Re: [gnugo-devel] dfa size reduction

[Dave Denholm]

Paul Pogonyshev <address@hidden> writes:

> actually, it doesn't reduce the size of dfa, but the size of the list of 
> attributes. i just noticed that there are _very_ many repeating pairs in
> those lists and hacked in a new function named compactify_att() into dfa.c.
> it calls itself recursively, removing all repetitions found.
> 


FWIW, I think I did mention that one way to reduce the number of duplicates
being generated is to use one attribute list shared between the intermediate
dfas. Would make dfa generation slightly faster, but probably not significantly
so. And it would change the meaning of the intermediate size changes slightly.


dd
-- 
address@hidden          http://www.insignia.com