**** BEGIN LOGGING AT Wed Feb 03 02:59:58 2016 Feb 03 08:34:43 good morning Feb 03 11:38:59 dv_: wouldn't you be better off just using COMPATIBLE_HOST? Feb 03 11:39:02 doesnt work Feb 03 11:39:26 COMPATIBLE_HOST="(armv7*|aarch64*)" -> "incompatible with host arm-poky-linux-gnueabi (not in COMPATIBLE_HOST)" Feb 03 11:40:43 here's the recipe in question: http://pastebin.com/raw/yfE2k97H Feb 03 13:03:19 am I wrong to think RDEPENDS_${PN} is automatically converted into DEPENDS by adding -native to each rdep when you BBCLASSEXTEND+="native" ? Feb 03 13:13:19 ..but if I set RDEPENDS_${PN} = "x" and build for native, it does not automatically DEPENDS on "x-native". should it? (assuming BBCLASSEXTEND native) Feb 03 13:32:33 mago_: if so, wouldn't that automatically convert run-time dependencies into build dependencies for -native packages? Feb 03 13:35:13 joseppc: yeah, but there are no runtime dependencies for native packages Feb 03 13:39:08 joseppc: (so i think that makes sense) Feb 03 13:39:37 mago_: I didn't know that, it seems strange, why would a package that depends on python on your target wouldn't depend on it for the native environment? Feb 03 13:53:36 morning Feb 03 14:05:44 can I somehow make certain bbclasses only useable from within the same layer they are in? Feb 03 14:06:02 I want to backport some bbclass in oe-core master into a jethro branch of a custom layer Feb 03 15:59:29 dv_: oh, right, that's a bit irritating. you're right, TARGET_ARCH is not specific enough in this case. Feb 03 16:00:32 hm, this openembedded-architecture list is a bit of a pain in the neck. Feb 03 16:00:34 * pb_ unsubscribes Feb 03 16:01:14 why? Feb 03 16:01:51 nobody ever seems to post to that list without cross-posting to a million others, so I find myself getting about six copies of every message. Feb 03 16:02:38 which gets a bit tedious after a while :-} Feb 03 16:02:48 maybe you need better e-mail client :) Feb 03 16:03:23 heh, yes, probably. Feb 03 16:05:10 recommendations? it appears my mail client is full of known security holes :-/ Feb 03 16:13:28 gmail works for me Feb 03 16:18:19 mutt works for me, but isn't very good eliminating duplicates Feb 03 16:21:19 gmail does a sterling job :) Feb 03 16:22:00 kergoth: around? Feb 03 16:22:31 somewhat. whats up Feb 03 16:22:49 kergoth: do you have any opinion on moving the typed data stuff from oe.data to bitbake? Feb 03 16:22:59 kergoth: maybe exposing it as data.getTyped(name)? Feb 03 16:23:19 Richard and I are both agreed that we need variable typing in bitbake, absolutely, but not as an afterthought Feb 03 16:25:33 pb_: so the recipe seems fine? Feb 03 16:25:45 pb_: I'd send this in to meta-openembedded , in meta-oe/recipes-support Feb 03 16:25:47 * kergoth ponders Feb 03 16:34:17 kergoth: agreed Feb 03 16:40:32 dv_: yeah, I think so Feb 03 16:40:49 g'day kergoth Feb 03 16:41:04 rburton: not saying your proposed name is an afterthought, just saying we need to give it careful thought in general :) Feb 03 16:41:21 hey pb_ , how've you been? Feb 03 16:42:36 yeah, pretty good. just in the process of moving to a new office. Feb 03 16:43:03 kergoth: my proposal is to think about it a bit more ;) not sure i want to merge the current implementation to bitbake Feb 03 16:43:53 * kergoth nods Feb 03 16:44:36 rburton: I don't think we'd want to use it, but I was surprised at how well this prototype behaved: https://gist.github.com/kergoth/2788410 Feb 03 16:45:49 never knew about .groupdict() Feb 03 16:45:50 that's awesome Feb 03 16:46:43 The nice thing about doing something along those lines is we don't have to worry about *when* the conversion from string to type happens, since it was always that type Feb 03 16:46:49 of course it has other downsides, but.. Feb 03 16:46:53 food for thought Feb 03 16:47:39 using flag and conversion has complexities like what if i don't know what type this is, or someone changed the flag without changing the value and broke the conversion on me, or broke my assumption about which type this value is Feb 03 16:47:51 of course, you could enhance getTyped to accept a type, and choke if that type mismatches the type in the flag Feb 03 16:48:14 i.e. getTyped("FOO", "int") -> bb.fatal("expected type 'int' doesn't match defined type 'boolean'") Feb 03 16:49:01 yuck Feb 03 16:50:00 lots of little corner cases to concern ourselves iwth.. like everything else in this project :P Feb 03 16:50:50 i'd like to see a new file format proposed, but it'd be tough for bitbake to support that today. right now *how* things are done is tightly intertwined with *what* will be done, which makes it difficult Feb 03 16:52:55 going back to the earlier mention of email clients, gmail is nice in some ways, but it has two big annoyances for me. 1. mail i send is de-duped when it comes back, so mailing list filters / labeling doesn't fully apply to my own sent mail, and 2. there's no freaking 'reply to list' function Feb 03 16:53:09 so we're stuck with 'reply', 'reply all', or manually munging the to/cc Feb 03 17:17:30 kergoth: yah, I don't think I've ever found an email client that I'm totally happy with Feb 03 17:18:18 I suppose that thanks to the wonders of free software I ought to be able to create one, but somehow life always seems too short. Feb 03 17:18:35 and I can't help thinking that, given all the others suck so much, it must be quite a hard problem. Feb 03 17:18:47 heh, indeed Feb 03 17:18:51 i expect the same is true of browsers Feb 03 17:18:57 since they all seem to suck in one way or another Feb 03 17:19:00 it's not just the free software mail clients that suck Feb 03 17:19:05 good point Feb 03 17:19:28 yeah. well, browsers do seem to have reached a level where all browsers are more or less "good enough" for casual use. Feb 03 17:19:46 I can use firefox or chrome more or less interchangeably most of the time, and neither of them enrages me particularly. Feb 03 17:20:21 I rotate between them every 6 months or so when i get fed up with whichever one I'm using :) Feb 03 17:20:26 but yes, absolutely, when I said I've never found an email client that I like, that wasn't just free ones. I have never found a client that I am happy with at all, free or proprietary. Feb 03 17:20:37 chrome seems like it's only fast with tiny numbers of tabs now Feb 03 17:20:39 which makes me sad Feb 03 17:20:58 yeah. which is odd, since it has a process per tab, you'd think it would scale indefinitely. Feb 03 17:21:15 I suppose maybe each process is pegged at 100% cpu and it's only fast until the number of tabs exceeds the number of cores in your computer. Feb 03 17:24:29 JaMa: I have seen this error with latest meta-qt5 master: http://pastebin.com/raw/zs0hXFe0 Feb 03 17:24:33 does this ring any bells? Feb 03 17:30:54 dv_: probably needs to skip dev-so now after last changes in oe-core Feb 03 17:31:33 just saw that there's also jansa/master , perhaps it fixes it too Feb 03 17:42:32 dv_: nope Feb 03 18:58:15 hi, is there a way to create a tarball that content the delta between 2 images. I mean I have 2 images the standard one and one with extra tools for development and we want to have a tarball with those tools Feb 03 19:08:42 mkdir a ; cd a ; tar -xpf standard.tar ; cd ..; mkdir b ; cd b tar -xpf extra.tar ; cd ../a ; for i in `find . ` ; do rm ../b/$i ; done ; cd b ; tar -cpf - . > /tmp/extras.tar Feb 03 19:09:15 debugging typos is left as an exercise to the reader. :) Feb 03 19:14:25 paulg_: it's a way I guess Feb 03 19:16:02 chouimat, not sure you can expect anything more than that. Feb 03 19:17:44 paulg_: I need it to be integrated to the build and something more like when we generate the sdk or toolchain tarball. Feb 04 01:55:54 otavio: we should be good to go if you can explain "Keep test cycle updated" Feb 04 01:56:18 is that on your builds or something we do locally? Feb 04 01:56:53 http://freescale.github.io/doc/release-notes/1.8/index.html#machine-maintainers **** ENDING LOGGING AT Thu Feb 04 02:59:59 2016