As Jose says, I've been using the incremetal option and my experience is
very nice. I use this option to index PDF files, and It's very useful to
me(saves a lot of time).
A feature I miss in this option, is the one to remove a file from the
index. I actually filter the search result with a file content, to avoid
appearing these files in the result.
Anyway a think it's a very interesting option, and It makes swish-e
stronger. As a have said to Jose, I think It's way we have to explore and
Is this helpful?? What do you think??
P.D. : Sorry because of my poor English. ;-)
>CVS 2.5 version contains the fix. It also has some other
>fixes for bugs reported by Ander (thanks Ander).
>Test with current cvs version. But keep in mind that this piece
>of code is very experimental. Perhaps Ander can tell us about
>On 6 Jan 2004 at 13:01, Bill Moseley wrote:
>> On Tue, Jan 06, 2004 at 06:58:38AM -0800, Patrick O'Lone wrote:
>> > Hello,
>> > What does the --enable-incremental option do? There's really no
>> > documentation around about it. Also, in src/config.h, there is an
>> > option called USE_BTREE - is that experimental as well or any better
>> > for larger indexes? Thanks for any responses!
>> --enable-incremental enables the USE_BTREE backend, and allows
>> *adding* new data to the index (there's no way to "update" the index
>> by removing old data). It's should be useful for things like mail
>> archives that continue to grow.
>> > BTW - I couldn't compile the program with the --enable-incremental,
>> > it died while compiling src/db_write.c, on line 743. I'm sure it's
>> > because a pointer to SWISHE *sw is not available locally to the
>> > setTotalWordsPerFile() function call.
>> Yes, I see that error, too. Jose, can you take a look?
>> Bill Moseley
Txat euskalduna >>> http://www.euskalerria.org/solasgunea
Received on Thu Jan 8 08:42:38 2004