Re: bitmap AM design

Neil Conway <neilc@samurai.com> writes:
> (BTW, is poor concurrency really the biggest issue with hash indexes? If 
> so, there is some low-hanging fruit that I noticed a few years ago, but 
> never got around to fixing: _hash_doinsert() write-locks the hash 
> metapage on every insertion merely to increment a tuple counter. 

Given the short amount of time that lock is held, this wouldn't
win anything worth noticing.  Also, it's not "merely" to increment a
counter --- the counter drives decisions about whether to split buckets,
so any decrease in accuracy would lead directly to losses in overall
performance.

The lack of WAL support is a much bigger issue.
        regards, tom lane