**** BEGIN LOGGING AT Sat Jan 18 02:59:59 2014 Jan 18 04:29:46 one down, one to go... **** BEGIN LOGGING AT Sat Jan 18 09:59:03 2014 Jan 18 11:20:50 JaMa: hmm, having the packagegroups always rebuilding isn't ideal :/ Jan 18 11:22:58 agreed, that's why I was suggesting to use the same PACKAGE_ARCH for them as the packages included in them Jan 18 11:23:31 which also isn't ideal, but is somewhere in the middle Jan 18 11:24:25 rebuilding when the (r)dependencies change, but rebuilding just once per architecture Jan 18 11:31:34 JaMa: once per tune effectively Jan 18 11:32:04 it seems sad to have to force them all to that behaviour :( Jan 18 11:40:12 agreed, but at least they are "built" super fast Jan 18 11:40:54 once per tune, unless they are really allarch or MACHINE_ARCH :) Jan 18 11:44:11 JaMa: how many cases of this problem do you run into? Jan 18 11:54:19 I've backported sstatesig.py change 2 days ago and 2 cases (in 6 different images) since then Jan 18 11:55:14 I've already proposed work around for our cases, where the package with just one library (so possibly renamed by debian.bbclass) is moved from packagegroup to some other recipe RDEPENDS Jan 18 11:55:56 but because such renames don't happen too often, I guess that there is more cases like this, just hard to trigger them Jan 18 11:56:13 and people wouldn't understand why it's happening and how to fix it :/ Jan 18 12:48:29 morning all Jan 18 12:50:39 RP: I seem to be unable to send email due to some poxy firewall here at heathrow, but the answer to your questin about SPE in gcc is yes: it's just a tuning flag like any other, and you can generate either spe or non-spe code with any ppc gcc binary by saying "-mspe" or "-mno-spe" as appropriate. Jan 18 12:51:05 I'm not quite sure why oe makes such heavy weather of the SPE thing. Conceptually it's no different to, say, -mfpu=neon on ARM. Jan 18 12:52:01 actually, maybe that's not quite true. I think -mspe does have abi impact, so perhaps it's more like -mfloat-abi=hardvfp or something. Jan 18 13:09:56 gm Jan 18 14:02:01 * JaMa sending couple of patches for my part of bug squashing weekend :) Jan 18 14:40:12 jackmitch|home, remember the boost didn't build on Arch bug? Jan 18 14:40:28 can you update it with your data point the it worked on your Arch machine? Jan 18 14:41:10 Crofton|work: sure, do you have a bug number Jan 18 14:42:21 Crofton|work: no worries, got it Jan 18 14:42:22 https://bugzilla.yoctoproject.org/show_bug.cgi?id=5484 Jan 18 14:42:27 heh ;) Jan 18 14:51:31 https://bugzilla.yoctoproject.org/buglist.cgi?bug_severity=janitors&resolution=---&query_format=advanced&bug_status=UNCONFIRMED&bug_status=NEW&bug_status=ACCEPTED&bug_status=IN%20PROGRESS%20DESIGN&bug_status=IN%20PROGRESS%20DESIGN%20COMPLETE&bug_status=IN%20PROGRESS%20IMPLEMENTATION&bug_status=IN%20PROGRESS%20REVIEW&bug_status=REOPENED&bug_status=NEEDINFO&bug_status=WaitForUpstream Jan 18 15:01:32 * RP is having "fun" with the bitbake manual Jan 18 15:01:51 trying do disentangle whitespace and content changes :( Jan 18 15:06:03 * JaMa not saying anything about whitespace consistency :) Jan 18 15:11:29 I need to figure out how to debug X stuff Jan 18 15:12:01 the OE spash appears during boot OK, but X (from x11-base image feature) doesn't display Jan 18 15:12:10 I see the server and matchbix processes Jan 18 15:14:35 [2642378.909] (EE) Failed to load module "modesetting" (module does not exist, 0) Jan 18 15:15:56 [2642980.650] (EE) FBDEV(0): FBIOBLANK: Device or resource busy Jan 18 15:16:11 [2701363.539] (EE) Server terminated successfully (0). Closing log file. Jan 18 15:16:15 those are the errors Jan 18 15:18:14 did you install xf86-video-modesetting? Jan 18 15:19:02 no Jan 18 15:19:08 what will that do? Jan 18 15:19:34 probably fix 'Failed to load module "modesetting"' :) Jan 18 15:19:42 yeah Jan 18 15:19:50 I am tryin gto imporve my knowledge :) Jan 18 15:20:02 I wonder why that is not installed from the IMAGE_FEATURE? Jan 18 15:21:31 it's a machine thing Jan 18 15:21:46 k Jan 18 15:22:02 would it lead to no display? Jan 18 15:22:14 the device or resource busy look s"bad" Jan 18 15:34:44 is there an image feature that replaces busybox with real tools? Jan 18 15:36:47 I realize image feature is a hammer, but not everything is a nail Jan 18 15:37:44 without explanation, that statement carries little information Jan 18 15:38:11 images are meant to give you a predefined set of packages Jan 18 15:38:22 all this image features stuff breaks that assumption Jan 18 15:38:34 I'm OK with that Jan 18 15:38:49 I am trying to eat our own dog food Jan 18 16:27:26 Crofton: I don't think there is IF for that Jan 18 16:27:26 I just install coreutils and live with both included in my images Jan 18 16:28:04 OK Jan 18 16:34:01 I hat ephotrums :) Jan 18 16:34:01 phorums Jan 18 16:41:47 Crofton|work: I don;t think we can totally replace busybox right now, we don't have 100% overlap Jan 18 16:42:21 it seems like a "reasonable thing for people to want" Jan 18 16:42:31 * Crofton|work curses X Jan 18 16:42:56 Crofton|work: so reasonable nobody's sent patches? :) Jan 18 16:43:07 seriously, someone send them... Jan 18 16:43:10 I suspect they do what JaMa said :) Jan 18 16:44:01 Crofton|work: right, that is certainly easy enough Jan 18 16:44:36 yeah, I'll hav eto poke at it some Jan 18 16:44:54 I hate gettign busybox on machine with large sd cards Jan 18 16:45:28 I want to get X11 goin gon theis lcd and oxroid-xu for fosdem frst Jan 18 16:45:37 my typing is getting really bad Jan 18 16:46:48 http://pastebin.com/0e1mmK3f Jan 18 16:48:47 any idea why pulse is complaining? Jan 18 16:52:55 the only thing that looks pulse-related there is the system vs user config Jan 18 16:53:03 yeah Jan 18 16:53:34 either run it system wide or start it on login session like normal, i guess Jan 18 16:54:20 my main concern is why doesn't it just work out of the box Jan 18 16:54:39 does the image even have login session stuff? Jan 18 16:55:07 looks like you need at least the --system thing for now Jan 18 16:55:59 if there's no login/display manager then you need to run a system-wide config Jan 18 16:56:01 i look slike the X startup process tries to start pulseaudio Jan 18 16:56:31 is it starting a root xsession? Jan 18 16:56:40 I assume so Jan 18 16:57:03 basically this is care-x11-image Jan 18 16:57:03 then you need the --systemconfig Jan 18 16:57:40 i don't use pulse in my xorg image since i don't have a login manager yet Jan 18 16:58:11 bigger issue is no display when X starts :) Jan 18 17:01:25 so you're both deaf and blind? Jan 18 17:04:39 and stupid Jan 18 17:09:17 can't hear/see != stupid... Jan 18 17:09:46 unless you're deaf and blind from taking viagra Jan 18 17:09:53 the added problem is I am not good at depending x startup issues Jan 18 17:09:58 on un tested hw Jan 18 17:10:55 if I start wo X/psplash, I think I expect a login prompt on the lcd panel Jan 18 17:11:07 psplash does look ok Jan 18 17:12:28 viagra, deafness, blindness, /me learns lots of new interesting stuff from #oe :-D Jan 18 17:12:52 yep Jan 18 17:13:03 this is a good exercise Jan 18 17:13:05 well, yeah, i would expect one of a) console login, b) dm login, or 3) straight into X Jan 18 17:13:42 one area I want to see OE improve is have more stuff work out of the box Jan 18 17:14:51 for every combination of distro/machine features? Jan 18 17:15:35 every sane combo at least :) Jan 18 17:16:03 the pulse thing for example Jan 18 17:16:14 we are trying to run it as root, so it need the option Jan 18 17:16:15 making the X images use a dm would help Jan 18 17:17:07 if the minimal X image booted up to slim that would be moot Jan 18 17:17:09 90% of my problem is that desktop X has just worked for so long, I forget how it works :) Jan 18 17:18:05 Just Works with things like pulse requires a full desktop for "default" configs to work Jan 18 17:18:45 *full as in conslekit/policykit/dbus/display-manager etc... Jan 18 17:19:53 at least a minimal "sane" session handling chain as it were Jan 18 17:20:46 that's on my list for the layer i just broke out Jan 18 17:22:01 ie, getting slim in my xorg/openbox image and adding a default "pi" user instead of booting straight into a root desktop Jan 18 17:23:19 i'm open to suggestions for a better name than "meta-alt-desktop-extras" Jan 18 17:24:59 i thought about calling it "alt-wesley-crusher-die-die-die" but that name was taken... Jan 18 17:25:38 rofl Jan 18 17:25:48 usenet joke anyone? Jan 18 18:25:25 now back to "wtf is up with mupdf?" Jan 18 20:41:55 okay, it builds again Jan 18 20:42:10 just not the tools... Jan 18 20:53:04 okay, how do i get stubs-soft.h installed in my sysroot? Jan 18 20:54:02 it's there for bbb but not for rpi Jan 18 21:14:30 is it a TUNE issue maybe? Jan 18 21:30:12 anybody? Jan 19 02:23:55 what sets x11 in DISTRO_FETURES Jan 19 02:31:32 or mor eto th epoint, how do I remove x11 from DISTRO_FEATURES so I can add directfb Jan 19 02:41:50 i guess you could set it in your image or try the filter thing Jan 19 02:42:07 * nerdboy just fixed mupdf Jan 19 02:42:13 *again* Jan 19 02:45:31 and this time it builds a shared lib and the x viewer Jan 19 02:46:05 * nerdboy grumbles something derogatory about wonky upstream builds Jan 19 02:46:13 I need to turn off x11 so I can build a directfb image Jan 19 02:46:25 crofton -- define your own distro and don't put x11 in your DISTRO_FEATURES Jan 19 02:46:37 include console-image ? Jan 19 02:48:00 fray, seriously? I need to make a distro conf to kill x11? Jan 19 02:48:20 if you want to prevent x11 from being in the distro_features, you need to set them.. the best way is to define a distribution Jan 19 02:48:53 distro=poky and a custom image recipe with inherit core-image-minimal and include console-image Jan 19 02:49:04 that should work... Jan 19 02:50:09 DISTRO_FEATURES_remove = "x11" probably works as well.. Jan 19 02:51:20 holding back? Jan 19 02:52:10 that worked **** ENDING LOGGING AT Sun Jan 19 02:59:59 2014