**** BEGIN LOGGING AT Wed Sep 19 02:59:58 2012 Sep 19 05:06:48 night Sep 19 06:48:50 hi Sep 19 08:47:19 morning all Sep 19 08:47:48 good morning Sep 19 08:49:30 'mornig Sep 19 08:55:19 hi guys Sep 19 08:55:31 | The directory that should contain system headers does not exist: Sep 19 08:55:31 | /home/hrw/HDD/devel/canonical/aarch64/build/tmp-eglibc/sysroots/x86_64-nativesdk-oesdk-linux/usr/include Sep 19 08:55:41 fresh tmpdir, bitbake nativesdk-eglibc Sep 19 08:56:18 /home/hrw/HDD/devel/canonical/aarch64/build/tmp-eglibc/sysroots/x86_64-nativesdk-oesdk-linux/usr/local/oecore-x86_64/sysroots/x86_64-oesdk-linux/usr/include exists Sep 19 08:58:02 ideas? Sep 19 08:58:19 directories populated from sstate files Sep 19 09:36:01 hi bluelightning, florian, all Sep 19 09:37:00 hi pb_ Sep 19 09:37:03 hey pb_ Sep 19 09:37:40 * florian wonders if its easier to make an OE based filesystem for this board instead of using ptxdist Sep 19 09:44:39 I would love to see someone with experience create a pro/con table of all the embedded build systems - they all seem much for much when I take a quick look at others Sep 19 09:45:20 I admit that this would be useful for quite some people Sep 19 09:46:30 florian: What are your reasons for using ptxdist, manufacturer supplied, built in board support? Sep 19 09:47:43 jackmitchell: The hardware manufacturer ships with a ptxdist based BSP. Sep 19 09:49:27 My first impression was good: A manual and a few archives only. But the reality is a little bit frustrating... Sep 19 09:49:50 interesting, is it a monolithic mix of patches and binaries or is it just a matter of porting the kernel? Sep 19 09:51:01 It feels a little bit like the first time using OE but with an inconsistent manual and suffering from outdated package descriptions which seem to be impossible to find in the BSP ;) Sep 19 09:52:07 :/ good luck! Sep 19 09:52:13 Its supposed to be a build system for whole filesytems with a kconfig based configuration tool and soem defined toolchains. Sep 19 09:54:10 I got a toolchain distribution package and tried to build a toolchain. After several fixes to the build script it started to build... and failed after building 12 toolchains for various architectures. Sep 19 09:57:26 florian: it sounds like you're in for a fun day! ;) Sep 19 09:59:48 jackmitchell: indeed... luckily the sales guy from the hardware manufacturer failed as well and therefore the customer knows that it is a tough job already Sep 19 10:05:43 jackmitchell: re comparisons, it would have to be done by someone independent; I think many of us OE folks would find it hard to be objective about other build systems Sep 19 10:06:20 I do look at some of the slightly newer ones and wonder why they started yet another system instead of getting involved with OE... Sep 19 10:07:44 yes, that tends to be my feeling too - I think people probably don't put enough time into trying others before getting frustrated and deciding they can do it better themselves Sep 19 10:08:31 or maybe from the pengutronix view, they do a lot of kernel/toolchain work so their system is probably made to suit their testing needs. Sep 19 10:12:34 nano roxx as a target ;) Sep 19 10:14:06 jackmitchell: hard to make such comparison - I used only OE for serious work Sep 19 10:16:02 The only other system I've seen with some serious traction is BuildRoot - whenever I've delved into that it seems to re-invent the wheel alot, it also seems as though it has some mega duplication of package issues Sep 19 10:16:24 from what I could gather, it has packages inside packages Sep 19 10:16:59 so dependancies of programs were packaged inside the program directories - it was all very alien :) Sep 19 10:21:33 hi all Sep 19 10:35:30 hi all Sep 19 10:36:17 if an .ipk is built by two recipes, is there a way to pick the non-default one (the one with older package-version) Sep 19 10:36:38 I tried PREFERRED_VERSION/PREFERRED_PROVIDER, but it didn't work Sep 19 10:38:35 I don't think there is. In fact, I think the older one will get deleted under those conditions unless it has a different PACKAGE_ARCH. Sep 19 10:38:45 See the oe-core list, there was some discussion of this fairly recently. Sep 19 10:40:11 pb_ the ipk file names have different version number, so they don't get over-written Sep 19 10:40:22 but the older version of the ipk file end up in morgue Sep 19 10:41:13 that is build/tmp/deploy/ipk/core2/morgue Sep 19 10:45:37 oh, are you using oe-legacy? Sep 19 10:45:54 I don't think oe-core has the morgue anymore. Sep 19 10:51:36 pb_: no, I am using oe-core Sep 19 10:52:18 jackmitchell: iirc OE was created when BuildRoot was not enough for OZ devs Sep 19 10:53:28 jackmitchell: but both changed a lot in last 8-9 years Sep 19 11:11:17 hrw: ah, interesting, thanks! Sep 19 11:16:21 did someone built meta-toolchain recently? Sep 19 11:19:11 hrw: I did in the last few days, but it failed when compiling my application with ggc crt* issues Sep 19 11:19:36 so it built, but was unusable for my purpose (still unsure why it won't work) Sep 19 11:19:53 whether it is the toolchain or my application is still to be decided ;) Sep 19 11:33:10 with oe-core.. how can I not package the the source files in a -dbg package? Sep 19 12:31:16 here is the base_contains syntax from the openembedded manual: ${@base_contains(’’, ’’, ’’, ’, d)} Sep 19 12:31:27 it documents everything ,except this mysterious "d" at the end Sep 19 12:31:32 what is it? why is it necessary? Sep 19 12:45:01 dv_: it's the data dictionary - it's what the function needs to be able to access and expand variable values Sep 19 13:00:23 Hi guys Sep 19 13:00:59 I get the following error when I try to run bitbake and i can't find how to fix it Sep 19 13:01:04 ERROR: Multiple .bb files are due to be built which each provide virtual/kernel Sep 19 13:03:25 DiKay: do you have PREFERRED_PROVIDER_virtual/kernel set in your configuration (usually machine configuration)? Sep 19 13:03:31 pb_: do you still remember update-alternatives? Sep 19 13:05:30 ho right, indeed Sep 19 13:05:46 with value "linux-yocto" Sep 19 13:22:25 When I set PREFERRED_PROVIDER_virtual/kernel="linux", how can I know which bb file associated to "linux" ? Sep 19 13:33:02 DiKay: it will be whichever recipe has PN = "linux"; then the version preference logic selects which version if there is more than one Sep 19 13:33:34 PREFERRED_VERSION_linux = "2.6.32.27" Sep 19 13:33:38 (PN by default being the part of the file name before the first underscore) Sep 19 13:33:38 could work Sep 19 13:34:20 unfortunately that doesn't since "linux" is in OVERRIDES (from TARGET_OS) Sep 19 13:35:38 thus that means that statement will apply to all recipes (!) Sep 19 13:37:03 I made my own linux-vanilla recipe so I could see what's going on Sep 19 13:37:52 (bitbake -g is helping me !) Sep 19 13:44:03 why have all the linux recipes vanished and there's only linux-yocto now ? Sep 19 13:44:43 they didn't, they moved to BSPs which still use them Sep 19 13:44:50 DiKay: and linux.inc is in meta-oe Sep 19 13:45:48 yep, seen it Sep 19 14:22:39 * pb_ stabs webkit-clutter Sep 19 14:34:35 * JaMa stabs segfaulting opkg Sep 19 14:36:39 JaMa: make sure its really dead and not crawling back in behind your back Sep 19 14:37:06 JaMa: Like putting some concrete on it and putting it into the river Sep 19 14:38:18 I feer I'll have to patch it and heal it instead :/ Sep 19 14:40:14 I did wonder if we should approach the OpenWRT folks and try to come up with a joint effort to maintain opkg since we are seemingly the only remaining users Sep 19 14:48:16 yeah, or come up with a joint effort to just burn opkg and invent something else that sucks a bit less. maybe even dpkg+apt is a sensible choice nowadays, I dunno. Sep 19 14:48:44 it probably is true that the computing landscape has changed a bit in the 10 years since i/opkg was first designed. Sep 19 14:50:06 * pb_ hands pwgen a working internet connection Sep 19 14:52:41 there was a murmur on the opkg mailing list a while back from the maintainer about a complete rewrite but there was nothing further about it since Sep 19 14:53:23 pb_: I'm not sure the design has ever been the serious issue, it's the diabolical implementation ;) Sep 19 14:54:37 yeah, agreed. I don't think the design is inherently broken, though I do wonder whether the world has changed in such a way that the opkg design is no longer necessary and some other package manager could be made to work. Sep 19 14:55:02 yocto is using rpm, right? presumably that does work for them, and although I would hesitate to recommend it for anyone else I guess it is a possibility. Sep 19 14:57:50 pb_: poky uses RPM by default yes Sep 19 14:58:13 it has its own issues, naturally... Sep 19 14:58:30 I'm not convinced a perfect package manager exists, they all have their problems :/ Sep 19 14:58:53 heh it looks like opkg should segfault everytime new package has 2 entries in Provides while old version had just 1 Sep 19 14:59:03 awesome... Sep 19 15:00:06 svn blame to see why that if which breaks it was added later for some reason Sep 19 15:01:16 bluelightning: Except of course :) Sep 19 15:01:43 xim: personally I have no favourite... Sep 19 15:04:51 bluelightning: What, you're not ready to accept as your saviour? Sep 19 15:05:32 http://code.google.com/p/opkg/source/diff?spec=svn277&r=277&format=side&path=/trunk/libopkg/pkg.c is the issue :/ Sep 19 15:05:36 xim: unless is some package manager that I've never seen/used before, no :) Sep 19 15:05:38 I guess people aren't as religious about package managers as editors, but still :) Sep 19 15:06:46 JaMa: hmm, 2009, so quite an old change... Sep 19 15:06:47 I like dpkg, but have never given anything else a serious chance. I guess there are plenty others who have it the other (obviously wrong :p) way around Sep 19 15:09:16 bluelightning: all would be easier with oe-core changes to opkg sent upstream I guess Sep 19 15:12:49 JaMa: I agree, we need to do that... I think we did a round of upstreaming for opkg a while ago, I guess some more have accumulated since then Sep 19 15:14:18 JaMa: re 3133, are you able to trigger that with less than the large number of layers you have enabled? Sep 19 15:14:55 (without in any way suggesting that that is a factor) Sep 19 15:15:24 bluelightning: that was only opkg-utils not opkg itself Sep 19 15:16:38 JaMa: hmm looks like you're right Sep 19 16:11:59 * pb_ stabs clutter Sep 19 16:12:08 who thought it was a good idea to have a variable called "bool"? Sep 19 16:25:52 oh good grief, webkit has its own private copy of eglplatform.h. what madness is this? Sep 19 16:26:38 pb_: Das. Ist. Spartaaaaaaa! Sep 19 16:58:40 bluelightning: re 3133 in bugzilla, sorry it took so long, slow server and I was multitasking with upstreaming/fixing opkg Sep 19 17:02:12 basic gcc problem, generated "hello world" a.out file to desktop, then when I type a.out in terminal , it says command not found. What up? Sep 19 17:06:56 type ./a.out Sep 19 17:08:09 Thanks, that worked , whats going on ? Sep 19 17:08:50 e_spin: learn basics Sep 19 17:09:10 this isn't right channel for that Sep 19 17:09:39 I'll be back! Sep 19 17:29:41 re Sep 19 17:30:18 re Sep 19 21:43:22 Is there any way to change where the build system grabs the kernel src from? Sep 19 21:43:47 Attempting to build webos but need to specify the rpi kernel as the patchs are not mainline yet Sep 19 21:58:01 if the recipe name is different, define PREFERRED_PROVIDER_virtual/kernel = "recipe-name" Sep 19 21:58:05 usually the machine defines this Sep 19 22:09:45 aaa801: you're aware of meta-raspberrypi ? Sep 19 22:10:06 aaa801: https://github.com/djwillis/meta-raspberrypi Sep 19 22:10:20 already has a specific kernel recipe Sep 19 22:11:35 damn. anyone who hasn't yet seen the_silver_searcher, *highly* recommend switching to it. it's like ack, which is much better than grep, but it's smoking fast compared to ack Sep 19 22:11:43 http://geoff.greer.fm/2011/12/27/the-silver-searcher-better-than-ack/ Sep 19 22:13:19 couldn't someone just fix the existing tool rather than writing a new one? Sep 19 22:13:40 (/grump) Sep 19 22:14:22 ack was written in perl Sep 19 22:14:30 so.. blech Sep 19 22:14:46 one less perl dependency is a Good Thing™ :) Sep 19 22:35:03 I wonder anyone worked on pythong 3.x recipes ? Sep 19 22:35:14 kergoth: did you play with it at all Sep 19 22:35:21 nope Sep 19 22:35:22 oooh.... kinky pythongs....... Sep 19 22:35:28 lol Sep 19 22:37:22 I wonder if it really makes sense to use 3.x Sep 19 22:37:32 but anyway I will try Sep 19 22:39:39 good luck Sep 19 22:41:42 kergoth: thanks I will need a lot of it Sep 19 22:41:59 actually I will try to make it coexist Sep 19 22:42:03 with 2.7 Sep 19 22:42:20 * khem typo's butbake Sep 20 00:32:16 good day ! Sep 20 00:32:49 hi Sep 20 00:32:53 has anybody here worked with the sierra development env? Sep 20 00:35:45 never heard of it, and it'd be off topic for this channel, unless it involves oe or yocto Sep 20 00:36:24 Trying to see what is the benefit of using their platform vs embedded linux .... Sep 20 00:36:28 google is not helpful Sep 20 00:37:09 Crofton - no one has done a comparison ... google doesn't know everything Sep 20 00:37:49 avvero: regardless, it's not on topic for this channel. this channel is not about embedded linux in general, it's about the openembedded project / build tools specifically. Sep 20 00:37:58 #edev or #elinux would probably be a better bet Sep 20 00:38:09 yep Sep 20 00:38:30 kergoth : thanks! I will try over there ... Sep 20 01:13:02 bluelightning: i deleted the kernel folder from webos and put the rpi recipies in Sep 20 01:13:09 will any aditional changes be needed Sep 20 01:13:20 rage he left.. Sep 20 02:43:05 gnight **** ENDING LOGGING AT Thu Sep 20 02:59:57 2012