[PATCH 2/6] compositor-drm: Allow instant start of repaint loop.

[Mario Kleiner]

On 04/08/2015 07:49 AM, Daniel Stone wrote:
> Hi Mario,
>
> On Wednesday, April 8, 2015, Mario Kleiner <mario.kleiner.de at gmail.com
> <mailto:mario.kleiner.de at gmail.com>> wrote:
>
>     It should be fine, i was involved in this stuff for lots of XOrg
>     DRI2/DRI3 work, but i also tested it on single display and
>     dual-display for each of the two outputs and for both outputs
>     simultaneously and different refresh rates, resolutions etc. --
>     hence my other patches in the series fixing fullscreen/modesetting
>     problems etc. i found when trying to do the multi-display timing testing
>
>
> Out of interest, are these tests available somewhere?
>

Hi Daniel,

they are not standalone tests. I just manually ran various demos and 
tests from my toolkit, checking if they roughly behave as expected. The 
software is Psychtoolbox-3 (http://psychtoolbox.org), a popular MIT 
licensed open-source toolbox to extend GNU/Octave and Matlab with 
functionality needed for neuro-science/medical research. Think of it as 
something like SDL but optimized for the needs of brain research. So all 
tests are written as scripts for execution within Octave, or in case of 
the compiled C plugins, as part of the toolkits builtin startup and 
runtime consistency checks. The X11/GLX variant of the toolkit is 
available on Debian/Ubuntu via apt-get octave-psychtoolbox-3, the latest 
versions from neuro.debian.net ppa. However, I just started implementing 
an experimental Wayland backend at the beginning of this year and this 
is not yet available via Debian, only as a compile time option when 
building it from source, with dependencies on libwaffle which is not 
bundled with Debian, so getting it up on Wayland isn't exactly pain free 
atm. I could step you through the procedure if you wanted, but it's 
probably more easy to wait a bit more for the wayland backends build and 
installation procedure to mature before some of this could serve as a 
test case or actual test. The goal is for the wayland variant to match 
or exceed the x11 variant in functionality and performance. Btw., 
running the x11 variant on XWayland instead of on a native X-Server is a 
good method to crash or hang the toolkit, XWayland, or Weston ;)

>         If you're going to respin this patch, you could take care
>         of that whitespace issue also(*).
>
>
>     What are actually the policies for indentation? I spent almost more
>     time fighting my text editor for indentation than writing the code.
>     The files use tabs instead of regular spaces for indenting like most
>     other projects? Maybe i have my tab-width wrong and therefore
>     struggle to get it right?
>
>
> Just like the kernel - hard tabs for alignment, tab width of 8.

Thanks! What a difference that tab width of 8 made, no more fighting 
with the editor :)

-mario

>
> Cheers,
> Daniel