[02:38] <duflu> RAOF: I take it an Android backend (or stub) for subpixel is coming?
[02:38] <duflu> Or not required?
[02:39] <duflu> Maybe x11 too
[03:34] <duflu> Awesome. Kernel 4.7 fixes my black screen by turning it into an equally plain orange screen
[04:40] <RAOF> duflu: Nested is up next. Android already sets it to _unknown, which might be as good as we can do?
[04:47] <duflu> RAOF: I would hope Android can do better. Will grep later. Also x11 should be able with xrandr info..?
[04:47] <RAOF> Yup. X11 should be reasonably easy.
[05:59] <duflu> RAOF: I can't find any such option for Android/HAL... Probably just as well while unity8 does client-side screen rotation, which would need work in that case
[06:00] <RAOF> unity8 could certainly make the appropriate changes.
[06:00] <RAOF> Of course, that requires the nested platform to do the right thing, which it doesn't at the moment :)
[06:00] <duflu> RAOF: I think it's mostly desktop monitors that U8 just isn't ready for
[06:01] <duflu> "low" DPI
[06:01] <RAOF> Indeed.
[06:01] <RAOF> Sub-400DPI :)
[06:01]  * duflu is reminded to investigate FRC
[06:01] <duflu> again
[06:04] <RAOF> FRC?
[06:05] <duflu> RAOF: I think it's also known as DRM_MODE_DITHERING_*  (https://en.wikipedia.org/wiki/Frame_rate_control)
[06:05] <RAOF> Oh, temporal dithering?
[06:05] <duflu> Yes
[06:05] <duflu> RAOF: I noticed recently Macbook Airs still need it
[06:05] <duflu> In 2016 just as they did in 2010 :P
[06:06] <RAOF> Seriously? I thought Apple tended to ship non-terrible hardware.
[06:06] <duflu> Yep, seriously.
[06:07] <duflu> It looks extra bad in Ubuntu like spatial dithering. I think we need to flip a switch
[06:08] <RAOF> Yay! Trade colour terribleness for 30Hz flicker!
[06:08] <duflu> Yep... and I only just convinced John Lea to stop shipping pre-dithered wallpaper
[08:04] <duflu> RAOF: What was it you said about relating surfaces to output info before? I have a plan, but don't want to conflict if you have one
[08:43] <RAOF> duflu: Pass a MirOutput* as a part of the surface-output event.
[08:43] <duflu> RAOF: That's won't work if you need to abstract it to a non-real MirOutput
[08:44] <duflu> ie. clone mode or overlapping multiple outputs
[08:44] <RAOF> ??
[08:44] <duflu> So you should probably wait for my general solution
[08:44] <duflu> Or go have dinner...
[08:45] <RAOF> True, but overlapping outputs don't make sense, and clone mode you can happily pick whichever you choose.
[08:46] <duflu> RAOF: In clone mode, two physically different arrangements should surely be generalized to disabling sub-pixel?
[08:46] <duflu> Similarly I need to proxy frame timing info
[08:47] <duflu> So it's not that we should say a surface is on output X, but the common traits of the outputs it is on are A,B and C
[08:53] <duflu> RAOF: I meant a surface "overlapping" multiple outputs. In such a case what is its subpixel arrangement, and frame interval, and physical resolution etc
[08:53] <duflu> The answer may not match any 'real' output
[08:59] <RAOF> In such a case, you pick one output.
[09:00] <RAOF> (Specifically, you pick the output with the highest scale factor ☺)
[09:02] <duflu> RAOF: Yeah I get that we can survive with that compromise. And in the single monitor case the answer is absolutely correct. I'm just trying to future-proof
[09:03] <RAOF> Surfaces should generally span multiple outputs only transiently, and you can't render them correctly anyway.
[09:03] <duflu> RAOF: I used to believe that. Till we were asked to do a monitor "wall" project
[09:04] <duflu> One surface, many monitors
[09:04] <duflu> Although in that case everything except the phase of display would match
[09:05]  * duflu now remembers what he was looking at
[10:30] <RAOF> For a video wall wouldn't you just do multiple surfaces?
[10:30] <RAOF> Or, alternatively, multiple buffer streams in a single surface.
[10:34]  * ogra_ would just make vlc work :P