Skip to main content.
home | support | download

Back to List Archive

Re: PHRASE SEARCH

From: Jose Manuel Ruiz <jmruiz(at)not-real.boe.es>
Date: Thu Apr 13 2000 - 16:18:36 GMT
Bill,

I have downoladed all the fixes. They will be aplied in
swish-e-1.3.2-PHRASEd

 
> Did you look at the http://sunsite.berkeley.edu/SWISH-E/Patches/ directory
> for any patches that could/should be included?  I'm not sure is SRE
> included any in his version or not, other than the stemmer.c bug.

This one is also useful. I will take a look at it. Perhaps, the first
thing
to do is implementing sorting by the Metaname of yor choice. 
I am moving part of the code from linked lists and binary trees to 
arrays in order to improve performance in long lists of results,
so sorting may be easier and faster. So spliting, may be easier, but
you have to repeat the search every time you get a portion of results... 

> One patch that would be nice is the one that provides paging within swish.
> I don't like the way I page right now in my programs -- I read the entire
> results back, sort the results, and then splice out my page.  I'm not sure
> if that could be setup in swish due to the special sort I'm using (I sort
> on data returned in ProperyNames).  My method also means I'm reading a
> rather large chunk of data back into the web server (running mod_perl) and
> that memory is used for the live of the server process.  So it would be
> nice if swish only returned the page I'm interested in.


This sounds really well but although I use Apache here, I do not know
how to write
an apache or perl module. Long time ago I wrote a basic authentication
module
for Apache and Oracle but Apache version was 1.1 and modules may have
changed a lot. 

I have just discovered Perl only two months ago. Of course, I have heard
of it but
I have not used it until now. I have always used C instead.

> 
> Someone once discussed on this list making swish-e an Apache module.  I'd
> love to try out that.  Swish will always be limited in speed as long as a
> fork has to be done for each request.  Might be better memory wise to have
> a swish-e daemon, but probably much easier to make it an Apache module, or
> even just an xs extension for perl that can be loaded into Apache with a
> PerlModule directive.
> 
> I've made simple xs modules before (for example, I've got the swish
> Stemmer.c module available to perl that I use for highlighting results),
> but I'm not good enough, probably, to make swish-e a module.  Someday I'll
> look at that.
> 

-- 

Jose Manuel Ruiz Ramos

jmruiz@boe.es
Received on Thu Apr 13 12:23:34 2000