>Given that swish has to do a complete search no matter how many
>items are to be reported with the results, it does seem like a feature
>that does not need to be inside the swish program itself.
I agree that it does not NEED to be in the program itself, but placing it
in the program would reduce the amount of processing time in a web-based
search for a couple reasons:
1) The compiled swish-e is faster than perl/php3/whatever. (Not a lot, but
it does make a difference.)
2) If it's handled outside of the program, then swish-e still has to do the
entire search AND the CGI has to parse the entire results in order to skip
the required number of entries. This also includes extra code in order to
keep track of what entry you're on and how many comment lines there are,
etc... This is ok when you're on the first page of results, but when
you're deeper into the pages it gets fairly slow.
In my opinion, it looks like a fairly easy feature to implement and it
would save time in parsing the search results.
Just my $.013 (two cents Canadian)
Received on Fri Jan 22 11:29:18 1999