On Thu, 8 Apr 1999, Aaron Ardiri wrote:
> i just looked at the API docs.
>
> WinRemoveWindow
> ---------------
>
> Purpose: Remove the specified window from the window list.
> ...
> Comments: Doesn't free the memory used by the window.
>
> nope.. that aint going to work.. thats for sure. from memory i
> remember the HardBall demo using WinDeleteWindow - and NOT
> WinRemoveWindow.
It doesn't crash when you use the Remove call.
Palm doesn't explain the differences and when to use each call, so I think
that is at the root of the problem.
> gcc has screwed up big time.
Actually it has:
Hardball.s (m68k... -S):
move.l (%a0,%d3.l),%d0
jbeq .L28
clr.b -(%sp)
move.l %d0,-(%sp)
trap #15; dc.w 0xa1f8
addq.w #6,%sp
.L28:
With -Oanything
Graphics.s :
move.l (%a2),%d0
jbeq .L10
clr.b -(%sp)
move.l %d0,-(%sp)
trap #15; dc.w 0xa1f8
.L10:
Note the missing "addq.w #6,%sp".
Removing the optimization makes it reappear, at least here
> any gcc people here? i want the source.. i fix this shit
> myself. - i bet they overlooked some of the asm("")
> implementations..
No, the optimizer is broken.
But removing -O5 didn't fix everything.
