Subject: Re: [htdig3-dev] Running 3.2.0b2: Compression, Score and Modiefied
From: Geoff Hutchison (email@example.com)
Date: Thu Apr 13 2000 - 21:13:45 PDT
At 12:47 PM +0200 4/13/00, Geschke Steffen wrote:
>Since v3.2.0b2 the option "wordlist_compress" is set true by default.
>However, htsearch and htdump doesn't work correctly with a
>compressed database (at least for a standard configuration).
Are ypu saying that you've tried it with wordlist_compress set to
false and you *don't* see this behavior?
>If the search is successful, the matching score is always 1%.
Actually, I think this may be an inadvertant result of some
adjustments I made to the scoring engine. Toivo Pedaste sent in this
report about a week ago:
At 12:44 PM +0800 3/29/00, Toivo Pedaste wrote:
>It turns out the scores from the words is values like 1.3 and 0.005
>while the backlink contributes values in the thousands. This is
>without setting any factors in the config files.
What I did was I set the word score to be divided by the word
frequencies. This is generally a good thing since it will make less
common words unimportant. However, I suspect in your case, all the
scores became 0 and hence were given a 1% ranking. :-(
I don't know what to say except you should try bumping up the _factor
attributes. Since you can modify them on the fly (no reindexing),
this shouldn't be too hard. I think we'd all be interested to know
what you find.
>The $(MODIFIED) variable doesn't work in output templates
>(for both, beta 1 and 2), although document list (printed
>by htdump) contains 'm:' values for all documents.
Huh. Do the m: values look like what you'd expect. Try something like this:
date --date "[fill in your value]"
(otherwise it's probably a really big number.)
I just wonder whether it's a bug in the htsearch code or the indexing
code. The dates come through fine for me though.
To unsubscribe from the htdig3-dev mailing list, send a message to
You will receive a message to confirm this.
This archive was generated by hypermail 2b28 : Thu Apr 13 2000 - 19:06:27 PDT