Bad Dialog size on large screens about facebook-android-sdk

[Korriged]

A patch:

1/ change the declaration of attributes DEFAULT_LANDSCAPE and DEFAULT_PORTRAIT in FbDialog with:

// Patch bug #30 - BEGIN
// Refer to http://github.com/facebook/facebook-android-sdk/issues/issue/30
static LayoutParams DEFAULT_LANDSCAPE = null;
static LayoutParams DEFAULT_PORTRAIT = null;
// Patch bug #30 - END

2/ Add the following code to method FbDialog.onCreate(Bundle savedInstanceState), just after super.onCreate(savedInstanceState);

// Patch bug #30 - BEGIN
// Refer to http://github.com/facebook/facebook-android-sdk/issues/issue/30
// Gets screen size (depends on screen orientation):
int height = ((WindowManager)getContext().getSystemService(Context.WINDOW_SERVICE)).getDefaultDisplay().getHeight();
int width = ((WindowManager)getContext().getSystemService(Context.WINDOW_SERVICE)).getDefaultDisplay().getWidth();
if(width > height) {
// Landscape
int temp = height;
height = width;
width = temp;
}
// Height = larger dimension
// Sets screen size: height-60; width-20
LayoutParams DEFAULT_LANDSCAPE = new LayoutParams(height-60, width-20);
// Sets screen size: height-40; width-60
LayoutParams DEFAULT_PORTRAIT = new LayoutParams(width-60, height-40);
// Patch bug #30 - END

[Pretz]

I have a slightly different solution than Korriged in my fork of facebook-android-sdk:
http://github.com/Pretz/facebook-android-sdk/commit/4d6d0f0d583c49c804b67c93073ef82e159b50b6
I've sent a pull request to Facebook so hopefully my changes will get merged upstream.

I scale the pixel values by the resolution density, but Korriged solution which insets the dialog from the screen size is also viable. However, it will be inset differently on high and low resolution screens, since 60px will be a different physical difference on an hdpi screen as opposed to an mdpi screen; probably the best solution would scale those values based on the screen density.

[soneff]

Hey, sorry for not being able to look into this sooner. I just read up on Android screen configurations, and I think that Pretz's solution is more robust (essentially density independent pixels approach). I'll pull your branch, test it, and try to push a change today.

[soneff]

I've tested on all the devices I have and it looks alright, so I've push this out. Thanks Alex! Let me know if you have any more problems / fixes.