Re: WinRT changes for review

On Sun, Oct 2, 2016 at 3:04 PM, hardcoredaniel <[email protected]
(mailto:[email protected])> wrote:
"



"
1. I have not found anything useful to put into ddpi. I filled in the value 
for "ResolutionScale", but I do not know whether the raw DPI values need to 
be scaled anyway - so this was experimental only. I can also put in the mean
of rawdpix and rawdpiy instead.

"



If those values give return values that are equal to the values returned by 
the Win32 backend, I'd generally be fine with this.






"



No, currently they are not. I put in the numeric value of this enum instead:




https://msdn.microsoft.com/de-de/library/windows/apps/windows.graphics.
display.displayinformation.resolutionscale.aspx




My impression was that the DPI values need to be scaled by this. (Like iOS 
physical DPI changes depending on whether you request a highDPI window or 
not). But maybe this is not true.




If these values




https://msdn.microsoft.com/de-de/library/windows/apps/windows.graphics.
display.displayinformation.rawdpix.aspx

https://msdn.microsoft.com/de-de/library/windows/apps/windows.graphics.
display.displayinformation.rawdpiy.aspx




are always correct, then I can simply copy&paste from the Win32 backend to 
compute ddpi.


 
"




 
"

2. I would be fine with a hint, one reason less to maintain an out-of-tree 
SDL :-) Last time we discussed it we did not conclude whether the hint is 
necessary, or triple-buffering should be enabled by default, because the 
additional memory consumption is not relevant nowadays.

"



I've started work on a patch (for a hypothetical, SDL_HINT_RENDER_BUFFER_
COUNT), and will try to post it to Bugzilla soon, for review.




"



Cool :-) Please announce your patch on the list as well so I can take a 
look.


 
"






 

"

3. The implementation from the patch will only try to set the 
"PreferredLaunchWindowingMode" property if the switch to windowed/fullscreen
worked. An implementation outside SDL would have to query the fullscreen 
state first (to check whether the call worked) and then set the property 
accordingly. 
However, if the property should be set independently of the result of the 
function call, it can be set externally. 
Too bad that "SetWindowFullscreen" cannot return success or failure...

"



Having internal SetWindowFullscreen functions be able to report errors 
sounds interesting.  I'm wondering if there are downsides here, such as 
quirky behavior on select platforms (I vaguely recall OSX-fullscreen'ing 
being async, for example), that would prevent this operation from making 
sense, in some cases.  :-/




"



Hm, in that case a "get()" function for the fullscreen state would probably 
be a better idea than a success/failure return value from the "set()" 
function.

 
"









-- David L.







_______________________________________________
SDL mailing list
[email protected]
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org"

_______________________________________________
SDL mailing list
[email protected]
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org