Is Minimal Perfect Hashing the wrong algorithm?

Many compiler writers use Minimal Perfect Hashing to generate some
hash tables such as a key word table.


But this algorithm minimizes table size. Shouldn't it minimize lookup
time with a cap on table size? That is shouldn't it look at the time
required to create a hash from an entry?


And in this light maybe synonyms aren't so bad if the second entry in
the bucket is something rare and misses are also rare.
--

Return to the comp.compilers page.
Search the comp.compilers archives again.