poky-contrib.. let me look quickly Feb 01 22:55:22 yeah i noticed the appends were out of date, and also there's an issue with the provider preferences for libc, it seems Feb 01 22:55:27 thanks Feb 01 22:55:29 actually it looks like meta-mingw was updated 11 days ago Feb 01 22:56:00 I'm checking now if that syncs to my Jethro work Feb 01 22:56:04 ah, didn't see that, nice Feb 01 22:56:29 * kergoth tries again Feb 01 22:57:17 only 3 appends against old versions now..it's a start i gues Feb 01 22:57:19 not still a bit out of sync with what I have.. let me rebase and see how far off.. Feb 01 22:58:14 I should have it rebased on maybe 10 minutes.. Feb 01 22:58:20 on / in I mean.. Feb 01 22:59:27 cool ,thanks Feb 01 23:02:03 ok.. poky-contrib mgh/meta-mingw Feb 01 23:02:20 there are a small number of patches above the upstream.. this SHOULD work on both jethro and master.. Feb 01 23:02:25 (I've not built it yet though to verify) Feb 01 23:02:31 (doing that in a minute or so) Feb 01 23:04:04 the pathces I posted have a chunk that change behavior.. no more generating '.sh' files on windows, instead we generate proper .bat files.. Feb 01 23:04:18 ahh it didn't build.. looks like gdb and binutuils versions have changed sufficiently to stop it Feb 01 23:06:44 ...build starting... Feb 01 23:07:42 I've pushed what I'm building.. but I'm guessing I've got maybe 15 minutes until it completes.. Feb 01 23:08:33 np, thanks for looking into it Feb 01 23:09:35 I've been trying to keep things up to date there.. Feb 01 23:16:00 i haven't played with it too much yet, just starting to look at builds with the mingw SDKMACHINE for windows targeted sdks Feb 01 23:42:09 hmm gdb 7.10.1 didn't build under meta-mingw Feb 01 23:42:24 wonder if I have a patch for that that didn't get into my public tree (might still be in internal testing too) Feb 02 00:01:00 ahh found the gdb issue.. the way readline is defined changed in the 7.10.1 version.. Feb 02 01:49:19 How do I ensure that eglfs_brcm is quilt with QPA? Feb 02 01:49:25 Built* **** ENDING LOGGING AT Tue Feb 02 02:59:58 2016