I am new to this list, but I am using SWISH-E successfully in indexing over
19000 files.

Those files are in different directories and there are no repeated files. I
generate the index for each directory and then I merge them... This last
step uses a lot of memory (it seems something like 200MB).

I believe it would be easier if could have "absolute" ranking values
instead of relative ones (normalized to 1000). Then I could just make the
index for each directory, make multiple searches and choose the best ones
and normalize them in a "post processing" program.

Is it possible? Would that be useful to more users? Is it possible to put
that as an option (say a -A flag)?
That would allow other uses, like ignoring matches below a given value
(otherwise, even if low, they are normalized to 1000)...

Best regards

Jose Carlos



--------------------------------------------------
GlobalPartner Ltda.
Al. Jau, 1494
01420-002   Sao Paulo  SP
Tel.: (011) 881-6455
FAX:  (011) 280-0783

Jose' Carlos R. de Andrade
Diretor

andrade@globalpartner.com.br
--------------------------------------------------