[Xapian-discuss] Omega: behavior msize when collapsing results

[Olly Betts]

On Wed, Mar 24, 2010 at 04:48:41PM +0100, Jeroen Vaes wrote:
> So, what causes this behavior? Is this correct (in that case it would
> seem that the API documentation is wrong), or did I encounter some weird
> bug? And does anyone have a solution?

It sounds like a bug.  We've had a few in this area before, and getting these
statistics right under all the possible combinations of options while doing
various optimisations is somewhat tricky.

I believe we've backported all the relevant fixes to 1.0.x, but it would
be useful to check with 1.1.4 to see if that exhibits the same issue in
case we missed one.

If it fails with 1.1.4, then we really need a testcase - unfortunately it's
hard to track down such bugs without one.  Is the dataset something you can
make available, and small enough to do sensibly do so?

Cheers,
    Olly