**** BEGIN LOGGING AT Wed Nov 06 02:59:58 2013 Nov 06 04:13:13 khem: hi. i see that you have added chromium recipe in meta-browser. have you ever used that on an ARM platform? Nov 06 08:21:47 ant_work: kernel 3.12 should be ok for h1940, but I didn't test it (tested only rc) Nov 06 08:21:59 good morning Nov 06 08:22:08 morning Nov 06 08:22:29 ant_work: as for userspace - opie should work with this PDA Nov 06 08:22:48 ok, atm we have linux-yocto-dev to 3.12 Nov 06 08:23:12 linux-yocto is still 3.10 Nov 06 08:23:38 and I guess 3.10 was bad for h1940... Nov 06 08:24:36 Crofton: ok Nov 06 08:26:19 ant_work: I guess so Nov 06 08:27:22 the patch to boot older arm kernels built with recent oe-core toolchain? Nov 06 08:27:40 Crofton, ant_work: did you find the patch to boot older arm kernels built with recent oe-core toolchain? Nov 06 08:28:54 iirc there are at least 2 patchs needed to the kernel for that Nov 06 08:30:25 koen: I used linux-2.6.30 and I solved it simply using udev 164 Nov 06 08:30:31 hello Nov 06 08:30:33 I think both Crofton and woglinde did solve Nov 06 08:30:39 hrw: hi Nov 06 08:31:03 ant_work: ok thx, I missed IRC for a few days Nov 06 08:31:05 mckoan: that's not a toolchain issue, but a syscall issue Nov 06 08:31:38 or rather Nov 06 08:31:42 the kernel boots fine Nov 06 08:32:18 koen: IIRC my kernel booted fine but hung after init Nov 06 08:32:25 right Nov 06 08:32:34 lacking CGROUPS et al. Nov 06 08:32:43 koen: IIRC I solved with udev 164 Nov 06 08:32:44 you need the patches to hookup sys_accept4 and some sysfs layout patches Nov 06 08:33:09 hmmm Nov 06 08:33:14 I take that back Nov 06 08:33:23 what you need is to upgrade to a more recent kernel :) Nov 06 08:34:24 koen: of course now I'm happy with a fresh 3.12 :-D Nov 06 08:34:27 it is however an improvement: one can build old kernel with oe-core avoiding oe-classic just downgrading udev Nov 06 08:35:51 unsure about latest gcc, though Nov 06 08:35:51 oh, and some nasty renames wrt headers Nov 06 08:37:28 that reminds me my evening yesterday Nov 06 08:37:47 I was wondering how many ARM devices are fully supported in 3.12 kernel Nov 06 08:38:06 and is any of those newer then 1-2 years Nov 06 08:40:36 hrw: define "fully supported" Nov 06 08:40:49 "boots with ramdisk and serial" is how most vendors define it Nov 06 08:40:58 or "with 2000 patches on top" Nov 06 08:41:17 koen: whatever is on board works - do not count opengles closed stuff Nov 06 08:41:40 koen: so i2c, spi, gpio, can, screen, buttons, ethernet, serials, whatever, etc Nov 06 08:42:21 powermanagement as well? Nov 06 08:42:30 if so, you need to go back 4 years or so Nov 06 08:42:38 hrw: so just the pandaboard then :-D Nov 06 08:42:48 hrw: and old beagleboards Nov 06 08:43:07 suxx Nov 06 08:44:02 actually I think things like s2c24xx dev boards and mx31 dev boards all still work, so stuff from that era Nov 06 08:44:11 but modern stuff like Arndale sucks Nov 06 08:44:25 arndale is so last year Nov 06 08:44:31 arndale octa is all the rage now Nov 06 08:44:43 which I guess is just as broken or worse? Nov 06 08:45:10 well Nov 06 08:45:24 it looks to be the same deal as with arndale Nov 06 08:45:28 XorA: they put 5420 or 5410 on it? Nov 06 08:45:33 except with more promises of better behaviour Nov 06 08:45:35 last I tried arndale with mainline, it worked but no audio (didn't need it tho) and LPAE was broken Nov 06 08:45:36 5420 Nov 06 08:45:47 I lost hope when it comes to Exynos cpus Nov 06 08:45:52 suihkulokki: and usb which means no network Nov 06 08:45:54 and without lpae no kvm, which was what I wanted to test Nov 06 08:46:18 #oe is polluted with #linaro people ;D Nov 06 08:46:33 * koen rubs eyes Nov 06 08:46:43 damn, I really thought this was #Linaro :) Nov 06 08:46:45 RMK has super patch of doom to fix LPAE issues for good, but the patch is so huge its probably never going to merge Nov 06 08:47:02 btw - what is 'M' in RMK? Nov 06 08:47:14 I dont know Nov 06 08:48:33 XorA: what's sticking with usb? Nov 06 08:49:09 hrw: the allwinner community stuff is looking good in mainline as well, but I don't know the exact status Nov 06 08:49:13 suihkulokki: that special hub initialisation that needs done, I have not seen a patch enter mainline for that but I might have missed it Nov 06 08:50:32 koen: the fun would be when Chinese chips will get better support in mainline than other vendors ;D Nov 06 08:50:44 well Nov 06 08:51:01 most of the 'new' chinese SoCs use stock ARM IP blocks Nov 06 08:51:02 koen: but AllWinner/MediaTek/etc are using ARM components so it should make life easier Nov 06 08:51:08 heh Nov 06 09:36:23 morning all Nov 06 10:06:49 morning all Nov 06 10:08:59 hi bluelightning Nov 06 10:09:12 hi woglinde Nov 06 10:11:36 hi ro Nov 06 10:11:39 aehm rp Nov 06 10:14:23 morning all Nov 06 10:15:11 hi Richard Nov 06 10:17:01 hi RP, hrw Nov 06 10:27:33 hello Nov 06 10:28:10 I used the cronie recipe and try to edit /etc/crontab file but crond does not seems to read it Nov 06 10:28:59 however when I use crontab command (which edits a crontab specific to the user) it worked well Nov 06 10:29:27 so why a file /etc/crontab is provided in this case? Nov 06 10:45:59 /var/spool/cron* Nov 06 10:48:02 yes there is a crontab per user Nov 06 10:48:16 so /etc/crontab is not read? Nov 06 10:53:23 hi. i am new to openembedded and have some small question. After several failures i got finally the sources for the Alcatal OT 890 Device that is based on a MTK6516 MediaTek SoC. I was able to recompile the needed kernel with my changes and i am trying now to create a Mobile Linux Distro. My question is would i be able to build such a Mobile Linux Distro with OE for this Device? I have seen... Nov 06 10:53:25 ...there is a similar device called SCIPHONE in OE also based on a MediaTek MTK SoC. Beside this what for GUI can be build and ported with OpenEmbedded. Exist the Possibility to Port as a example Ubuntu Touch to OE if i dont like the guis like Opie and so on? Thanks for the answers. Nov 06 10:55:08 xperia: as long as your device has working framebuffer you can run X11 on it. and then? whatever Nov 06 10:56:52 hrw: cool sound very good. i like also a lot the very clean enginered OpenEmbedded easy understandable structure compared to android. porting android to non supported devices is a real pain for me at least. Will give OpenEmbedded for sure a go. Nov 06 11:31:09 morning all Nov 06 11:31:53 morning pb_ Nov 06 11:33:53 does anyone know if meta-perl is still "current"? There have only ever been three commits in that directory, all on the same day, and it contains a grand total of six files Nov 06 11:33:55 (three of which are README, COPYING and layer.conf) Nov 06 11:34:39 pb_: there are a couple of problems there Nov 06 11:35:22 pb_: Stygia sent a bunch of patches to add recipes there but they needed work, and he hasn't followed up on that yet Nov 06 11:35:40 pb_: also I'm not sure the maintainer of that layer is especially active Nov 06 11:36:23 I'll poke Stygia when he comes online, supposedly he has a large collection of perl recipes to contribute Nov 06 11:36:23 did that layer get abandoned, or did it move somewhere else? Nov 06 11:36:24 right. he certainly doesn't seem to be especially proactive: he's only added two recipes of his own. Nov 06 11:36:29 ok, cool Nov 06 11:36:34 no, nothing moved AFAIK Nov 06 11:36:52 I have a bunch that I have collected which I could contribute, but I would have to do some work to those as well and I don't want to do that if it's going to be a waste of time. Nov 06 11:38:31 it does sort of seem as though it ought to be possible to make an automatic recipe generator for the things that are "just" perl modules from CPAN, which would save a bunch of typing. Nov 06 11:39:01 right, aside from corner cases Nov 06 11:39:20 I seem to recall that last time I was trying to make xmltv work it was a fairly soul-destroying experience having to write recipes for all the perl modules it needed. Nov 06 11:41:30 Stygia: ah, just the person... Nov 06 11:41:41 Stygia: I just emailed you re meta-perl stuff Nov 06 11:42:41 bluelightning, Oh? Nov 06 11:42:53 my mail watcher thing doesn't show anything. Nov 06 11:43:12 I know the commits I made previously were broken as fuck, I've been waiting till I'm 100% sure everything is ready. Though, many of the things definitively are, actually. Nov 06 11:43:28 I'll do it today. :) We've just been busy here at the company lately. Nov 06 11:43:41 We have a huge order for Polish trains to fill, 200+ units to sell, which would make us, so. Nov 06 11:43:48 But, my part is working great (including OE stuff). Nov 06 11:43:51 So I'll start pushing stuff. :) Nov 06 11:43:54 Stygia: ok, np Nov 06 11:43:57 I even made a repo for it. Nov 06 11:44:04 Stygia: as I said in the mail I'd be happy to look over the recipes if you could point me to the repo Nov 06 11:44:27 Yup, but the repo we actually have is private for now. Nov 06 11:44:58 I'm moving it to another repo which'll be public. Nov 06 11:45:13 We have recipes for our private stuff in this current one, including git info, so. Nov 06 11:45:28 Not auth info but still. Nov 06 11:46:51 None of the perl stuff, though. Nov 06 11:46:51 Oh. Nov 06 11:46:59 I made a hackish bbappend to rrdtool, to allow the perl extension to work. Nov 06 11:47:04 nto sure if that's wanted or whatn. Nov 06 11:50:07 Stygia: which company you work for (if I may ask)? Nov 06 11:51:04 Movis, a startup in Denmark working to build a small embedded box, that you could say place in a bus or train, which then provides video streaming, magazines, ebooks, audiobooks, etc, and internet access. Nov 06 11:51:08 Movis for mobile video streaming. Nov 06 11:51:24 interestng sounds Nov 06 11:51:31 If you're north American, keep in mind that public transportation is huge. Nov 06 11:51:36 But that grammar sounds German. Nov 06 11:51:42 Stygia: I am from Poland Nov 06 11:51:47 bluelightning, https://github.com/EmilRP/public_bb_recipes Nov 06 11:51:49 Here you go. Nov 06 11:51:56 But this is not actively cleaned up, not 100% yet. Nov 06 11:51:57 Been busy. Nov 06 11:52:07 hrw, Hah, we're selling to Przewozny Regionale. Nov 06 11:52:12 I can't spell that, but still. Nov 06 11:53:10 Well, _trying_ to sell to them, but still. Nov 06 11:53:38 bluelightning, In general, if it looks like a total mess it was either created long ago, or not by me. I _think_ I've got the gist of things w.r.t perl recipes by now, so. Nov 06 11:53:59 Stygia: ok, great - thanks Nov 06 11:54:14 bluelightning, No problem. Nov 06 11:54:22 If you find issues, please let me one one-by-one so I can fix it upstream, alright? Nov 06 11:54:35 I removed that ugly perl bbappend we had, removed a lot of duplicate recipes for core modules, etc, lately. Nov 06 11:54:50 If I could travel back in time to when I started this would be awesome now. Nov 06 11:54:50 bl/uOh, one thing. Nov 06 11:54:53 bluelightning, One thing. Nov 06 11:55:03 My boss would really be happy if I could somehow link these commits to Movis. Nov 06 11:55:46 Stygia: typically I guess that would be accomplished by having these committed under your name + company email addr Nov 06 11:56:15 bluelightning, Yes, that would be fine. Nov 06 11:57:12 We don't want our name plastered through the recipes or anything, just somehow sign them off by a company-associated name/address. Nov 06 11:58:34 MIT/BSD at your leisure, BTW, I seem to remember you liking MIT. Nov 06 11:58:44 Anything but GPL and I don't personally object at all, nor would my boss. Nov 06 11:59:31 I think the best thing would be just to create a script that commits each one on top of meta-perl and then have you run that and then send the patches from there Nov 06 12:00:15 Oh? Ah to make the patches sequential properly etc. Nov 06 12:00:28 It is a lot to keep track off with 100+ recipes. Nov 06 12:00:34 That's what ls -l | wc -l says anyway, 107. Nov 06 12:00:51 right, you can take care of ordering afterwards with git rebase -i Nov 06 12:01:05 Which I read up on BTW, thanks for that. Nov 06 12:01:11 I'm much less derp with git. Nov 06 12:01:14 the script is just to have one commit per recipe and have your name on them Nov 06 12:01:25 and not have to do that by hand :) Nov 06 12:01:26 Hmm, right. Nov 06 12:01:31 Heh, yup. Nov 06 12:01:58 something like for (@files) { `git commit $_ -sm "added $file"`; } would do I think. Nov 06 12:02:16 Don't parse that too much, probably isn't valid, but it should cover the gist of the idea. Nov 06 12:02:24 Ah, and an intereger to do x/x Nov 06 12:03:50 a bash loop will do it Nov 06 12:04:39 Stygia: some of us are allergic to perl and just hit an immediate parse error :) Nov 06 12:04:55 RP, Pshaw, that's not even as bad as perl can get. Nov 06 12:05:12 Stygia: I say this as someone who once worked on a few perl applications too :/ Nov 06 12:05:24 bluelightning, Anyway, thin kI'll wait ano matter what until you've read it through. Nov 06 12:05:27 Stygia: that is positively readable :) Nov 06 12:05:35 RP, Ah, well. Perl is alright once you get to know it, but it isn't the best. Nov 06 12:05:45 Readable perl is up to you, if you write write-only, then it'll be write-only. Nov 06 12:05:56 Maybe I'm deluted but I have no trouble reading my own perl, at least. Nov 06 12:05:56 Mostly. Nov 06 12:06:01 Stygia: the trouble is it doesn't seem to stick with me. I guess its what you're used to Nov 06 12:06:30 Hmm, well. Nov 06 12:06:37 Perl makes sense once you think in context and lists. Nov 06 12:06:44 The only thing that still bothers me is rested ref structures. Nov 06 12:07:06 Because Data::Dumper is so difficult compared to, say, print_r in PHP (*shudders*, php) Nov 06 12:07:20 * RP notes that of the few computery books I own, there are two perl reference ones Nov 06 12:07:46 Stygia: Sadly I know exactly what you mean there... Nov 06 12:08:22 Yup... Nov 06 12:08:28 Working with a JSON api makes it worse. Nov 06 12:08:38 You get a data structure, every level being a ref instead of a proper hash or array. Nov 06 12:08:58 So it ends up being my @thing = @{ ... }; on every level of the parsing. Nov 06 12:09:02 Ah, well. Nov 06 12:09:02 Could be worse. Nov 06 12:09:02 on the topic: recent m4 and gettext native do require that perl on buildhost has threads. It is a new requirement, is it agreed? Gentoo seems discouraging it Nov 06 12:09:04 Could be PHP. Nov 06 12:09:28 I'm not sure about perl threading, it works but many people say it's shit. Nov 06 12:09:37 And, has excess resource consumption since it copies all local resources wholesale. Nov 06 12:09:45 I've read it was bad with 5.0 Nov 06 12:10:21 I was happily building w/out '-ithreads' up to last week Nov 06 12:10:22 Could be, I just remember people on #perl bad-mouthing it lately, telling me to use IO::Async Nov 06 12:10:27 Which is delicious BTW. Nov 06 12:10:47 I wrote a RRDtool logger script with IO::Async, ended up really neat IMO> Nov 06 12:10:53 ant_work: this is the first I've heard about this Nov 06 12:11:01 is on the ML Nov 06 12:11:22 Stygia: libdvdread, libpostproc and fribidi are all in meta-oe and appear not to be different from the versions there, so those should be dropped Nov 06 12:11:36 [oe-commits] Laurentiu Palcu : m4: upgrade to 1.4.17 Nov 06 12:11:45 bluelightning, Ah, yes those should have been deleted. Nov 06 12:12:27 bluelightning, I took them out of the repo. Nov 06 12:12:34 Stygia: additionally, libdbd-sqlite-perl and libdbi-perl are in meta-perl already (though they may be different) Nov 06 12:12:45 Stygia: also liburi-perl is in OE-Core Nov 06 12:12:49 bluelightning, Ah, well right. Nov 06 12:12:49 RP: I have an Ubuntu 12.04 32bit on the same disk and it could build m4-native from scratch. Gentoo could not with default perl flags Nov 06 12:12:58 bluelightning, Oh? OE-core has CPAN recipes? Nov 06 12:13:11 Stygia: one or two, where they are dependencies of other non-perl recipes Nov 06 12:13:22 (and only then) Nov 06 12:13:45 I wouldn't be adversed to having a small number in the Core which tested the key perl classes Nov 06 12:15:09 RP: I'd say let's just get these in meta-perl and go from there... Nov 06 12:15:26 (as a first step, I mean) Nov 06 12:15:27 bluelightning: sure, that is a good first step Nov 06 12:15:30 bluelightning: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t2&id=59823e6b93386e42c48020f98744295dd0ba88bd Nov 06 12:16:08 * RP is kind of surprised this is working Nov 06 12:16:12 nice... Nov 06 12:20:01 bluelightning: next is to see if this makes a difference to configure speed Nov 06 12:22:05 autotools and tests are slowest parts Nov 06 12:22:34 hrw: 40% of the build time is in configure these days Nov 06 12:23:04 RP: I am building Python3 now (Fedora) - >40% is in tests now Nov 06 12:26:05 Stygia: a very small number of these recipes are in meta-security atm as well, but that in and of itself is not a concern... however comparing one of them I see some interesting differences Nov 06 12:26:07 Stygia: http://pastebin.com/2ZYWG4MH Nov 06 12:27:48 Stygia: (the DEPENDS line from the original recipe is bogus though) Nov 06 12:27:55 RP: that looks like a poor mans per-recipe sysroot :) Nov 06 12:28:26 koen: the idea is that this code could allow for that Nov 06 12:28:35 Stygia: libhtml-parser-perl_3.71.bb has a typo "SUMARRY" Nov 06 12:28:38 koen: this is just an experiment to see how well it does/doesn't work Nov 06 12:29:47 RP: I bet it will quickly find missing deps Nov 06 12:30:18 koen: with the work Martin has done we're actually looking pretty good, but yes, it would likely find some Nov 06 12:30:38 Stygia: in fact if you have DEPENDS += "perl" in any of these recipes that inherit cpan, you can drop that line Nov 06 12:31:25 rp: i've been wondering just how much of configure is spent looking for stuff we can add to the site cache Nov 06 12:32:00 rburton: I'm leaning towards caching the result of the first time a given test runs Nov 06 12:32:18 rburton: the site cache was intended for cross compile values we cannot find another way Nov 06 12:32:41 rburton: keep in mind that of the 40% in configure, 20% is autoreconf and 20% running configure (approx) Nov 06 12:33:12 This patch is looking at making autoreconf faster as if it has less m4 files it should be quicker Nov 06 12:33:50 rburton: results cache would work on the other 20% Nov 06 12:43:06 bluelightning, I'll have a look, sorry, went to lunch. Nov 06 12:43:54 bluelightning, Are you sure it complies without the do_compile block? Nov 06 12:44:02 It is an old recipe, so I'm not sure it needs it per se. Nov 06 12:46:22 bluelightning, I'm a bit confused about this diff. Nov 06 12:46:33 The things with - in front of them. Nov 06 12:46:42 Are these things you patch vs the current, upstream recipe? Nov 06 12:47:17 bluelightning, Hmm, so would I be reading it right if I say this diff is the original recipe, patched with changes from mine? Nov 06 13:08:50 Stygia: correct (see the top of the diff) Nov 06 13:12:29 bluelightning, Yes, right. Nov 06 13:12:44 bluelightning, Sorta satisfying when my recipe is 'better' than the upstream one. Nov 06 13:13:30 Stygia: so the EXTRA_CPANFLAGS and do_compile stuff aren't necessary you think? Nov 06 13:14:39 Not that I've seen, I'm using libnet-dns without them. Nov 06 13:15:11 hmm ok Nov 06 13:19:36 Stygia: you shouldn't need to add the value of ${PN} to RPROVIDES_${PN} btw Nov 06 13:21:01 bluelightning, Oh? Just RPROVIDES = is enough? Nov 06 13:21:09 Stygia: no, that's not what I meant Nov 06 13:21:12 I thought it was best practice to append _${PN} to fields like that. Nov 06 13:21:27 Ah, then? Nov 06 13:22:08 Stygia: in libclass-accessor-perl you have it setting RPROVIDES_${PN} to include libclass-accessor-perl; that's already implicit Nov 06 13:23:00 FWIW I'm going through and tidying up the R* values now, I'll send a patch Nov 06 13:23:04 bluelightning, Oh, that's an error then. Nov 06 13:23:08 Ah, cool, thanks. Nov 06 13:23:11 R*? Nov 06 13:25:00 RDEPENDS, RPROVIDES etc. Nov 06 13:25:25 just found one recipe does DEPENDS_${PN}, that isn't valid Nov 06 13:41:41 Stygia: libdigest-sha1-perl has LICENSE set twice, which value is the correct one? Nov 06 13:42:47 bluelightning, The top one, Artistic | ... Nov 06 13:43:01 Stygia: ok Nov 06 13:50:03 Stygia: libgps-point-perl has RDEPENDS on perl-module-calar-list-utils - is that the right package name? Nov 06 13:51:54 bluelightning, Ah, no, that should be perl-module-scalar-list-utils Nov 06 13:52:27 Stygia: what provides that - I can't seem to find it Nov 06 13:53:15 bluelightning, Perl itself. Nov 06 13:53:38 bluelightning, It has that dynamic recipe creation stuff, where if you rdepend on perl-module-X it creates a module based on provided file that matches what you try and depend on. Nov 06 13:54:13 Stygia: dynamic package creation - but the packages are created from what the perl recipe produces, not what other recipes depend on Nov 06 13:54:54 bluelightning, Oh? I think I read somewhere, or heard from someone, that if you depend on something, the perl recipe provides it. Nov 06 13:55:11 it claims it provides perl-module-* yes Nov 06 13:55:18 but that doesn't mean it actually does at runtime Nov 06 13:55:38 bluelightning, Either way, Scalar::Util is core, so it should be perl-module-* Nov 06 13:55:41 bluelightning, Hmm, well. Nov 06 13:56:03 Scalar::List::Util is the package which provides Scalar::Util and List::Util Nov 06 13:56:11 So maybe that should be perl-module-scalar-util or pel-module-list-util Nov 06 13:56:55 one, or both? Nov 06 14:18:48 bluelightning, I'm not sure, let me check. Nov 06 14:19:23 bluelightning, GPS::Point wants Scalar::Util, the other thing is an error then. Nov 06 14:28:18 bluelightning, You'll note I fix things in this public repo (and my private one as well) and push it as you point things out. Nov 06 14:41:37 RP: any idea about the filespath ordering issue ? Nov 06 14:42:05 ant_work: what bug are you running into? Nov 06 14:42:13 ant_work: I knew there was something I was going to look at :/. Meant to look but haven't yet sorry Nov 06 14:42:22 I can see here that if defconfig is present, then it is copied to .config : https://github.com/openembedded/oe-core/blob/master/meta/classes/kernel.bbclass, but I provide no defconfig. Does oe do make defconfig at some point? Nov 06 14:42:24 I'm running into a FILESPATH bug after updating Nov 06 14:42:39 see SRC_URI computing orde Nov 06 14:43:22 ant_work: this fails now: https://git.linaro.org/gitweb?p=openembedded/meta-linaro.git;a=blob;f=meta-bigendian/recipes-core/initscripts/initscripts_1.0.bbappend;h=355fbab2156a98d2b434b8b5380e58fd645fd79a;hb=HEAD Nov 06 14:43:33 koen: http://lists.openembedded.org/pipermail/openembedded-core/2013-November/086005.html Nov 06 14:43:38 (apart from the fact that alignment.sh is the wrong thing to do on armv6+) Nov 06 14:43:58 koen: I fear some defconfig issues are coming from that same problem Nov 06 14:47:58 The trouble is the override ordering was completely broken, e.g. arm would be preferred over armv7. The policy was always supposed to be apply the most specific, not the least specific Nov 06 14:52:27 RP: surely the /foo//bar of the layer must have precedence Nov 06 14:56:51 ant_work: I need to look at what its doing exactly Nov 06 14:57:17 sure, formfactor is a perfect test Nov 06 14:57:53 just check in $S what is fetched Nov 06 15:03:02 Stygia: right, thanks Nov 06 15:03:10 Stygia: however you've changed it to perl-module-scalar-list-utils but that does not exist Nov 06 15:03:41 bluelightning, I think I changed it to perl-module-scalar-util? Nov 06 15:04:49 Ah, forgot to commit that, sorry. Nov 06 15:05:29 Never mind the lazy commit messages in this public one, it's not the one I'll actually use anyway. Nov 06 15:05:43 Stygia: sure, I'm not expecting you to keep the patches I send as commits either :) Nov 06 15:08:02 bluelightning, Oh, I'm changing whatever you point out. Nov 06 15:08:09 bluelightning, But I take it you'll send me a proper patch file to apply? Nov 06 15:08:24 bluelightning, Though of course, any overt improvements I might as well incorporate in my own repostirory. Nov 06 15:08:26 repository. Nov 06 15:08:30 Stygia: for the R* stuff and anything else I notice, yes Nov 06 15:08:58 Stygia: I figure let's fix up what you have and then you can create the meta-perl patches from there Nov 06 15:10:45 bluelightning, Yup exactly, thanks. Nov 06 15:10:47 bluelightning, R*? Nov 06 15:11:23 RDEPENDS and RPROVIDES? Nov 06 15:11:28 yes Nov 06 15:11:50 I'm fixing the formatting and pushing them towards the end of the recipe, since they are to do with packaging Nov 06 15:16:23 bluelightning, Oh? What's wrong with the formatting? And, I take it having them near the end is non-standard, or? Nov 06 15:16:39 Stygia: no, they should be put near the end Nov 06 15:16:51 Stygia: it's not consistent Nov 06 15:17:33 bluelightning, Yes, that's what I meant, sorry. Nov 06 15:17:36 bluelightning, Ah, right. Nov 06 15:17:51 it's mostly fine, I just figured I might as well touch it up while I was moving stuff Nov 06 15:18:01 bluelightning, Sorry about that, which do you prefer, " \ (newline) thing \, or " thing \ (newline) thing \?" Nov 06 15:18:03 ? Nov 06 15:18:42 I'm making it the latter Nov 06 15:24:39 bluelightning, Alright. Nov 06 15:24:54 bluelightning, I'm thinking maybe you could use a regex for it, but if you're walking them through anyway... Nov 06 15:25:03 it's ok, I'm almost done Nov 06 15:26:44 bluelightning, Cool, fantastic. Nov 06 15:26:59 bluelightning, Thank you man, I should be doing this myself, but I've just been busy. Nov 06 15:27:06 bluelightning, I hope the 100+ perl recipes should at least make up for it. Nov 06 15:27:16 bluelightning, And I hope mostly they aren't garbage. ^_^ Nov 06 15:27:22 np, I'm keen to see this squared away :) Nov 06 15:27:47 bluelightning, squared away? Nov 06 15:28:05 Stygia: sorted out... I believe that's US military slang originally Nov 06 15:28:14 bluelightning, Ah, alright, good. Nov 06 15:28:30 bluelightning, English isn't my first language so hard to catch up on all the slang. Nov 06 15:32:44 Stygia: can you add a SUMMARY for libmime-base64-perl and check that HOMEPAGE is correct? Nov 06 15:34:04 bluelightning, Yup, that HOMEPAGE is clearly wrong! Nov 06 15:34:34 bluelightning, There, pushed. Nov 06 15:52:43 bluelightning, I have to go reinstall my OS very soon. Nov 06 15:52:49 bluelightning, Troublesome, I know, but I'll be gone for maybe an hour. Nov 06 15:53:36 Stygia: ok, np Nov 06 15:54:08 bluelightning, See you soon if things go well. Nov 06 15:55:46 rburton: are you interested in crazy configure speedup prototype code? Nov 06 15:55:55 rp:heh, go on then Nov 06 15:56:56 rburton: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t2 Nov 06 15:59:28 rburton: doesn't look much but does scary things to builds Nov 06 16:00:32 oh god almighty :) Nov 06 16:01:01 the question is does stuff like AC_CHECK_LIB use the cache Nov 06 16:01:55 rburton: I'm wondering if we can write a tool to collect/sanitise these entries into a coherent central cache Nov 06 16:02:15 rburton: as it is anything which changes cflags explodes atm :/ Nov 06 16:02:48 and sadly that is rather a lot of recipes Nov 06 16:04:39 probably worth instrumenting the cache accessor and seeing how many cache misses there are Nov 06 16:08:29 rburton: its all horrible shell script, I doubt its practical or possible Nov 06 16:09:22 or alternatively rburton, why don't you show me how that is done? Nov 06 16:10:35 the implementation of AC_CACHE_VAL is in general.m4 Nov 06 16:10:41 its "just" m4 ;) Nov 06 16:11:29 rburton: I guess the crazy boilerplate must come from somewhere Nov 06 16:12:01 layers upon layers of m4 :) Nov 06 16:12:36 hmm, 18MB config.log files Nov 06 16:12:50 really need to write something to combine these files and remove dups Nov 06 16:15:58 * rburton tries Nov 06 16:17:01 rburton: the issue is that each cache file contains all the entries of the previous ones so it gets ever larger and each time, more files get included Nov 06 16:17:28 I'm kind if impressed it even vaguely builds though Nov 06 16:19:55 whoops broke my autoconf Nov 06 16:25:53 RP: http://pastebin.com/XzY6djnL Nov 06 16:27:06 rp: we could have a preload recipe that runs a configure script exercising a lot of the standard tests Nov 06 16:27:27 like the system headers, compiler flags Nov 06 16:27:29 etc Nov 06 16:28:02 (that patch was a two liner, we could easily ship it) Nov 06 16:30:22 rburton: we already have a notion of this idea with siteconfig.bbclass Nov 06 16:31:44 hmm, core-image-sato shows 12 failures with my patch... Nov 06 16:32:38 rburton: the thing which now scares me is that different cache files have different values for some of these :/ Nov 06 16:32:53 such as? Nov 06 16:33:16 rburton: lt_cv_path_LDCXX Nov 06 16:33:59 that is worrying Nov 06 16:55:03 kergoth: btw your idea with the symlinks worked perfectly, thanks Nov 06 16:55:12 rootfs built without errors Nov 06 16:55:33 rburton: 6152 of 6189 tasks suceeded for core-image-sato, three failures Nov 06 16:57:25 dv_: nice, glad to hear it Nov 06 16:57:27 np Nov 06 16:57:41 Hello everybody. I wanna ask if OE contains a package for open source project which configures the network setting of a box, (ip config, netmask, wireless essid, ..) Nov 06 16:57:42 already started a discussion for moving towards dora Nov 06 16:57:53 good luck :) Nov 06 16:57:54 vadmeste_: connman? Nov 06 16:58:04 existing projects that are using danny are a problem though. a danny->dora migration is not trivial, and will require lots of testing. Nov 06 16:58:41 yeah, we basically have to do a file-by-file audit of every layer we maintain every time we do an update to the next release, here Nov 06 16:58:51 it's good to do from time to time anyway, though.. Nov 06 16:59:23 I can imagine. same would happen here. Nov 06 16:59:52 and judging from some rather ugly bitbake scripts I've seen for our software, I think we should also audit our stuff :) Nov 06 16:59:57 hehe Nov 06 17:00:04 rburton: thanks for responding.. I forget to mention that I need a web interface not a daemon.. Nov 06 17:04:37 rburton: weren't you supposed to be writing one of those? *cough* Nov 06 17:04:47 bluelightning: shush Nov 06 17:10:28 e Nov 06 17:31:33 rburton: this is scary. I'm seeing dependencies of ""../../c++/.libs/libbeecrypt_cxx.so" for beecrypt which rpm doesn't like for some reason Nov 06 17:44:25 rp: ouch Nov 06 17:44:45 the dependency is invalid.. Nov 06 17:44:57 dependencies must be absolute paths (assuming you are talking package dependencies) Nov 06 17:45:16 also that value looks to be internal to some build tree which is bad as well Nov 06 17:45:48 RP: that brings back memories of busted rpaths :-) Nov 06 17:45:54 fray: yes, something completely broke the configuration Nov 06 17:46:16 fray: I have a choice of three autoconf variables which did it... Nov 06 17:46:24 * RP blacklists them all for now Nov 06 17:51:37 rburton: http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t2&id=a38adc04bbd687bb8d577069ed3b0a36abb2e77f actually builds core-image-sato Nov 06 17:51:47 ;) Nov 06 17:52:16 rp: with no difference to the output? :) Nov 06 17:52:35 rburton: that is still tbd. I do have buildhistory to check though Nov 06 17:53:45 "Can't install rpm-libs-5.4.9-r63.5@i586: no package provides /media/build1/poky/build/tmp/sysroots/qemux86/usr/lib/libbeecrypt.so" Nov 06 17:53:54 * RP ammeds to "nearly builds core-image-sato" Nov 06 17:54:10 I'm sure I fixed this, I just need to nuke the build as its corrupted now... Nov 06 17:54:43 ;) at least it looks more reasonable.. :) Nov 06 17:54:49 (even though it's wrong) Nov 06 17:54:49 Anyhow, this hack shows signs of something that may be vaguely more workable Nov 06 17:55:17 * RP cleanssstate's rpm Nov 06 17:55:41 and yes, its entirely understandable I broke sstate given what I'm doing here... Nov 06 17:56:08 gm Nov 06 17:56:15 morning Crofton Nov 06 17:56:29 working on the left coast sucks Nov 06 17:56:52 especially when I leave the laptop in the office and can't catch up on irc tell I get in Nov 06 17:57:28 Crofton: you say it like we say important stuff :-) Nov 06 17:58:20 you guys under estimate yourselves Nov 06 18:01:52 * RP notes the image does build now Nov 06 18:01:55 * RP -> afk Nov 06 18:07:55 bluelightning, Back online FYI. Nov 06 18:08:53 Stygia: ok, I sent you another mail in the mean time Nov 06 18:09:14 bluelightning, Ah, I need to reinstall my mail watcher! Nov 06 18:27:53 bluelightning, Probably not going to respond today, getting this system back up to spec will take a bit. Nov 06 18:28:15 Stygia: ok, I'll leave it in your capable hands Nov 06 18:30:24 bluelightning, I'm not sure if I should be flattered or suspect sarcasm. Nov 06 18:33:14 Stygia: no sarcasm intended Nov 06 18:33:21 * bluelightning heads out, bbl Nov 06 18:34:26 bluelightning, is a nice guy, unlike most of us :) Nov 06 18:36:30 Yes, I know. :) Nov 06 18:36:32 Ah, he left Nov 06 18:36:36 But yes, I'm aware. Nov 06 18:36:42 I'm just traumatized by #perl Nov 06 18:37:14 heh Nov 06 18:42:15 hi crofton Nov 06 22:43:15 anyone know why my image seems to want orc 0.4.18, not the 0.4.16 we have in meta-oe? Nov 06 22:45:33 Crofton: 0.4.18 is in meta-oe master Nov 06 22:45:41 yeah Nov 06 22:45:43 Crofton: at a guess you switched branches recently Nov 06 22:45:45 I was just realizing Nov 06 22:45:50 that exact thing Nov 06 22:46:12 I was looking at something a freind sent me for meta-sdr, but he is working off dylan I think Nov 06 22:46:33 It might be nice to young, so you can be clever fast Nov 06 22:46:58 if I lived in CA, I would quit work and sit in the sun all day Nov 06 22:47:19 I wouldn't say clever, just experienced... Nov 06 22:47:26 well, experienced in this specific problem Nov 06 22:47:29 * Crofton guesses you did that recently :) Nov 06 22:47:53 the move orc to oe-core thread had me thinking the rev was right Nov 06 22:48:04 and started me to look at the version I have Nov 06 22:48:27 ok, time to sort out the mess Nov 06 22:49:21 khem`, ping Nov 06 22:50:52 bluelightning, send in the FOSDEM table app Nov 06 23:00:50 Crofton: right, will do so now Nov 06 23:00:56 thanks Nov 06 23:04:48 Crofton: just posted patches to move orc to oe core Nov 06 23:05:06 awesome Nov 06 23:05:55 btw I noticed some tar 1.27 fixes werent added to the dora branch Nov 06 23:06:10 without these fixes, it is not possible to use oe-core on hosts with this tar version Nov 06 23:08:33 Crofton: done Nov 06 23:08:49 awesome Nov 06 23:12:28 dv_: are you sure? at least 2 fixes are there Nov 06 23:12:31 http://git.openembedded.org/openembedded-core/commit/?h=dora&id=3d5a6d0a480a0fa98260a3b3ffc71b8d9e3e58af Nov 06 23:12:36 http://git.openembedded.org/openembedded-core/commit/?h=dora&id=69c26e795c117aabfaf313abbfd10e70ede633d9 Nov 06 23:12:43 it builds fine here with tar-1.27 Nov 06 23:14:11 oh. nevermind, I misread the branch position Nov 06 23:14:25 it looked as if dora and master were split apart Nov 06 23:19:34 also, image creation time is a huge issue Nov 06 23:19:39 it should be faster! Nov 06 23:21:44 Crofton: +1... Nov 06 23:22:01 ndec, I am trolling sort off Nov 06 23:22:08 me too... Nov 06 23:22:15 I know that is a hard operation to speed up Nov 06 23:26:09 I'm trying to root the Roku box. Anyone interested? Nov 07 01:11:00 I think I will write a script to create a cmake toolchain file from the sdk env vars Nov 07 01:11:07 by reading the adt source Nov 07 01:11:14 WHAT HAS MY LIFE COME TO? Nov 07 01:11:35 now I think I will to go central computer and buy some odds and ends **** ENDING LOGGING AT Thu Nov 07 02:59:58 2013