>>>>> "Paul" == Paul J Lucas <pjl@ptolemy.arc.nasa.gov> writes:

Hi again:

    >> I know, I use it and it's pretty fast, but still with big
    >> indices, or with small indices but lots of request, and being
    >> CGIs as they are, you could end up, as it's happened to us,
    >> with 10 or 12 copies of the index in memory.

    Paul> 	Are you sure about that?  The mapping parameter

Not really...

    Paul> specified in the mmap(2) call in file_vector.c is MAP_SHARED
    Paul> that, to my understanding, means that multiple processes
    Paul> will share a mapping of the same file; hence there is at
    Paul> most one copy in memory at any given instant.

OK, then we were thinking about the old swish-e, not the new
swish++. In any case, the Java solution is platform-independent, while
swish++ depends on having the mmap call in the system; the Mercurio
engine could run in any machine with a Java VM, including Windows NT,
or whatever.

Let's just say they are two alternative solutions to the same problem,
OK?

J

-- 
JJ Merelo                         | http://kal-el.ugr.es/htbin/jj-plex
Grupo Geneura ---- Univ. Granada  | http://kal-el.ugr.es/