[mohaa] Comments on SH server

[Ryan C. Gordon]

> The solution is very likely the same thing that was causing the crashes;
> the memory pool. malloc() and free() are very expensive when they get used
> for EVERY event that occurs (hint: WALKING generates events...everything
> passes events around the server)...doubly so as the address space starts
> to fragment. I think once I go and clean this up the memory pool code
> correctly, it'll use less CPU. I hope.

Oh, and...

...optimization is the last thing that happens. Then again, judging by the
response to alpha2, we're rapidly passing the debugging phase.  :)

I'm probably going to repackage this same binary as "beta1" tonight, so
people that are scared of alpha code will start to test it. There probably
won't be any changes between beta1 and alpha2.

--ryan.