2008-11-28 17:40:24

ruud
Developer emeritus
omega
Real name: Ruud van Melick
From: a galaxy far far away
Known languages: li,nl,en,(de)
Website

Re: Feedback to: Help us test the release candidate for the upcoming 4.0.7

Victor, what if you change this line:

$search = " and ($cols) $s_filter";

into:

$search = " and (match (`'.join('`, `', $cols)."`) against ('$q')) ".$s_filter;

Does that search case-insensitive… and which MySQL version are you using and what is the collation and charset of your TXP tables and is the db_charset set to utf8 in your config.php?

This really belongs in a separate topic btw, because this is not something that changes in 4.0.7.

Last edited by ruud (2008-11-28 17:41:04)

Offline

 

2008-11-28 17:57:45

wet
Developer
omega
Real name: Robert
From: Lenzing, Austria
Known languages: de-AT, en, PL/M-80
Website

Re: Feedback to: Help us test the release candidate for the upcoming 4.0.7

I think we’ve been there, done that. Wasn’t too performant, IIRC.

Offline

 

2008-11-28 18:19:25

ruud
Developer emeritus
omega
Real name: Ruud van Melick
From: a galaxy far far away
Known languages: li,nl,en,(de)
Website

Re: Feedback to: Help us test the release candidate for the upcoming 4.0.7

That’s a mixed RLIKE / MATCH solution. I was referring to a pure MATCH solution, which should be much faster, but relies completely on MySQL fulltext search for results (which does have some limitations: minimum search word length and skipping words that occur too often)

Offline

 

2008-11-28 18:22:39

wet
Developer
omega
Real name: Robert
From: Lenzing, Austria
Known languages: de-AT, en, PL/M-80
Website

Re: Feedback to: Help us test the release candidate for the upcoming 4.0.7

ruud wrote:

which does have some limitations: minimum search word length and skipping words that occur too often

I think this is why Alex split the search in two parts. Word length was an annoying issue.

Offline

 

Powered by FluxBB