Mail Archives: djgpp/1997/02/10/05:45:01
Douglas Eleveld writes:
>When I draw onto a winddow, I first check if the mouse is in the window
>and if it is not, then I just draw, and if it is in the window, I erase the
>mouse before drawing in the window. That means that sometimes I'll be
>drawing on the screen with a visible mouse pointer, but not over the
>drawing region. Is this dangerous? I haven't had any problems so far.
It depends on the video mode. In mode 13h or linear framebuffer SVGA
modes it is quite safe, but in banked SVGA modes you could run into
difficulties. The problem is that the mouse pointer is drawn in a timer
interrupt, which could preempt your drawing code. Say you call
rectfill() to draw something at the bottom of the screen. This selects
the appropriate bank for writing to the bottom of video memory, but what
if the user moves the mouse in the middle of the rectfill() operation?
The mouse drawing code will switch banks to draw the pointer up at the
top of the screen, so when it returns from the interrupt and resumes the
rectfill, the wrong bank will be selected. The worst that can result is
garbage on the screen, though: there's no way it can crash your program.
A possible fix would be to make the mouse code save and restore the
current bank, but this is buggy under several VESA drivers that I've
tried. Another option would be to disable interrupts while you do the
drawing, but that isn't a good idea unless you are going to finish and
reenable them _really_ quickly...
/*
* Shawn Hargreaves - shawn AT talula DOT demon DOT co DOT uk - http://www.talula.demon.co.uk/
* Ghoti: 'gh' as in 'enough', 'o' as in 'women', and 'ti' as in 'nation'.
*/
- Raw text -