2016-01-11 00:08:16	jfred-matrix	Hmm okay, fresh attempt failed - here's the build log: https://gist.github.com/jfrederickson/569cc04f48e9e2f7e89e
2016-01-11 00:11:02	@kido	https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60752 it just seems you don't have enough ram to compile gcc, there's not much we can do about that :/
2016-01-11 00:12:53	@kido	Use swap or something, how much ram do you have on your build machine?
2016-01-11 00:25:27	@kido	Afk
2016-01-11 01:22:47	jfred-matrix	Ohhhh it's a RAM issue? No worries then, I'll bump up the RAM allocation on that VM
2016-01-11 01:24:10	jfred-matrix	The VM only had 2 GB allocated to it, but the physical host has 36
2016-01-11 01:28:21	jfred-matrix	I really need to get serial consoles up and running on my VMs so I can see if they're OOMing
2016-01-11 01:28:34	jfred-matrix	it's not at all obvious when you're SSH'd in
2016-01-11 01:30:41	jfred-matrix	actually it should show in dmesg, whoops
2016-01-11 02:21:07	jfred-matrix	Hah, yep, during the build it's using slightly more than 2 GB... welp
2016-01-11 03:23:35	jfred-matrix	Hrm... okay about how much disk space does this thing need to build?
2016-01-11 05:46:35	jfred-matrix	Woo, finally got asteroid running! :D
2016-01-11 05:47:26	jfred-matrix	the default image doesn't appear to have any apps though - should there be a settings app?
2016-01-11 11:09:52	@kido	jfred-matrix: cool! Yes, there are supposed to be several apps installed, more importantly do you have adb?
2016-01-11 11:11:37	@kido	It's realy curious that toxip built asteroid just a day before you and had apps and no adb, especialy since I didn't make any modification in between
2016-01-11 13:54:10	jfred-matrix	kido: Yep I can get a shell with ADB. Also... it looks like my watch's battery died in ~7 hours overnight
2016-01-11 13:56:44	@kido	alright for adb, then I realy don't understand what happens to toxip. for the battery it's more or less normal, nothing has been done for energy consumption atm
2016-01-11 14:13:08	jfred-matrix	Are the asteroid apps in a repository/can I install them over adb? (If I manage to get an internet connection on this thing that is. XD)
2016-01-11 14:13:41	locusf	I wonder if you could just use ssh+mb2
2016-01-11 14:14:12	jfred-matrix	and I suppose for the battery you'll probably need some kind of aggressive sleep mechanism heh
2016-01-11 14:14:48	jfred-matrix	just so I'm not missing something though... how would you launch an app in asteroid?
2016-01-11 14:14:55	jfred-matrix	if there's an app drawer I haven't found it
2016-01-11 14:15:02	jfred-matrix	just the three home screens
2016-01-11 14:16:57	jfred-matrix	ohhh never mind it's a swipe up/down
2016-01-11 14:17:00	jfred-matrix	derp
2016-01-11 14:17:54	jfred-matrix	can't seem to open the settings app though
2016-01-11 14:18:36	jfred-matrix	battery shows as undefined%
2016-01-11 14:22:38	jfred-matrix	in any case, at least now I can play around with QML on this thing :)
2016-01-11 14:31:29	@kido	jfred-matrix: the asteroid apps are available in source code forms under github https://github.com/AsteroidOS they are built by openembedded when you build asteroid so you can find them in asteroid/build/tmp-glibc/work/armv7a-vfp-neon-oe-linux-gnueabi/asteroid-settings/+gitAUTOINC-r1/ the source code is under git, the configure/compile/install cross compilation scripts are under temp and the target
2016-01-11 14:31:29	@kido	directory wheree they are installed is image/, you can adb push the image folder on the asteroid's root directory and the app will be installed
2016-01-11 14:32:22	@kido	so a basic workflow if you want to hack asteroid-settings would be to go to asteroid/build/tmp-glibc/work/armv7a-vfp-neon-oe-linux-gnueabi/asteroid-settings/git/, modify the files you want, run ../temp/run.do_compile ; ../temp/run.do_install and adb push ../image / 
2016-01-11 14:32:58	@kido	try to run asteroid-settings and paste me the result of the end of journalctl
2016-01-11 14:33:22	@kido	I suspect you use an old version of qml-asteroid
2016-01-11 14:34:09	@kido	if you want to get an internet connection it's probably tricky and I never achieved that, I suppose that for now an "easy" way would be to setup PPP over adb
2016-01-11 14:34:49	@kido	on the long term we should use the kernel's usb otg infrastructure to setup an ethernet over usb connection https://www.kernel.org/doc/Documentation/usb/gadget_multi.txt
2016-01-11 14:35:22	@kido	that would also allow ssh which would finaly be a real shell (adb shell sucks like hell), usb mass storage and so on...
2016-01-11 14:36:46	@kido	for the battery the infrastructure is the one from sailfishos so basicaly mce does all the dirty job and it seems there was some problem with the kernel being a bit too recent, I tried to fix it but I didn't spend a lot of time on it
2016-01-11 14:37:24	@kido	for the battery showing undefined% it comes from statefs still not being completely functional
2016-01-11 14:38:18	@kido	if you want to play with QML don't forget to check the wiki for info
2016-01-11 14:40:06	@kido	and if you want to help on all the problems you mentioned here (statefs, mce, settings app, internet connection, launcher usability, openembedded build errors...) that would, of course, be greatly welcome :)
2016-01-11 14:41:12	@kido	I'll be able to help more intensively this week end, I have a shitload of exams during the next few days :)
2016-01-11 15:15:21	jfred-matrix	kido: Here you go - https://gist.github.com/jfrederickson/39cd21d3a3596c42feb0
2016-01-11 15:15:53	jfred-matrix	but thanks for the tips :)
2016-01-11 15:17:08	jfred-matrix	I'll definitely think about launcher usability - re: openembedded build errors, I think all my problems were due to lack of RAM and disk space, so that probably deserves a mention on the install page
2016-01-11 15:18:08	jfred-matrix	Also notable, I do have a bunch of these in my journalctl log: Jan 11 01:54:06 dory mce[434]: modules/battery-statefs.c: tracker_open(): /run/state/namespaces/Battery/State: open: No such file or directory
2016-01-11 15:18:55	jfred-matrix	that repeats every few seconds... so, it definitely seems to be something related to statefs
2016-01-11 15:19:16	@kido	jfred-matrix: git pull in meta-asteroid then bitbake asteroid-image, if you want to repush the image it will work but if you don't want to push the image you can just push the missing dependencies with:
2016-01-11 15:19:17	@kido	adb push tmp-glibc/work/armv7a-vfp-neon-oe-linux-gnueabi/nemo-qml-plugin-systemsettings/+git8e98f8c8069592d3dd21011743fdfb4d2efcc50a-r1/image /
2016-01-11 15:19:18	@kido	and
2016-01-11 15:19:39	@kido	adb push tmp-glibc/work/armv7a-vfp-neon-oe-linux-gnueabi/nemo-qml-plugin-dbus/+gitc4d98477aaea95d7cbbf0414ed7aeb802942737f-r1/image /
2016-01-11 15:20:11	@kido	jfred-matrix: yes that's the problem behind the undefined% battery
2016-01-11 15:20:17	jfred-matrix	Aha, thanks! I'll give that a try after work
2016-01-11 15:20:39	@kido	for openembedded build errors maybe you could add to the wiki the minimum resources you had to use
2016-01-11 15:21:52	@kido	mhh you'll also probably miss two other dependencies
2016-01-11 15:22:09	@kido	adb push tmp-glibc/work/armv7a-vfp-neon-oe-linux-gnueabi/usb-moded-qt5/+git0a4b8ffe0cd775cce7a2366c8363d29d617b8e60-r1/image /
2016-01-11 15:22:10	@kido	and
2016-01-11 15:22:22	@kido	adb push tmp-glibc/work/armv7a-vfp-neon-oe-linux-gnueabi/usb-moded/+giteadebc6ce7230e2789ff0999a8b15a49a4aceaf3-r1/image /
2016-01-11 15:22:56	@kido	those adb push and adb shell hacks are terrible but once we get a reliable internet connection over usb we can use the package manager, openembedded keeps a repository for us
2016-01-11 15:23:11	 *	jfred-matrix will try and get a network connection on his watch XD
2016-01-11 15:23:33	@kido	;)
2016-01-11 15:23:51	jfred-matrix	maybe bluetooth tethering makes more sense honestly, since the watch will be connected to a phone via bluetooth anyway
2016-01-11 15:24:04	@kido	sure
2016-01-11 15:24:36	@kido	but you also got to think about porters
2016-01-11 15:25:12	@kido	I miss the good old days when I booted the watch blindly without any shell or ui at all
2016-01-11 15:26:16	locusf	PAN might work
2016-01-11 15:26:19	locusf	with connman even
2016-01-11 15:27:04	@kido	locusf: never setup that but it seems to suit our needs indeed
2016-01-11 15:27:44	locusf	I've used PAN manually at times, since 2005 or something
2016-01-11 15:28:03	locusf	no need to setup complicated wifi ap's + lower power consumption
2016-01-11 15:28:09	locusf	and no wires vs usb
2016-01-11 15:32:08	jfred-matrix	locusf: On that note, some protocol over PAN may well be the way to go re: phone-watch comms
2016-01-11 17:25:26	locusf	jfred-matrix: well its a tad heavy
2016-01-11 17:26:47	jfred-matrix	hmm
2016-01-11 17:31:30	locusf	tcp stack vs BLE characteristics transfer
2016-01-11 18:48:41	locusf	kido: so I've gotten a bit closer to the issue why libiodata and friends don't work
2016-01-11 18:48:49	locusf	they are being built as qt4
2016-01-11 18:49:11	locusf	or at least eg libiodata has 2 choices to build for either qt4 or qt5
2016-01-11 19:12:39	locusf	kido: https://github.com/meta-qt5/meta-qt5/wiki/Building-with-OE#bad-include-paths-and-build-failing-to-find-qt5-headers ...
2016-01-11 19:47:35	locusf	FFFFFUUUUUUUU
2016-01-11 19:47:47	locusf	the build was reading some meego aged mkspec files ....
2016-01-11 20:43:02	@kido	locusf: ah ok, good job!
2016-01-11 20:45:17	@kido	locusf: if OE_QMAKE_PATH_HEADERS = "${OE_QMAKE_PATH_QT_HEADERS}" fixed the build you can pull request
2016-01-11 20:50:11	locusf	kido: it didn't
2016-01-11 20:51:56	@kido	mhhh :/
2016-01-11 21:09:36	locusf	but skmething else did happen
2016-01-11 21:09:56	locusf	the build went further than before