[gnugo-devel] May I ask a question about the hashing in gnugo 3.6

[Kai Wang]

Hi,
   I'm a student currently working on SlugGo project. I have some
quesitons about the hashing data structure changes in GnuGo 3.6. Since I
couldn't find the answers in GnuGo's documentation website, I think here
might be the right place to address my questions. If it's not, please
tell me or simply ignore this email.
    I found out that the hashing data structures have changed a lot
in GnuGo 3.6 compared to GnuGo 3.4, such as the removal of
the "Hashposition" data structure, the removal of
the "FULL_POSITION_IN_HASH" scheme, and the replacing of the "Read_result"
field which contained two unsigned ints with  the "data" field which is
only an unsigned int in "Hashnode" data structure.
    What's the collision occurence rate of the hash table? Given that the
"FULL_POSITION_IN_HASH" has been turned off, what would GnuGo 3.6 do
if there is a collision? And why the "Read_result" field was replaced
with the smaller "data" field.

Thanks a lot,

Kai