Float to Float4 Issue about oracle_fdw HOT 2 CLOSED

Yes, the problem is obvious and has to do with type names and their difference.
FLOAT in Oracle is the same as NUMBER, which is a "binary coded decimal" and can represent numbers with many significant digits fairly accurately.
FLOAT(n) limits the number of significant (binary) digits, while FLOAT doesn't.

FLOAT4 in PostgreSQL is a 4-byte floating point number in binary representation, and that means that the number of significant digits is quite limited. That explains the "rounding errors" you observe.

You would be much better off using FLOAT8 in PostgreSQL, or NUMERIC, if more precision is required.

The PostgreSQL data type that is most like Oracle FLOAT is NUMERIC, which is also a binary coded decimal.

from oracle_fdw.

Ahh, guess we should not have assumed that because of name similarity the data types are the same. Thanks for the great explanation. The FDW has been a great asset for our organization. Thank you.

from oracle_fdw.