Posted by info on 05/29/07 12:46

Thanks for your reactions,
There is this problem with Srinivas' solution and the solution Robert
supplies is actually what I already proposed myself. Namely separating
the case where all columns are null from those cases where some ar not
null, and this is very slow. (I'm talking about hundreds of columns
and millions of rows.)
I was actually thinking more of a solution to ignore the nulls, rather
than on the fly setting them to 0.
More suggestions are appreciated.
Regards,
Paul

[Back to original message]