Re: [Adonthell-devel] pathfinding & walking

[Kai Sterker]

On Sat, 10 Aug 2002 22:15:48 +0200 Alexandre Courbot wrote:

> > > Mmm, looks like it segfaults after running for a while (15-20
> > > minutes for me, tested two times). I don't know much, but the
> > > backtrace is quite weird:
> > 
> > Hm. I'll check that. Never had it running for that long so far.
> 
> Ah, actually my gdb is unable to read GCC 3.1's symbols. I recompiled
> Adonthell with 2.95 and it's much more clear now:
> 
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 1024 (LWP 22304)]
> time_event::execute (this=0x8a09358, address@hidden) at
> time_event.cc:70
> 70        if (Repeat != 0) Time += Interval;
[...]
> Well, it's still weird anyway!

Yeah. Not exactly the place where you'd expect a segfault. I think I
managed to reproduce it once, unfortunately without gdb running. Now
I've been waiting for hours and it works just fine.

At least I've converted some more schedules in the meantime. All but the
most difficult (Bjarn, Erek, Orloth, Oliver and Sarin) are using events
for walking as of now. The rest is disabled for now, And the best thing
is: Adonthell is running absolutely smooth on the P200 :). With sound
disabled it doesn't even use all the CPU anymore! So it'll be great on
the Zaurus as well, I expect.

I commit once again, the rest should follow tomorrow. Hopefully with
above bug fixed.

There is one more thing. Right now, the exacts states of NPCs aren't
saved or loaded. But I'll intend to change that as well. At the moment
loading/saving is be a bit rough.

Kai