Skip to main content.
home | support | download

Back to List Archive

Some more questions (re: client-side search on Windows)

From: Greg Ford <greg(at)not-real.reddfish.co.nz>
Date: Mon Apr 14 2003 - 05:58:29 GMT
Hi all,

I have now got a working client-side search application running in =
Internet
Explorer for Windows.

I have a few more questions (and observations)

( Q.  1.)

Why aren't words near the end of my files being indexed?

i.e.=20
when I run :
    swish-e -c swish.cfg=20

where swish.cfg includes the following:=20
# swish.cfg starts
IndexFile docs.idx
IndexDir .
IndexOnly .htm .html
WordCharacters .abcdefghijklmnopqrstuvwxyz0123456789/
PropertyNames description
PropertyNamesMaxLength 1000 description
PropertyNames decisionid
PropertyNamesMaxLength 8 decisionid
PropertyNames year
PropertyNamesMaxLength 4 year
MetaNames description keywords year
# swish.cfg ends

over a corpus of around  2000 html files (legal decisions)
and then search for the word "dismissed"=20
(which should be near the end of around half of all the files) i.e.:

    swish-e -f docs.idx -w dismissed

I get only 79 hits...

The indexing process is apparently only indexing the start of the files?
I am running swish-e 2.3.4.
(if I search for words normally found near the start of the file I get =
thousands of results).

( Q.  2.)
One day ... can you rename swish-e/src/string.h ?

My application which is based on the libswishe currently has to=20
    #include "../swish-e/src/swish-e.h"

I would rather add the swish-e/src directory to the
global search path (and #include <swish-e.h>) - but I can't! with VC++.

When I add swish-e/src to the global list of include directories
(or as an additonal include dir for my project) the compile fails=20
- because swish-e/src/string.h is found ahead of  standard library =
(<string.h>).
As someone observed earlier  VC++ has a "broken build system... ", and
rather than fix VC++ it might be easier to rename string.h instead?


( Q.  3.)
Why do you call the -L (limit) option "experimental"?
I'm keen to use it, I have a fairly static/well defined corpus of HTML =
files.

( Q.  4.)
Is anyone urgently waiting for a working ActiveX wrapper for Swish-e ?
I've got mine working quite well now, (based on Swish-e 2.3.4).=20
This is similar to SwishEx in principle, but an entirely new codebase.=20
I use it with a client-side JavaScript, but it should work as a =
server-side ASP=20
too...
(It is in C++ using the ATL template library on MSVC++).
If you are interested, I would be happy to send a copy...

Anyway - thanks Swish-e developers for a great search engine,=20
and thanks in advance for any help on the above...=20

Greg Ford




*********************************************************************
Due to deletion of content types excluded from this list by policy,
this multipart message was reduced to a single part, and from there
to a plain text message.
*********************************************************************
Received on Mon Apr 14 05:58:39 2003