Lars Volker 0aec98f674 IMPALA-2749: Fix decimal multiplication overflow ...

When multiplying double and decimal values, we used to cast all doubles
to decimals before doing the multiplication. Due to the precision of two
decimals being added during multiplication, the effective value range of
the resulting decimal type could become very small and overflows could
happen.

This change switches the behavior to cast to double precision types when
at least one of the input operands is of type float or double. In such
cases we will not have exact results in general and we assume the user
would normally not expect exact results from an inherently inexact
datatype.

Change-Id: Idd28c5471506c68a860beb0778d98c8d25825f9f
Reviewed-on: http://gerrit.cloudera.org:8080/1820
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Reviewed-by: Marcel Kornacker <marcel@cloudera.com>
Tested-by: Internal Jenkins

2016-01-23 23:59:27 +00:00

2.4 KiB

Raw Blame History

View Raw