> Thanks, but a zipped db is too slow for my needings.
Then db files in the classpath will not help.
The main problem for both zipped databases and databases in the
classpath is: you can't do real random access. You can only read from
a stream. Random access is required, so it is simulated like this:
When seek(pos) larger than current position, then a skip is made
(which is basically a read). If the seek pos is smaller, then the
stream is closed and re-opened. This is very slow.
> There is no way to make the performances of zipped db comparable to
> that ones of uncompressed files?
No. Not that I know of.
Regards,
Thomas