We have encountered a problem in persisting Java byte type to SqlServer. Basically Java byte has a range from -128 to 127, whereas SqlServer tinyint has a range from 0 to 255. We intend to address the mismatch by mapping Java byte to smallint.

So far, we have managed to customize the type mapping, and verified that it indeed created a table with smallint column. At runtime, we were able to save non-negative value to the column. However if we have a negative value (say -1), we would hit exception ‘[JDBC SQLServer Driver]Value can not be converted to requested type.’

I took a quick look at org.hibernate.type.ByteType, and see it is mapped to JDBC tinyint. Does that explain what we experienced at runtime? How do we influence Hibernate to map it differently at runtime?

Hi foggy,

I just have read

https://hibernate.onjira.com/browse/HHH-7060 and https://hibernate.onjira.com/browse/HHH-6815

today, and according what's written there, the problem is solved if you use Hibernate version 4:
if you use Hibernate version 4 then Java byte should be mapped to smallint.

Alternatively if you don't cant switch to hibernate4, you can also write and use a custom SQLServerDialect like following:



If you bootstrap the database with MyCustomSQLServerDialect then the byte-type properties should be mapped as smallint types.

Thanks, good to know that Hibernate is doing something on this issue.

We are still on Hibernate 3.3.2, which seems to predate SqlTypeDescriptor (3.6?). I will talk to folks to understand the impact of upgrading to 4.0 in the near future. One question about the proposed solution. Would that affect all Java types that are mapped to JDBC tinyint? It is conceivable that we would develop a UserType thatcan be losslessly represented by a tinyint, and it would inadvertently be treated as a smallint.

Yes, but concretely byte is the unique basic java type which is mapped to Standard SQL-build-in type TINYINT