[R] Getting Keypressed State in rgl

[Duncan Murdoch]

On 2/29/2008 7:16 AM, Duncan Murdoch wrote:
> Dieter Menne wrote:
>> Dear Rglers,
>>
>> when using a callback in rgl (Windows, if it matters)
>>
>> http://finzi.psych.upenn.edu/R/library/rgl/html/callbacks.html
>>
>> I would like to get the Asynchronous keyboard status (as least
>> Shift/Control, but preferably other key) to set markers in a plot.
>>
>> getGraphicsEvents seems to be limited to normal graphics windows.
>>
>>   
> That's a general hardware problem, not really an rgl problem: you just 
> want to query the keyboard.  It definitely depends on the platform.  
> Windows has GetKeyState, GetKeyboardState, and GetAsyncKeyState, each 
> with slightly different behaviour.   For example,
> 
> GetKeyState(VK_SHIFT);
> 
> will tell you if either shift key is down.  There are also VK_LSHIFT and 
> VK_RSHIFT constants to distinguish the two shift keys.  Here's an inline 
> version (untested):
> 
> 
> isshifted <- cfunction(signature(result="integer"),
>                        includes = "#include <windows.h>",
>                        body = "result[0] = getKeyState(VK_SHIFT)",
>                        convention = ".C")
> 
> 
> Duncan Murdoch
> 

Now I'm on Windows, I can see that doesn't work.  Here's a version that 
almost does:

library(inline)
isshifted <- cfunction( includes = "#include <windows.h>",
              body = "return ScalarInteger(GetKeyState(VK_SHIFT));")

The problem with this one is that it puts "#include <R.h>"  ahead of 
"#include <windows.h>", and that causes a compile error because of a 
clash in the use of names.

If you edit the generated source you can fix this, but I suppose 
cfunction should allow you to put the windows include first, and I 
suppose R should try to avoid clashing with it.

Duncan Murdoch