[IMPALA-404] Bigint and float's compatible type is double.

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 1.0.1
• Fix Version/s: Impala 1.1
• Assignee: Alexander Behm
• Reporter: Alexander Behm
• Created: 07/Jun/13 17:08
• Updated: 15/Jun/13 00:45
• Resolved: 15/Jun/13 00:45

Description

Our compatibility matrix states that bigint is compatible with float if both are promoted to double.
This seems incorrect to me because a bigint can be converted to a float without loss of precision.

This causes problems for, e.g., inserts. The following query will fail to analyze, although there is no danger of losing precision.

insert into float_t values (cast (123 as bigint))