**** BEGIN LOGGING AT Sun Oct 23 02:59:58 2016
Oct 23 05:12:53 <Herrie> morning
Oct 23 05:12:59 <Herrie> morphis: ping
Oct 23 05:36:35 <ka6sox> heya Herrie....whats up?
Oct 23 05:38:03 <Herrie> ka6sox: Nothing much, have a bit of a cold that doesn't wanna go, otherwise things are good
Oct 23 05:38:07 <Herrie> How about you?
Oct 23 05:38:14 <Herrie> Back home or out & about somewhere?
Oct 23 05:38:29 <ka6sox> finally back home...trying to catch up!
Oct 23 05:39:35 <ka6sox> I see good progress here in the logs
Oct 23 05:40:00 <Herrie> ka6sox: Yeah it's pretty good :) We've been able to clean a lot of old bugs & issues ;)
Oct 23 05:40:23 <Herrie> Lots were caused by systemd startup order & not writing a simple flag after first use ;)
Oct 23 05:41:42 <ka6sox> d0h!
Oct 23 05:41:55 <ka6sox> good that you'all were able to get it sorted out.
Oct 23 05:43:05 <Herrie> Yup :) Now looking to see if we can move to newer Yocto (Morty) soon after it gets released but some challenges there still
Oct 23 05:45:43 <ka6sox> upgrading to a new release always has its challenges and opportunities...
Oct 23 05:45:52 <Herrie> ka6sox: Yup ;)
Oct 23 05:46:11 <Herrie> Just it won't build it seems for most targets... Was able to sort most though
Oct 23 05:46:24 <Herrie> I know little to no c++ so it's always hacking aorund :P
Oct 23 05:50:29 <ka6sox> what are the currently supported targets?
Oct 23 05:57:24 <Herrie> ka6sox: Galaxy Nexus (maguro), Nexus 4 (mako), Touchpad (tenderloin), Nexus 7 2012 (grouper) and Andolamin is trying to get the Raspberry Pi's working again
Oct 23 05:57:32 <Herrie> And we have good old qemux86 as well
Oct 23 05:59:29 <ka6sox> :)
Oct 23 05:59:41 <ka6sox> good
Oct 23 06:03:10 <Herrie> We might be looking into some more recent target as well, but we'd want to get things running properly first on current targets.
Oct 23 06:03:18 <Herrie> Since any fix is pretty generic anyway
Oct 23 08:22:49 <elvispre> Ping Herrie
Oct 23 08:23:22 <elvispre> Herrie: Did you establish when was the last time BT [audio] was working?
Oct 23 08:36:07 <Herrie> elvispre: Seems never on Mako.. At least I tried everything back to March and there were always issues
Oct 23 08:37:05 <elvispre> Herrie: :-) Well, I was using it to listen to music for a while so it can't have been that bad.
Oct 23 08:37:41 <elvispre> Herrie: But my best guess is March in that case.
Oct 23 08:38:25 <elvispre> Herrie: (From my diary of when I was on the bike rollers!)
Oct 23 08:44:22 <elvispre> Herrie: In my log I see several "Method ... on interface org.bluez.Audio doesn't exist" messages. An exhaustive "find" on the tmp-glibc folder doesn't find any references to org.bluez.Audio so that's my best idea about the problem at the moment.
Oct 23 08:47:06 <Herrie|Pre3> elvispre: That seems to be Luna-Next-Cardshell that gives that error if I'm correct.
Oct 23 08:48:03 <Herrie|Pre3> elvispre: It's quite complicated with BlueZ, PulseAudio, ConnMan all being involved + our own plugin. If something is not working it's quite hard to track down.
Oct 23 08:48:17 <Herrie|Pre3> elvispre: I'm also able to get audio sometimes but not always.
Oct 23 08:48:27 <Herrie|Pre3> And then there's systemd and policies as well still :s
Oct 23 08:49:05 <Herrie|Pre3> Just the indicator doesn't seem to work in LNC and the status in connman doesn't seem to change to connected/paired even though pairing succeeded
Oct 23 08:49:11 <Herrie|Pre3> At least that's what I could see.
Oct 23 08:49:39 <Herrie|Pre3> BlueZ4 is a bit old and will be removed from ConnMan though adding it back in is putting back a single file.
Oct 23 08:50:02 <Herrie|Pre3> BlueZ5 upgrade gives a lot of other headaches on kernel side.
Oct 23 08:51:05 <Herrie|Pre3> We might want to have a look into Mer/Jolla's ConnMan version instead of mainline one. Since we're using a lot of similar things with Mer guys it might make sense to have their ConnMan version instead.
Oct 23 08:51:17 <Herrie|Pre3> But that might break other things again of course :s
Oct 23 08:51:40 <elvispre> Herrie: Another "opportunity" to learn about new things :-P
Oct 23 09:06:42 <Herrie|Pre3> elvispre: Yeah let me get this Morty stuff working first. Already getting way too deep into C++ for my liking :P
Oct 23 09:06:56 <Herrie|Pre3> Then I'll play around a bit with Mer's ConnMan
Oct 23 10:07:05 <Tofe> Morning!
Oct 23 10:12:54 <Herrie> Tofe: morning!
Oct 23 10:13:00 <Herrie> Does this look proper to you? https://github.com/webOS-ports/FreeImage/commit/c01653d6b12939877e79d2787d3695865e8b1ee3
Oct 23 10:15:44 <Tofe> wow, 'char' is unsigned on ARM, I've learned something today!
Oct 23 10:16:02 <Tofe> your commit seems fine
Oct 23 10:19:34 <Herrie> Tofe: OK
Oct 23 10:20:23 <Tofe> gosh, it's worse than I thought http://blog.cdleary.com/2012/11/arm-chars-are-unsigned-by-default/
Oct 23 10:20:51 <Herrie> Tofe: libhybris is giving issues too: hooks.c:1473:5: error: built-in function '__sprintf_chk' must be directly called {"__sprintf_chk", __sprintf_chk} and hooks.c:1474:5: error: built-in function '__snprintf_chk' must be directly called {"__snprintf_chk", __snprintf_chk}. I can simple comment those 2 out in hooks.c? It seems to work....
Oct 23 10:22:36 <Tofe> no, that would possibly introduce very weird bugs
Oct 23 10:22:54 <Tofe> let me look at the source code for these lines
Oct 23 10:26:04 <Tofe> with morty, we bump gcc from what version to what version?
Oct 23 10:28:02 <Tofe> and glibc too ?
Oct 23 10:29:25 <Tofe> 5.4 to 6.2 I guess for gcc
Oct 23 10:30:36 <Tofe> and glibc 2.23 to 2.24 it seems
Oct 23 10:34:50 <Tofe> Herrie: do you have access to the work directory?
Oct 23 10:35:40 <Herrie> Tofe: Yeah
Oct 23 10:35:44 <Herrie> It's on my local build machine
Oct 23 10:36:30 <Tofe> Because __sprintf_chk should be defined in usr/include/bits/stdio2.h (at the beginning), and the latter is included by stdio.h, which is already included by the hooks.c file here
Oct 23 10:38:12 <Tofe> ah wait, now I understand the error message
Oct 23 10:43:06 <Tofe> Herrie: we have two ways of fixing this. 1. create a function in hooks.c, named "my_sprintf_chk", with the same signature as the builtin one, and just call the builtin one. Or 2. use the "-fno-builtin" flag when building libhybris.
Oct 23 10:43:16 <Tofe> (at least when building hooks.c)
Oct 23 10:43:41 <Tofe> I guess the second way is more adapted to the situation.
Oct 23 10:47:07 <Tofe> Trhis sums up to adding to "hybris/common/Makefile.am" a line like "libhybris_common_la_CFLAGS += -fno-builtin"
Oct 23 10:47:29 <Tofe> I don't know if we can set it only for one .c file
Oct 23 10:48:34 <Tofe> ah yes: "hooks_CFLAGS += -fno-builtin"
Oct 23 10:49:25 <Herrie> Tofe: OK let me give that a go when my build is ready
Oct 23 10:49:32 <Herrie> Might make sense to PR this upstream?
Oct 23 10:51:58 <Tofe> yes, totally
Oct 23 10:52:27 <Herrie> Shouldn't have adverse affects for gcc < 6 ?
Oct 23 10:55:39 <Herrie> Tofe: Any preference for where this should be in Makefile.am?
Oct 23 10:56:36 <Tofe> before line 29
Oct 23 10:56:56 <Tofe> for gcc<6 it should work too I think
Oct 23 10:57:15 <Tofe> I can test it live on my krogoth
Oct 23 10:59:47 <Tofe> ah, it's "=", not "+="
Oct 23 11:01:56 <Herrie> Tofe yeah figured that out
Oct 23 11:02:01 <Herrie> I told me it needed to be =
Oct 23 11:06:50 <Tofe> my CFLAGS line seems to have no effect
Oct 23 11:10:47 <Herrie> Tofe: For me it builds with it..
Oct 23 11:11:31 <Herrie> Ah wait I reverted the change in the wrong hooks.c
Oct 23 11:11:38 <Herrie> I did it for Tenderloin instead of Mako :S
Oct 23 11:13:13 <Herrie> Yup doesn't work
Oct 23 11:13:30 <Tofe> ok let's put the flag for the whole target
Oct 23 11:14:12 <Tofe> (add it to the  libhybris_common_la_CFLAGS line)
Oct 23 11:14:13 <Herrie> we could suffice by adding it for libhybris_common_la_CFLAGS ?
Oct 23 11:14:20 <Herrie> Yeah that's what I'm trying now
Oct 23 11:15:06 <Tofe> --> error: '__sprintf_chk' undeclared here (not in a function)
Oct 23 11:15:35 <Tofe> let's fallback to way 1
Oct 23 11:17:36 <Herrie> Tofe: Yup seems so :S
Oct 23 11:21:35 <Tofe> oh no, it's variadic functions with "..."
Oct 23 11:32:26 <Tofe> Herrie: ygm
Oct 23 11:34:53 <Herrie> Tofe: Thnx will give that a go in a minute
Oct 23 11:47:50 <Herrie> Yup that works thnx
Oct 23 11:50:43 <Herrie|Laptop> Tofe: Seems there will be a few more to tackle ;) But I'll look into that later.... Sorting them 1:1 while JaMa is moving ;)
Oct 23 11:50:50 <Herrie|Laptop> See if I can get something to build at least
Oct 23 12:21:53 <Herrie|Laptop> Tofe: I noticed a few bugs in Browser in fullscreen on Touchpad, not sure if it's also as visible on N4. Not sure where to look to address these to be honest:
Oct 23 12:22:27 <Herrie|Laptop> 1. After rotating right and back left the resizing of the screen seems off. I need to minimize app and maximize to get the correct aspect ratio.
Oct 23 12:22:49 <Herrie|Laptop> Also it seems that after rotation & rotating back, the on-screen controls aren't working
Oct 23 12:23:05 <Herrie|Laptop> I.e. the coordinates are off. I tried this on YouTube, since it's best visible there
Oct 23 12:23:22 <Herrie|Laptop> When I try to tap on the slider for example it will always pause the video, instead of moving the slider
Oct 23 12:23:42 <Herrie|Laptop> So I have the idea that something is off with the touch coordinates after rotating & rotating back.
Oct 23 12:23:55 <Tofe> might be a qtwayland issue, but not sure
Oct 23 12:23:58 <Herrie|Laptop> Not sure this is Browser, LunOS Components WebEngineView or LNC
Oct 23 12:24:08 <Herrie|Laptop> Or even QtWayland like you say
Oct 23 12:24:17 <Tofe> could be webengine too, yes
Oct 23 12:24:37 <Herrie|Laptop> Will try to see if I can replicate on another target sometime as well
Oct 23 12:24:46 <Herrie|Laptop> I'll try to make a video to show what I meman
Oct 23 12:24:58 <Herrie|Laptop> Always shows better then my description LOL ;)
Oct 23 12:25:12 <Herrie|Laptop> It's not very annoying, just I noticed while I was playing with YouTube the other day
Oct 23 12:29:56 <Herrie|Laptop> I'll add it to the issue trakcer so we can keep track of it
Oct 23 12:50:54 <Tofe> ok good
Oct 23 12:51:27 <Herrie|Laptop> Let me first try to get this Morty stuff building :P
Oct 23 12:51:28 <Tofe> Herrie|Laptop: in Enyo, are radio buttons necessarily in an horizontal layout, like I see in StyleMatters?
Oct 23 12:51:42 <Herrie|Laptop> And send JaMa a list of things he needs to patch LOL ;)
Oct 23 12:51:46 <Herrie|Laptop> Tofe: Let me check
Oct 23 12:55:47 <Herrie|Laptop> Tofe: Hmmz I was expecting to see something else with radio button :P
Oct 23 12:56:51 <Tofe> me too!
Oct 23 12:57:55 <Herrie|Laptop> I'd say it could be either horizontal or vertical
Oct 23 12:58:06 <Herrie|Laptop> Most implementations I've seen are vertical
Oct 23 12:58:37 <Herrie|Laptop> jQuery seems to have it implemented for both: http://2.bp.blogspot.com/-wXlo3QTwkXk/TzJxOkTzOQI/AAAAAAAAAco/wvpdd5sLojI/s1600/jQueryMobile_RadioButton.png
Oct 23 12:58:52 <Herrie|Laptop> Which to me makes sense
Oct 23 13:01:58 <Tofe> Yes, makes sense. Still, in Onyx there seems to be only the horizontal variant...
Oct 23 13:02:42 <Tofe> However my main problem here is the way they are rendered: the first and last buttons of the group have a different look
Oct 23 13:03:34 <Tofe> I fear I'll have some issues implementing that in QML
Oct 23 13:03:44 <Herrie|Laptop> Tofe: You mean the rounding?
Oct 23 13:03:49 <Herrie|Laptop> Or...
Oct 23 13:03:50 <Herrie|Laptop> ?
Oct 23 13:03:51 <Tofe> yes
Oct 23 13:04:31 <Tofe> Currently a delegate doesn't know if he's the first in the list of not
Oct 23 13:05:10 <Herrie|Laptop> Can't you do something in the onCompleted of the model when all delegates are known?
Oct 23 13:05:16 <Herrie|Laptop> I.e. apply styling then
Oct 23 13:06:05 <Tofe> Nope, that'll be on the app side, and I'm implementing the theme
Oct 23 13:06:56 <Tofe> Though I *could* try to minimize the work needed on app's side
Oct 23 13:07:26 <Tofe> Like an additional property, "totalCount", which would inform the delegate about beeing the last one
Oct 23 13:07:59 <Tofe> I'll give it a try, we'll see.
Oct 23 13:08:12 <Herrie|Laptop> :)
Oct 23 13:08:24 <Herrie|Laptop> How is it going in general? Most don't seem too complicated?
Oct 23 13:08:31 <Herrie|Laptop> At least going by the examples :)
Oct 23 13:09:29 <Tofe> It's mostly ok
Oct 23 13:09:48 <Tofe> I can upload an updated version of my work, so that you can see where I am now
Oct 23 13:11:46 <Herrie|Laptop> :)
Oct 23 13:13:37 <Tofe> Done, you can checkout my branch
Oct 23 13:16:09 <Tofe> so far, I'm happy with Buttons, TextArea, TextField, CheckBox, ComboBox, Drawer, Frame, GroupBox, Popup, Switch
Oct 23 13:17:34 <Tofe> The remaining big ones are RadioButtons, TabBar, Menu, Sliders...
Oct 23 13:19:29 <Tofe> I'm not sure where to put the ToggleButton
Oct 23 13:21:10 <Herrie|Laptop> ToggleButton = Switch?
Oct 23 13:21:50 <Herrie|Laptop> ToggleButton in Enyo is more or less the Switch type in QML inmo
Oct 23 13:21:53 <Herrie|Laptop> imho
Oct 23 13:23:55 <Tofe> yes, but there is possibly a text for on and off
Oct 23 13:30:22 <Herrie|Laptop> Tofe: Yeah true
Oct 23 13:30:38 <Herrie|Laptop> But Switch has SwitchStyle that you could use
Oct 23 13:37:45 <Tofe> Well yes, but still. Oh and there's not *Style anymore in QtQuickControls2
Oct 23 13:37:51 <Tofe> no*
Oct 23 13:38:12 <Tofe> I'm directly modifying the view of the widgets themselves
Oct 23 13:39:42 <Herrie|Laptop> Tofe: Ah, didn't look into QtQuickControl2 yet ;) DIdn't know the styles were gone...
Oct 23 13:40:40 <Tofe> They moved the behavior into C++ code, and only the view of the controls is in QML now
Oct 23 13:41:00 <Tofe> It's much lighter, and also quicker
Oct 23 13:44:08 <Herrie|Laptop> Better performance is always good :D
Oct 23 19:50:40 * elvispre notes for the record: From the Lune OS Café Mocha changelog, May 2016: "bluez4: Add dbus policy so that Bluetooth headset can connect and route audio"
Oct 23 19:54:49 <Tofe> Herrie: I've pushed the radiobutton styling to my branch, if you're interesting to see how it looks like
Oct 23 19:55:04 <Tofe> interested*
Oct 23 20:01:34 <Herrie> Tofe: Will check in the morning, got a bit of a headache so hitting bed early today
Oct 23 20:02:37 <Tofe> ok :) sleep well!
**** BEGIN LOGGING AT Mon Oct 24 01:30:16 2016
Oct 24 01:33:45 <DougReeder> Tofe, can you implement the radio button styling, using a container to round the corners, so individual buttons don’t need to know if they are first, last or middle?
**** ENDING LOGGING AT Mon Oct 24 02:59:57 2016