On Wed, 2004-02-04 at 13:36, Leopold Toetsch wrote:
> > As I understand it (and correct me if I'm wrong), SDL_keysym needs a
> > byte of padding on my architecture within SDL_KeyboardEvent.
> Brr. I don't know. I've to ask my debugger for that :) If you have
> SDL_keysym alone, that needs 3 bytes additional alignment. But don't ask
> me about the combined structure.
Exactly my worry. Does it depend on the compiler? Architecture?
Libraries? C specification? Phase of the moon?
> But I'd say that there are four adjacent Uint8 taking one word, then is
> another word (SDLkey sym). Or a structure starts aligned ...
I ran this program in lieu of the debugger:
#include <stdio.h>
#include <stdlib.h>
#include <SDL/SDL_events.h>
int main ()
{
SDL_KeyboardEvent kbevent;
printf( "0x%08X event\n", &kbevent );
printf( "0x%08X event.type\n", &kbevent.type );
printf( "0x%08X event.which\n", &kbevent.which );
printf( "0x%08X event.state\n", &kbevent.state );
printf( "0x%08X event.keysym.scancode\n", &kbevent.keysym.scancode
);
printf( "0x%08X event.keysym.sym\n", &kbevent.keysym.sym );
printf( "0x%08X event.keysym.mod\n", &kbevent.keysym.mod );
printf( "0x%08X event.keysym.unicode\n", &kbevent.keysym.unicode );
return(0);
}
Here's the output on my machine:
0x7FFFF7E8 event
0x7FFFF7E8 event.type
0x7FFFF7E9 event.which
0x7FFFF7EA event.state
0x7FFFF7EC event.keysym.scancode
0x7FFFF7F0 event.keysym.sym
0x7FFFF7F4 event.keysym.mod
0x7FFFF7F8 event.keysym.unicode
> We don't have a notion of nested structures. They could go just inline,
> i.e. without extra syntax. Pointers to structures are of course missing
> too, that's a different thingy.
Inlining's not a problem; it only has to be done once. This is one spot
where C's lack of abstraction actually helps. :)
-- c
