Re: b-tree index search algorithms

Samuel Vogel <s(at)muel-vogel(dot)de> writes:
> I'm currently on a university research project if performance could be
> increased by substituting different inter-node search algorithms instead
> of the currently used binary search.

Hm, what have you got in mind exactly?

> But I'm having troubles understanding how the general b-tree
> implementation (nbtree.h) is used to represent for example a simple
> primary key on an integer column. I've debug printed the
> 'scankey->sk_argument' and all attributes of the index tuples on the
> pages being traversed (simply ran 'DatumGetInt32' on both) but I never
> see one of the integers actually appearing in my table being logged when
> I do a select.

Not clear what you did wrong from this amount of detail, but integer
keys ought to be pretty obvious at the debugger level.

> This is why I assume that all column values are hashed before they are
> pushed into the b-tree,

PG's b-trees do not hash anything. If you're not seeing interpretable
key values then you're doing something wrong in your inspection
methodology.

regards, tom lane