nanogui@linuxhacker.org

> Ok, this is a known bug. It is because Mozilla may attempt to render
> images with a width or a height of zero. One way I found to stop this was
> to remove the assert commands from the fblin32.c file. The way I did this
> was wrap the assert.h include in an #if 0, #else then #define assert. ie

> I am hoping to find a better fix for this.

Peter -
    If you could, it would be great to get a backtrace as to exactly
which midlevel (Gd*) routine is on the stack when the low-level
driver assertion blows...  Then we need to return if h <=0 in that
routine, rather than the screen driver routine.


The reason that it's a bad idea to remove the assert is that the
linear framebuffer code could access outside of the mapped
display region and then you'll crash anyway or worse.

Regards,

Greg