**** BEGIN LOGGING AT Tue May 06 02:59:58 2014 May 06 06:33:20 morning May 06 07:43:00 good morning May 06 07:44:04 moin May 06 07:44:27 it is after midnight, after all... May 06 08:05:37 nerdboy: http://www.total-knowledge.com/~ilya/mips/ugt.html May 06 08:11:18 yeah, there's been various schools of thought on that topic... May 06 08:12:18 maybe it's best to avoid words that imply a certain time of day May 06 08:14:50 i think people just do what they want May 06 08:15:08 with only the occasional timezone outbreak May 06 08:18:39 developers also tend to follow the sun less than most people May 06 08:55:17 morning all May 06 09:56:41 JaMa: is there a meta-luna layer already? May 06 10:29:20 Hi, how do I force the build of a recipe? Usually I did a "bitbake -f -b RECIPENAME" after doing changes to the source code, which worked fine. But for some reason this isn't working anymore. May 06 10:31:27 changes to the source code in the work directory? bitbake -C compile RECIPE May 06 10:32:02 -C is awesome May 06 10:33:26 rburton: ah ok, I've seen the -C option, but i mssed the compile. Thanks! May 06 11:58:52 Hi, I'm trying to use Yocto Autobuilder to create OE builds May 06 12:00:04 However, somehow my slave misses some layers. I have the layers in my config (under repos), they appear in the slaves source directory, but they don't appear in the slaves build directory May 06 12:00:51 Which step is responsible to copy the layers from slaves source to slaves build directory? May 06 12:50:25 The BitBake manual says "Layers allow you to isolate different types of customizations from each other." That's fine. But I'm having trouble understanding what a layer *is*, not does. Is it a single .bb file? A directory of them? May 06 12:52:29 osse: http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#metadata-machine-configuration-and-policy-configuration May 06 12:53:24 simply put: a directory following a certain layout containing recipes, classes and configuration files May 06 12:57:46 erbo: thank you! May 06 13:10:46 osse: layers are normally called meta-something, so meta-networking, meta-multimedia, meta-ti are all separate layers May 06 13:11:25 mind: meta-queastions is *not* a OE layer. May 06 13:12:41 rburton: thanks. I'm just getting in to this. There are so many new words to learn :p May 06 13:58:02 rburton: not yet May 06 13:58:25 rburton: only meta-webos-ports May 06 13:59:53 JaMa: i missed the discussion about luna at the oe meeting, was anyone taking minutes? May 06 14:08:03 I get the following error while running bitbake core-image.minimal: http://sprunge.us/HCAB - The git clone command fails, and sure enough the dir it complains about contains nothing but a lock file. I assume one of the previous commands failed/did the wrong thing but I don't know where to look. May 06 14:08:38 (Also: I'm not sure if this is the right place for this kind of questions. Feel free to tell me to shut up) May 06 14:24:22 rburton: it was mostly short demo, you can find link to image in this channel, my plan now is to strip it from distro-specific bits and provide it as alternative image May 06 14:33:47 osse: try bitbake -c cleansstate virtual/kernel && bitbake virtual/kernel May 06 14:36:21 mckoan: That first one ran without errors. The second one is still running and at the do_fetch step for linux-imx-... with some 200 tasks remaining. May 06 14:36:27 That seems to be a good thing :) May 06 14:37:17 mckoan: I suppose 'bitmake -c cleanstate' is equivalent (in spirit) to 'make clean'? May 06 14:38:30 Btw, thank you! May 06 14:39:58 osse: cleansstate - double s May 06 14:40:26 cleansstate will delete the local build directory and any relevent sstate cache files May 06 14:41:15 which ensures that you'll do a build when you next run bitbake as -c clean will only remove the local build but not anything in sstate May 06 14:41:52 Ahh, ok. I didn't make that typo when I ran the command, by the way. May 06 14:46:50 moin May 06 14:47:26 moin May 06 14:47:34 it is after midnight, after all... May 06 14:47:46 ar ar May 06 14:48:22 it's even kinda early on the west coast May 06 14:48:36 * nerdboy 's wife dislikes getting up early May 06 14:49:26 Here on the west coast it's time to go home from work. Most likely a different west coast May 06 14:50:09 highly probable even May 06 14:52:43 Hello. How does one choose the root filesystem type? I would like to use squashfs. I am close to getting my image to build with the read-only root FS feature (I think). I have enabled support for squashfs in the kernel as well. What else is required? May 06 14:52:56 * nerdboy can hopefully start his daisy build today May 06 14:53:17 bryan: see IMAGE_FSTYPES May 06 14:53:50 * nerdboy considers putting a stumbling block in front of kergoth May 06 14:54:27 i didn't get my remote shell up... May 06 14:55:17 time to run perl-cleaner... May 06 14:56:44 Thank you kergoth. I see it now in the documentation. This variable is not specified in my local.conf file. Where is it coming from? May 06 14:57:21 Or is it taking on a default value since it is not specified? May 06 14:57:46 YPTM: Participant passcode: 42001078 Dial-in number: 1.972.995.7777 US Toll Free number: 1.877.561.6828 BlackBerry users, click this link to join your conference as a participant: 1.972.995.7777x42001078# May 06 14:59:06 YPTM: Beth is on. May 06 14:59:25 Yocto Project Technical Meeting? May 06 14:59:35 * nerdboy is the nosy new guy May 06 14:59:45 Yes that is what YPTM is. :) May 06 15:00:24 YPTM: Armin is on May 06 15:00:41 YPTM: Mark H is here.. May 06 15:00:46 YPTM: Matthew is on May 06 15:01:01 YPTM: Tom Z on May 06 15:01:06 * nerdboy was on a business meeting until almost 1 am last night May 06 15:01:35 YPT:M: Ross joined. May 06 15:01:45 ain't irc wonderful? May 06 15:01:58 * nerdboy puts a sock in it May 06 15:02:17 YTPM: maxin here May 06 15:02:28 YPTM: Saul here May 06 15:02:39 YPTM Michael on the call May 06 15:03:20 YPTM: Darren has joined May 06 15:03:54 YPTM: Paul Eggleton joined May 06 15:04:03 * fray doesn't know offhand.. May 06 15:04:04 YPTM: nitin joined the bridge May 06 15:04:08 YPTM: AlexG on chat only May 06 15:04:09 YPTM: Richard joined May 06 15:04:57 YPTM: Sean is dialing in May 06 15:05:17 sjolley: AB downtime on Friday. May 06 15:06:46 Alex D is in May 06 15:07:34 * fray gets the question now.. I must have missed something the first time May 06 15:07:51 * darknighte missed the question, but heard some of the answer May 06 15:08:02 fray: dvhart: what was the question about tune files? May 06 15:09:01 he was wondering about adding support new new bsps,cores that don't previously exist.. May 06 15:09:14 if you put them in an OSV layer, BSP layer, etc.. then they are not generally useful.. May 06 15:09:24 he searches the layer index and found a reference in meta-mentor... May 06 15:09:36 but meta-mentor is just overriding tunes and not adding new ones AFAIK May 06 15:10:22 hmmmm. I will have to check with kergoth on that. May 06 15:10:45 for AMD, I went ahead an added the tune file into meta-amd, which seems appropriate to me. May 06 15:10:48 darknighte no, I know there are overrides as part of supporting the sourcery toolchain May 06 15:11:08 darknighte, my answer was add the tune files to where the toolchain is.. or where it's generally applicable.. May 06 15:11:19 * darknighte nods May 06 15:11:21 k May 06 15:11:27 starting with the BSP layer is a logical place, but once "other" people start to use the code, then move it to the core.. May 06 15:11:45 but if the core will never compile/optimize for that core/soc, then it makes no sense to be in the core May 06 15:13:27 https://wiki.yoctoproject.org/wiki/Stable_branch_maintenance May 06 15:15:20 * nerdboy new guy raises his hand May 06 15:15:54 fray: OEDAM == OpenEmbedded Developers - America Meeting May 06 15:16:05 for the curious: http://openembedded.org/wiki/OEDAM May 06 15:16:17 Jefro: any idea when you'll have your notes posted? May 06 15:16:43 didn't I say America? May 06 15:16:53 i I didn't, I meant to May 06 15:16:58 heh. you said architecture. May 06 15:17:05 ya, not what I meant to say.. May 06 15:18:13 darknighte working on minutes today May 06 15:18:28 (btw I'm not in the call, telecom issues at the home office) May 06 15:19:24 Jefro: thx. we should make sure to attach it to to the OEDAM page on the OE wiki May 06 15:19:47 as an aside, gentoo arm stages are armvXX-specific and either hard or softfloat, then users are expected to modify their base CFLAGS to be what the want, tuning, specific optimizations, etc May 06 15:21:17 i had to tweak it to work, but rpi bsp adds the potential to tune for arm1176jzfs May 06 15:21:45 nerdboy: We support different tunes too... May 06 15:21:51 just so i'm clear, are talking about somethgin different? May 06 15:21:54 YPTM is done. May 06 15:22:28 RP: yes, that's one of the things that needed fixing in meta-rpi May 06 15:22:44 * darknighte waves to nerdboy May 06 15:23:05 what i just said is essentially the same thing in both and it makes sense to me May 06 15:23:12 that's all May 06 15:23:25 nerdboy: that is indeed the same thing May 06 15:24:04 if it's changing, i guess i'd like to know more? May 06 15:24:39 nerdboy: no real changes that I know of May 06 15:24:42 * nerdboy thinks he might still be unclear May 06 15:24:49 oh May 06 15:26:10 nerdboy: we're not rebuilding gcc any more for every single tune, we can get away with doing it once per architecture May 06 15:26:24 nm... <= best rosanne rosannadanna voice May 06 15:26:27 nerdboy: the things like libc, libgcc and so on *are* still compiled per tune May 06 15:26:58 okay, then i remember that one coming up May 06 15:27:24 was that just to ratify policy? May 06 15:27:46 * nerdboy still playing anthropologist in yocto-land May 06 15:29:26 obviously i'm that guy that sits in front of the class and doesn't stay quiet May 06 15:29:48 hopefully in a good way... May 06 15:32:01 RP: what's the best way to pare down an emitted sstate-cache to just what's necessary to build? e.g. I know unless I intend to change package managers, it should be safe to drop do_package. I can of course create a second build dir that pulls from the first's sstate-cache as an sstate mirror, and use the sstate cache it produces from that as a subset, but I'm not sure that's ideal May 06 15:33:12 anyway, if it matters, my grumpy toolchain side likes that idea May 06 15:33:26 kergoth, not the 'best' solution, but I've set my sstate directory as a 'mirror'.. then created a new blank sstate dir.. May 06 15:33:40 do the build, the items are mirrored as they are used.. and I only have what was necessary to run that one build May 06 15:33:42 okay, that's what i just described May 06 15:33:56 * nerdboy helps george with the gnat-gcc toolchain May 06 15:34:03 sorry.. I missed that (doing 5 things at once this morning.. again) May 06 15:34:13 :) no worries, good to know that that's indeed a valid approach May 06 15:34:29 we're distributing sstate, and the thing gets massive, so anything we can do to get it down to a bare minimum would be useful May 06 15:34:33 i'll use that approach for now May 06 15:34:50 i tried just rm'ing the do_package archives, but that made bitbake awfully upset, it started rebuilding things and getting setscene errors May 06 15:35:04 hence switching to letting bitbake grab what it needs instead May 06 15:35:37 RP: do we have a way to tell bitbake to fetch sstate and then exit after that, or an event that's fired at that point, after calling the sstate check hooks, but before running either runqueue? May 06 15:35:39 there was a script that would let you prune it.. but I've not used it May 06 15:35:40 * nerdboy wonders if scanelf/ldd could help with that... May 06 15:36:02 sounds a bit like what revdep-rebuild does May 06 15:36:02 RP: i'll dig, but figured i'd see if you knew off the top of your head May 06 15:36:29 the "paring-down" part i mean May 06 15:36:52 no, that wouldn't be of use. we're interested in what bitbake requires, not what the binaries do May 06 15:37:13 yeah, still early for me... May 06 15:42:02 but it's an interesting mix of bash and python... May 06 15:42:26 * nerdboy likes interesting implementations May 06 16:31:03 denix: in case i didn't mention it last week, it was really cool to finally meet you irl May 06 16:37:21 Hello. I want to inject some files into the filesystem before actually building the rootFS image. How can I do this? May 06 16:41:39 bryan: before do_rootfs? maybe a prepend or even a base-files sort of recipe May 06 16:41:53 there are hooks for that sort of thing too in the rootfs creation process May 06 16:41:54 depends on what you need May 06 16:41:59 e.g. ROOTFS_POSTPROCESS_COMMAND May 06 16:42:16 for generic shell command injection May 06 16:42:24 that would be post, but is there a pre? May 06 16:42:57 * nerdboy has used too much custom ROOTFS_POSTPROCESS in the older stuff May 06 16:43:02 yep, there is May 06 16:43:28 i broke the old lab dir with the package/file info :/ May 06 16:43:37 I'd like to see a generic recipe that just grabs all the files from a directory the user specifies to shove into / in the rootfs May 06 16:43:51 Thank you both. Is there documentation somewhere on these hooks? May 06 16:43:54 its a common request May 06 16:43:56 probably May 06 16:44:06 there's a variable reference in the yocto documentation, i think May 06 16:44:15 * nerdboy has made bin packages too May 06 16:44:17 * kergoth has never actually read most of the docs :) May 06 16:44:34 which is fairly clean and tracked by package manager May 06 16:45:13 there are one or two examples in the main layers May 06 16:45:34 *examples of custom ROOTFS_POSTPROCESS stuff May 06 16:45:42 at least one anyway May 06 16:46:10 Ok. "ROOTFS_POSTPROCESS_COMMAND" actually does appear in the ref manual. When precisely would these command be executed? May 06 16:46:42 after the packages are installed, but before the images are created from the rootfs May 06 16:46:44 Is it right before the root filesystem is archived or compressed? Or after? May 06 16:46:50 I see May 06 16:46:51 iirc IMAGE_ROOTFS is the path to the root of the filesystem May 06 16:46:59 after would be the image postprocess command May 06 16:47:05 this is the rootfs postprocess command, so it's before the image bits May 06 16:47:19 Excellent. This is exactly what I wanted. Thanks! May 06 16:47:47 np May 06 16:48:14 another happy customer... May 06 16:59:02 In my next life I become an embedded engineer. Somehow every job I take has nothing to do with embedded but I end up in it May 06 17:04:59 apparently Belen is learning to solder May 06 17:09:18 do you smell burning flux or something? May 06 17:09:22 beware of the designer with a soldering iron May 06 17:09:57 * nerdboy waits to see what she comes up with May 06 17:23:45 khem: I just filed bug #6270 to help coordinate the fsl-ppc issue, can you comment on your progress in the bug and I will ping the FSL guys to look at this today also. May 06 17:23:46 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=6270 major, Medium+, 1.7 M1, raj.khem, NEW , [GCC-4.9] libgcc fails to build with ICE Issue with fsl-ppc May 06 17:24:55 sgw_: thanks assign it to me May 06 17:26:20 khem: already done, thanks for staying on top of it. May 06 17:26:31 no worries May 06 17:26:47 I am working with gcc community as well to get this fixed May 06 17:28:45 I have been using the instructions provided by Gumstix to make an SD card to boot from. Now I have my system set up to make a squashfs filesystem and I don't actually know how to put everything onto an SD card now. How doy others do it? May 06 17:28:59 *do* May 06 17:35:46 bryan: there are various scripts to create SDK cards May 06 17:36:18 wic seems promising, I wonder what it'd take to make it happier for non-x86 setups. currently some of its bits seem rather specific to bootimg/syslinux/etc May 06 17:48:07 khem: What are they? Where can I find info on them? May 06 17:56:59 Any objections to https://github.com/MentorEmbedded/meta-mentor/commit/efc648cd going into bitbake.conf? May 06 17:58:15 no worries here.. May 06 17:59:03 kergoth: make sense, should add the Savannah ones when they get added May 06 17:59:34 k May 06 18:00:50 khem: what did you add that could cause gcc 4.8 to fall over with your patches included, but not enabled May 06 18:00:55 tempting to say we should change the vardepvalue to be basename+checksum pairs, not the actual urls, since as long as the checksums match, where it came from is irrelevent May 06 18:21:32 hmm May 06 18:22:52 RP: do you know offhand if the dep scanning code picks up on variable references done through the python variable name lookup? e.g. FOO = "${@BAR + 'meh'}" — does it recognize that BAR is a dependency? I'm guessing it doesn't, and until it does, we should avoid referencing variables in that way. May 06 18:27:02 ah, yes, looks like it doesn't from my limited testing May 06 18:27:50 khem: ping May 06 18:28:43 that guy's blog from yesterday didn't mention being locked up like a veal cutlet May 06 18:28:59 * nerdboy should email him that one May 06 18:34:37 sgw_: yes May 06 18:35:07 sgw_: whats the failure May 06 18:35:32 khem: not sure if this is you or something in the whole gcc refactor, but we are falling over in master with 4.9 visible, but not enabled (GCCVERSION=4.8% May 06 18:36:05 nerdboy: thanks. it was nice meeting you too! May 06 18:37:08 sgw_: ok, point me to error May 06 18:37:30 same for khem of course... May 06 18:37:45 Hmm, I need to find the time to finish breaking up external-sourcery-toolchain one of these days. keeps falling onto/off the back burner May 06 18:37:51 khem: any of the AB failures listed before this May 06 18:38:09 it was mainly one person missing from my list... May 06 18:38:27 It looks like gcc-cross-initial and gcc-cross are getting 4.9, but libgcc is getting 4.8 May 06 18:38:30 kergoth May 06 18:40:45 sgw_: I sent a patch May 06 18:40:58 to OE-Core ml this morning May 06 18:41:04 or last night May 06 18:41:07 take that patch May 06 18:41:34 http://patches.openembedded.org/patch/71573/ May 06 18:41:43 that should hopefully fix it May 06 18:42:06 khem clearly I missed that one today! May 06 18:42:10 khem thanks May 06 18:51:09 I have a question which is not strictly a Yocto question, but still very relevant. If I am putting a squashfs filesystem on an SD card, what partition id code should be used for that partition? May 06 18:51:34 As in sfdisk. May 06 18:54:52 bryan, it probably doesn't matter.. May 06 18:55:00 I know a lot of people just use ext2 as their default.. May 06 18:55:19 K. Thanks. May 06 18:56:07 fyi, http://community.validation.linaro.org/scheduler/alljobs?search=Open&length=25#table May 06 18:57:10 http://community.validation.linaro.org/dashboard/image-charts/OpenEmbedded-CURRENT-bbb May 06 19:00:04 anyone know why the license_flags = "commercial" applies to gst-openmaxx and libomxil? May 06 19:00:50 openmax is still TI i believe, not sure the exact license May 06 19:01:05 denix should know... May 06 19:01:10 looks like LGPLv2 afaict May 06 19:01:31 denix: do you know what the license is for openmax? May 06 19:03:13 The You're Doomed License May 06 19:03:35 hehehe! cool, does that mean I can run doom on it? May 06 19:04:17 you get to lose your sanity to it May 06 19:07:54 i've seen doom on a pi... May 06 19:08:34 * nerdboy always liked the double-barreled shotgun and the BFG May 06 19:29:41 darknighte: OpenMAX is a standard by Khronos - http://www.khronos.org/openmax/ May 06 19:30:06 OpenMAX IL is an integration layer - http://omxil.sourceforge.net/ May 06 19:30:38 gst-openmax is a GStreamer plugin to talk to OpenMAX - http://freedesktop.org/wiki/GstOpenMAX/ May 06 19:31:24 all components are OpenSource, but hardware acceleration on TI platforms uses DSP and DSP codecs are usually commercial May 06 19:36:24 denix: any idea why openmaxil would be marked "commercial"? May 06 19:39:35 seems like the interface library being lgplv2 would "isolate" the commercial plugin bits? May 06 19:48:04 darknighte: ok, had to dig really deep... :) it's a cruft left from old COMMERCIAL_LICENSE global mechanism. originally, only proper commercial packages were marked, but then also packages that depend on those commercial ones were marked as well, so the "make world" would work... :) May 06 19:49:30 darknighte: and then global COMMERCIAL_LICENSE list was replaced with per-recipe LICENSE_FLAGS, so those commercial flags went into corresponding recipes, even those that aren't really commercial... May 06 19:50:33 hrmmm, so it sounds like I should be able to drop the commercial designation for those packages then? May 06 19:51:22 perhaps let me come at it from another direction. May 06 19:51:30 let me take it to the list... May 06 19:51:55 how does adding these flags enable "make world" to work? May 06 19:52:19 it would skip them, since they depend on real commercial packages May 06 19:52:31 I built an image with the "read-only-rootfs" feature. However, my root filesystem is still read/write. Is there something else I need to do? May 06 19:53:16 denix: hmmm, that would only apply if you actually attach a commercial plugin though, correct? May 06 19:53:26 the dependency, I mean. May 06 19:53:46 huh? May 06 19:53:48 bryan, the read-only-rootfs makes the system work when you BOOT with a readonly rootfs.. May 06 19:54:03 if you still are booting w/ 'RW', or your fstab remounts / to 'rw'.. we're not going to stop it May 06 19:54:13 denix: trying to re-word May 06 19:54:22 be sure 'ro' is passed to the kernel, and make sure you arn't remounts / 'rw' somehow.. May 06 19:54:36 openmaxil does not *always* depend on commercial packages. May 06 19:54:45 then rootfs is RO, and things are expected to work as the other filesystem elements are bind mounted to allow RW (via a tmpfs) in the correct places May 06 19:54:55 I see. To actually boot it as read-only, do I just alter fstab, or does it need to be done elsewhere? May 06 19:55:09 start w/ your bootloader, pass 'ro' to the kernel May 06 19:55:21 denix: specifically, in my case, these are no commercial bits being plugged in May 06 19:56:18 OK. How do I tell my bootloader to pass this to the kernel? May 06 19:56:39 which bootloader, which board, which kernel? May 06 19:57:44 denix is good at the rewording/precise terminology thing May 06 19:57:59 denix: I'll send a note to the oe-core list asking about it. May 06 19:58:07 U-boot, Gumstix DuOvero COM on a custom baseboard, Kernel is 3.6.0 May 06 19:58:17 darknighte: almost finish with the mail myself May 06 19:58:34 darknighte: the fact that you don't pull those in is because so much has changed... May 06 19:58:59 nerdboy: heh, giving me a hard time? :) May 06 19:59:24 seems to be part of nerdboy's personality ;) May 06 19:59:28 not particularly... May 06 19:59:31 it's a setenv, I just don't know which one May 06 20:00:04 ok. May 06 20:00:07 been schooled a couple times May 06 20:00:11 here we goo.. looks like "'bootargs" May 06 20:00:23 so try "print bootargs" May 06 20:00:34 then setenv bootargs ro May 06 20:00:57 if that doesn't work, you'll haev to find a u-boot/gumstix expert.. (I'm neither) May 06 20:01:08 darknighte: ok, email is out, please chime in May 06 20:01:32 denix: thx. I will. May 06 20:02:28 * darknighte waits patiently for email to arrive from list server May 06 20:02:45 sorry, denix no, DarkKnight duh... May 06 20:02:49 Do you mean try it on the system? How does one get a command prompt for the bootloader? May 06 20:03:12 heh, I thought you might mean me nerdboy... May 06 20:03:42 you're kidding, right? May 06 20:04:23 bryan, no idea.. I'd assume when you turn it on, the firmware drops into a shell where you confgiure the boot May 06 20:04:31 darknighte: who's DarkKnight and why is he using your personality? :) May 06 20:04:43 if not, I don't know what you hve to do to signal it.. you'll need to look at the uboot and gumstrix info for that May 06 20:04:58 Ok. Thank you for the tips :-) May 06 20:05:04 no problem.. May 06 20:05:14 denix: my *other*, alter ego. May 06 20:05:26 199 root rt 0 0 0 0 S 100.0 0.0 6:32.74 migration/38 May 06 20:05:27 the racing one? May 06 20:05:36 anyone every seen a problem where the migration tasks take so much CPU time? May 06 20:05:37 yep. I only use it for special occasions! May 06 20:05:39 bug in ps maybe? May 06 20:05:42 * DarkKnight is yet another yocto user and IRC idler May 06 20:06:10 :) May 06 20:06:11 also I'm wondering why I get highlighted and "msged" ;) May 06 20:06:30 hah, I should have checked to see if there actually was another handle in use! May 06 20:06:38 nice name DarkKnight ! May 06 20:06:39 tab complete and small type... May 06 20:06:51 no camel case here, bub! May 06 20:06:58 no problem May 06 20:07:34 good May 06 20:07:46 * nerdboy prfers underscores May 06 20:07:48 so, I guess I'll go easy on nerdboy for calling me DarkKnight after all May 06 20:07:57 heh :) May 06 20:08:21 it feels like list serv is hosted on mars or even further... May 06 20:08:28 * darknighte shudders at nerdboy's bad taste May 06 20:08:32 denix: nj! May 06 20:09:03 readability man May 06 20:10:15 remember those missing probes and exploding ariane V's ? May 06 20:10:29 * darknighte gives up waiting on denix "supossed" email and flips back to an editor May 06 20:10:29 they couldn't read the code... May 06 20:11:37 https://www.youtube.com/watch?v=kYUrqdUyEpI May 06 20:12:16 darknighte: check your mailbox May 06 20:12:36 think about that when you're writing code once in a while... May 06 20:12:48 "boom" May 06 20:14:59 denix: ? I don't see it. May 06 20:15:18 mentor one? May 06 20:15:22 nerdboy: I agree that code readability is important. May 06 20:15:36 obfuscate, obfuscate! :) May 06 20:16:18 denix: hmm, strange. I see it in my work account, but not my personal one. May 06 20:16:21 * darknighte shrugs May 06 20:17:23 you shrug too much... May 06 20:17:31 style is somewhat orthoganl May 06 20:19:20 * darknighte shrugs May 06 20:19:33 * darknighte shrugs again May 06 20:19:57 it's something most people don't think about much, at least from what i can see... May 06 20:20:38 nerdboy: I read an article some time ago extolling the virtues of an automatically enforced coding standard for that reason. May 06 20:21:08 surprisingly few think about comments and other stuff, even in a "professional" setting May 06 20:21:21 YMMV, as always, but I have found that in large groups, it helps to have fairly uniform code for maintainability May 06 20:22:07 stringent customers generally like tractability May 06 20:22:10 consistent styles help in two things - reviewers can concentrate on code instead of style which is a distraction May 06 20:22:10 Use_ALL_Styles_allTHEtime! May 06 20:22:17 * nerdboy included May 06 20:22:22 secondly you can write some post processing tools May 06 20:22:28 if styles are consistent May 06 20:22:29 gnu indent May 06 20:22:55 doesnt matter which one .. pick one May 06 20:23:01 is always my advise May 06 20:23:11 CamelCase May 06 20:23:27 i'me in violent agreement with the "shared" style, workflow, etc May 06 20:23:28 mix of different styles is a style on its own - see above May 06 20:23:40 just pick one and play nice together May 06 20:23:51 then we all get beer and cookies May 06 20:24:02 for (loop_index_variable = 0; loop_index_variable < 10; loop_index_variable++) ..... May 06 20:24:14 and some veggie stuff i suppose ;) May 06 20:24:37 I am agnostic to whatever it is May 06 20:24:59 I need to use that line sometime May 06 20:25:38 the (loop_index_variable? or the beer and cookies? May 06 20:26:16 Crofton: #define post_increment "++" ;) May 06 20:26:20 * nerdboy shrugs like darknighte May 06 20:28:05 i think most of my public examples are a mix of clean and brutal May 06 20:28:19 khem, epic May 06 22:13:32 khem: I need some help wrt -march=mips32r2 May 06 22:16:01 * nerdboy scrapes out his ear May 06 22:16:11 did i just hear that right? May 06 22:16:25 climate change discussion in g+ hangouts? May 06 22:17:06 *climate change impacts discussion May 06 23:28:50 ant_home: ok May 06 23:29:01 dont ask to ask :) just ask May 06 23:36:49 RP: I see that you left out http://git.openembedded.org/openembedded-core-contrib/commit/?h=kraj/gcc-4.9&id=cedeee72f8263e7397f8481501e0b516daae23aa May 06 23:37:02 from gcc-4.9 branch May 06 23:37:12 arent you seeing the issue ? May 06 23:37:24 on x86 ? May 06 23:40:03 RP said earlier there is some stuff in master-next that isn't merged yet.. May 06 23:41:56 ok May 06 23:42:24 having said that I think we should enable sse2 as default for i586 tunes May 06 23:42:46 it doesnt make sense to target _very_ old chips May 06 23:42:50 not for i586... there are still a bunch of i586 based systems that don't have any sse or MMX for that matter May 06 23:42:53 as defaults May 06 23:43:00 here are branch new chips that don't have sse or MMX May 06 23:43:03 really May 06 23:43:16 Intel Quark, whats on the Galileo board.. May 06 23:43:23 it's effectively a 486 w/ 586 instructions.. May 06 23:43:43 target marketing is 32-bit micro controller.. May 06 23:43:48 oh pentium4 resurrected :) I wonder if division errata was fixed May 06 23:44:13 i686 can have MMX.. it's the ones above that that are sse2 I believe.. May 06 23:44:28 thats why 'core2' is currently the generic CPU of choice.. it's 'old' but has modern features.. May 06 23:45:41 I have written a bitbake file for a library I would like to use, but I am having trouble getting it to work. Here is the file https://gist.github.com/russelhowe/1d8516953f6f46a4234e May 06 23:45:44 oh well then probably gstreamer is not a package for those uController either May 06 23:46:00 I'm guessing thats not so much a problem.. ;) May 06 23:46:22 I expect most usages are going to be IoT devices, embedded micro controllers, data aquisition, etc... May 06 23:46:36 russelh: whats the issue ? seems nice little recipe May 06 23:46:49 fray: yes I see May 06 23:46:53 It produces the packages bfl-dev and bfl-dbg, but no package named bfl. This means I can DEPEND on it in another bitbake file, but when it comes time to install, the dependency on bfl is not satisfied. May 06 23:47:01 fray: don't count out video displays, they might be in the mix May 06 23:47:23 russelh: does it generate a .so.X May 06 23:47:46 if not may be you have to ALLOW_EMPTY_${PN} = "1" May 06 23:49:51 khem: I thought it did, it does when I build it outside bitbake, let me check May 06 23:58:00 khem: yes, it does, in /usr/lib May 06 23:58:23 whats the name of library ? May 06 23:58:40 The library is liborocos-bfl.so May 06 23:58:54 yeah thats the problem May 06 23:58:55 * nerdboy busting a spleen May 06 23:59:02 again... May 06 23:59:05 .so is shoveled into -dev May 06 23:59:15 by default May 06 23:59:30 since its thought to be a symlink or linker stub/script May 07 00:00:07 so you should first add it to FILES_${PN} += "${libdir}/*.so" May 07 00:00:21 Oh, I see. It needs to see a version string. May 07 00:00:35 Let me try that, that sounds good. May 07 00:00:39 and then also add INSANE_SKIP_${PN} += "dev-so" May 07 00:01:08 yeah better if you have name it with symbol versioning naming convention May 07 00:01:18 like libfoo.x.y.z May 07 00:02:54 If I add the version string, do I need to tell bitbake to add the symlinks or does it figure that out automatically? May 07 00:10:44 https://secure.flickr.com/photos/linuxfoundation/14104504813/in/set-72157644040774060/ May 07 00:12:02 russelh: well symlink is not needed per say but it should be created by your components build system May 07 00:12:10 you dont have to tell bitbake anything then May 07 00:12:46 OK, thanks for the help. May 07 00:27:41 i love this guy... => "So no, I'm not required to be able to lift objects weighing up to fifty pounds. I traded that for the opportunity to trim Satan's pubic hair while he dines out of my open skull so a few bits of the internet will continue to work for a few more days." May 07 01:09:30 * nerdboy enjoyed the little burst of "style" discussion earlier May 07 01:09:57 would be really cool with more time and pitchers of cold beer May 07 01:10:19 * nerdboy homers "mmm... beer..." **** ENDING LOGGING AT Wed May 07 02:59:58 2014