Hi!

I do not know how to correctly store scores into the TT,
especially when it comes to q-search and nullmove.
Is it true, that we can take the score we get, no matter
if its from a "normal" search or a q-search and proceed
the following way?

Simplified and pseudo-code :-)

int search(alpha, beta, depth)
{
flag = FLAG_UPPER;

...

for all moves
{
MakeMove();

if(deep_enough)
score = qsearch(-beta, -alpha);
else
score = search(-beta, -alpha, depth - 1);
if(score alpha)
{
if(score = beta)
{
HashStore(depth, score, FLAG_LOWER);
return score;
}
flag = FLAG_EXACT;
alpha = score;
}
}

TestMateOrStaleMate();
HashStore(depth, alpha, flag);
}

The "deep_enough" from above could be something like depth 1.

And what about a null move search?
Before testing any moves we test the null move. Can this be done in a
similar manner as above?
(Again simplified)

....

if(deep_enough)
score = qsearch(-beta, -alpha);
else
score = searchnull(-beta, -alpha, depth - 3);

if(score = beta)
{
HashStore(depth, score, FLAG_LOWER);
return score;
}
if(score alpha)
{
alpha = score;
}

// go on with the "real" moves
....

And how would the "deep enough" be in that case?
depth 3?

And what about storing depth-1 or depth-2 in the TT
in the fail-high case above, as in a certain sense it is not
a depth-search?

Thanks,
-delphi-