NaN values will cause insertions to fail in Oracle 11/12
Closed this issue · 1 comments
jprafael commented
The OracleTranslator
is currently converting DbColumnType.DOUBLE
to oracle's DOUBLE PRECISION
. According to oracle's documentation we should be using BINARY_DOUBLE
instead since that is what conforms to the IEEE standard - also what is supported by Java and many other databases.
As a consequence of not doing so we fall into the limitations of DOUBLE PRECISION
- No support for NaN, -Inf or Inf.
- Different floating point handling which caused #18
Changing the column format and a little bit more code in OracleEngine.entityToPreparedStatement
fixes this.
This however has the downside of being not backwards compatible and potentially requiring changes in multiple client's DBs.
Let me know how you want to handle that.
dspereira004 commented
Closing this as it was already merged almost two years ago.