607332 - [Simple Search] should reduce the results with more keywords

Status: RESOLVED DUPLICATE of bug 260856

(Bugzilla :: Query/Bug List, defect)

Description

[MrX1980]

User-Agent:       Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b8pre) Gecko/20101026 Firefox/4.0b8pre
Build Identifier: 

Hello

the Bugzilla simple search should reduce the search results with every keyword.

The help say:
"Adding more terms narrows down the search, it does not expand it. (In other words, Bugzilla searches for bugs that match all your criteria, not bugs that match any of your criteria.)"
https://bugzilla.mozilla.org/page.cgi?id=quicksearch.html

That is not right at this moment.

Reproducible: Always

Actual Results:  
one keyword = few results
two keywords = more results
many keywords = many results

Expected Results:  
one keyword = many results
two keywords = less results
many keywords = more less results

Comment 1

[Frédéric Buclin]

Simple search != quicksearch. Both are two different things.

Comment 2

[MrX1980]

Ok, but "simple search" should however reduce the search results.
Sometimes with five keywords I have over 20.000 results.
This make no sense

Comment 3

[Max Kanat-Alexander]

Huh, you are right, and that should not be happening.

Comment 4

[Frédéric Buclin]

(In reply to comment #3)
> Huh, you are right, and that should not be happening.

Why not? If you have more keywords, you have better chance to catch some given bug, which is why you get more and more results. But they are ordered by relevance anyway.

Comment 5

[Max Kanat-Alexander]

@LpSolit The fulltext index is supposed to reduce your match results based on your input. It could be that MySQL's fulltext engine works in some different way than I expect, in which case this would be intentional and we would WONTFIX it, but I have to at least look into it to see if this is some bug of ours.

Comment 6

[Frédéric Buclin]

Maybe you need to combine both MATCH() into a single one?

http://dev.mysql.com/doc/refman/5.1/en/fulltext-search.html#function_match

suggests to use MATCH(col1, col2).


Unrelated, but could maybe make the query a bit faster:

http://dev.mysql.com/doc/refman/5.1/en/fulltext-natural-language.html

says that using MATCH() in the WHERE part automatically sorts results by relevance. This way, you wouldn't need the ORDER BY relevance part, nor the MATCH() as relevance in the SELECT part.

I didn't check if that's true for Pg and Oracle.

Comment 7

[Frédéric Buclin]

Dupe of bug 260856, see myk's comment 5 there.