[gambit-list] Gambit hash table cost?
sthilaid at iro.umontreal.ca
Thu Sep 4 09:46:06 EDT 2008
Recently, Marc had a talk on gambit table's weak pointers and he also talked
about eq tables which can be obtained by (make-table test: eq?). He also
mentionned that any non-gambit test used will require to write our own hash algo
(which is not something I want hehe).
I often want to use as my table keys s-data like '(apple tomato) or '(1 2 3)
which are lists where each elements can be compared with eq?. Now my question is
the following: What is most efficient, to use such keys in regular tables (that
uses equal? as its test) or to nest eq tables that will have as key only of of
the field, thus resulting in a kind of n-dimensionnal matrix table system...?
It would be conveniant if there we be a test supported by the tables (that would
possess a corresponding hash algo) that could verify lists with members that can
be compared with eq? wouldn't it? hehe
More information about the Gambit-list