Hi,
I've noticed that the result of the sort expression for sphinx matches is a
*float* value. I'm really disappointed, because *float* cannot "represent"
as much data as I would like to.
I need to do something similar to SPH_SORT_EXTENDED mode, however one
column values are slightly influenced by another column. I want to achieve
something like that:
2^33 * *@weight* + 2^32 * *attr_a* + *attr_b* - (*attr_c* * 10000)
(boolean) (int) (tiny int)
where attr_a is boolean, attr_b is int and attr_c is a tiny number.
This is quite similar to:
*@weight* DESC *attr_a* DESC *attr_b* DESC
I tried to do something like this:
2^17 * *floor*(8 * *sqrt*(*@weight*)) + 2^16 * *attr_a* + *sqrt*(
*attr_b* - (*attr_c* * 10000))
However, this approach was still not sensitive enough to properly sort by
either @weight or by attr_b - (attr_c * 10000), because the differences
between weights or between values of attr_b - (attr_c * 10000), therefore,
the @expr for many matches was rounded to the same float (the gaps between
consecutive values that can be represented by float are too large. In
consequence, lots of results have the same @expr value).
Do anyone know some workarounds, solutions, etc.?
Thanks in advance! I really appreciate all suggestions.
Cheers!
--
You received this message because you are subscribed to the Google Groups
"sphinx-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sphinx-users.
For more options, visit https://groups.google.com/d/optout.
