[Date Prev][Date Next]
Re: file backend
Jonghyuk Choi writes:
> How about support indexing through file offsets ?
> It seems not difficult to modify bdb/hdb index cache
> to hold file offsets instead of IDs.
Not sure if you mean search indexes or not having the entire database in
memory, but anyway I don't think so.
I don't know how bdb search indexes work, but I suspect it would be more
work to learn and implement them than I'm prepared to do for this
And without search indexes, a search must inspect every entry in the
database to find out if it matches - so if the database is not in
memory, it must in effect be rebuilt for every search.
With each entry in memory, this backend will perform like a bdb/ldbm
backend with unlimited cache but no indexes. So it should not be
used for very large files anyway.