[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kmail-devel
Subject:    Re: PATCH : scoring
From:       Guillaume Laurent <glaurent () telegraph-road ! org>
Date:       2001-04-10 18:22:22
[Download RAW message or body]

On Tuesday 10 April 2001 16:16, Michael Häckel wrote:

> First of all you should not score messages at all, if there is no scoring
> rule defined which affects the selected folder. That should at least cause
> no slowdown for folders without scoring.

That should be doable. We just need to add a way to check if there are rules 
for a given folder. :-)

> Otherwise simply the algorithm has to become faster :-)
> As far as I can see the score is only calculated out of headers and not out
> of the message text. That can become _much_ faster, if you don't parse all
> messages via mimelib into a KMMessage structure but read only the
> interesting part directely out of the unparsed message.

Indeed I had problems with that part. The only reason I used KMMessages was 
that I need a getHeaderByType(const QString &headerName) method. KMMsgBase 
didn't provide one.

> Assumed that the
> operatins with the content of the headers in libkdenetwork are fast.

The most expensive ones are regexp matches, AFAIK. That should be fast enough.

> You
> can do that in exactely in the same way, as it is already done in
> KMFldSearchRule::matches or you can move that 12 lines of code to a static
> funtion in kmmessage.cpp to reuse it.

Will look into that. Thanks.

-- 
					Guillaume.
					http://www.telegraph-road.org
_______________________________________________
Kmail Developers mailing list
Kmail@master.kde.org
http://master.kde.org/mailman/listinfo/kmail

[prev in list] [next in list] [prev in thread] [next in thread]