[Xapian-discuss] query within value range

[Richard Boulton]

Mark Clarkson wrote:
> So for making the programmers life that much easier is the performance
> expected to be better or worse than boolean when using a value
> processor?

Yes. :)

That's a particularly hard question to answer, because it depends on the 
  context.  Briefly:

Hopefully, the value table will be quite small, so if it's being used 
reasonably frequently, it should get cached in memory quite well. 
Therefore, the overhead of using a value instead of a boolean shouldn't 
be too high.

On the other hand, the information required for a single search using 
the value table will be more spread out than a postlist for a single 
boolean term, so you if the value table hasn't got cached well, you 
might well find that searches are slower.

On the third hand, to do a range query using booleans will often require 
an OR of lots of boolean terms, so the value option might be quite a bit 
faster, due to having to read less from disk.

Finally, if your search only returns a few results, the value will only 
be checked for those results, so the overhead should be minimal.  If the 
search returns lots of results, the value could be checked quite 
frequently, so you might get a much higher overhead.

So - without more information, your guess is as good as mine.

I'm working on some performance tests at this very moment, which include 
some tests of the performance of a search using a value range 
restriction, so I'll let the list know what I find.

-- 
Richard