This is the mail archive of the
xconq7@sources.redhat.com
mailing list for the Xconq project.
Re: scorekeeping and materials
On Sun, 2004-05-09 at 07:46, mskala@ansuz.sooke.bc.ca wrote:
> Okay, here's a stack trace from loading crasher.g (as a regular game, not
> in the designer) and then immediately selecting "save". Made under Linux
> on a PC, kernel 2.6.4, gcc 3.2.2, current XConq CVS version as of a couple
> days ago.
Thanks.
> And here's one from loading crasher.g and playing (moving the single unit
> around more or less at random) until it asserts on turn 18:
OK. I'm looking into this one. The problem seems to be with something
your AI opponent is doing. I set the other side's player to nobody, and
was able run around for 530 turns without a crash. With an AI opponent,
the crash happens on the time scale you are seeing.
Interestingly, the AI doesn't even know about change-type (yet), and so
the action struct must somehow be becoming corrupt either during
prepping or in the interim between prepping and execution.
This could wind up being a "fun" bug to track.
> has minimal game importance. The crash during play looks like it might
> actually be a different bug, but it also seems to be linked to
> see-terrain-always.
I'll try with see-terrain-always set to true and see what happens.
Again, thanks for the simple test case, and the bug report.
Regards,
Eric