Bug#688934: Another way to trigger this bug

[Sylvain]

Hi Bas,

Thanks a great lot for all your effort on this nasty bug, despite my
lack of responsiveness.

I took a few hours to investigate what exactly happens (adding a few
code comments in the process), and I see that in your testcase dmod,
the warp is an invisible sprite which is merged in the background and
thus never found by find_sprite().

If spr[0] is clean, the warp continues despite not finding the warp
sprite; if spr[0] is modified, then we hit the bug, as the code
assumes the warp animation is not finished.

I'm not sure in what way an unclean spr[0] will affect the game, but
it might, and cleaning it would affect compatibility with the original
Dink.

Consequently I used and documented your original fix from
http://www.dinknetwork.com/forum.cgi?MID=168476#168476

http://git.savannah.gnu.org/cgit/freedink.git/commit/?id=fff4b7cb8d6d2bf84482fdc83c2a21fd9d0379e1
http://git.savannah.gnu.org/cgit/freedink.git/commit/?id=402bf38f69cd7e5c47322b8087ba535d6f823283
http://git.savannah.gnu.org/cgit/freedink.git/commit/?id=5d692b447eb7a8b3c0f1128390641992a4dff484

WDYT?

(Btw, I don't have the same line numbers in my source files, possibly
you added test code around? ;))

- Sylvain

On Sat, Oct 13, 2012 at 01:11:19AM +0200, Bas Wijnen wrote:
> I found another way to trigger this: again with max sprites on screen,
> hit a duck. The attempt to create the flying head will again fill up
> spr[0] and trigger the bug on warp.
> 
> The problem in this case is in duck_brain in freedink.c.