I am using the Oracle9Dialect, 3.2.0.beta9a of Hibernate Tools, and am using JPA/EJB3 and Seam.

When reverse engineering Entity beans from the database, I have an annoying problem that I wonder if it could be fixed in Hibernate...

I have a few cases where the primary (or foreign) key of a table is a CHAR(x) field (typical for Country, Currency, etc.). When the Java code is generated, the columnDefinition is not overridden to be "char(x)", so upon deployment it breaks because Hibernate complains that the column is not what was expected (varchar2(x)):



I was wondering if the hbm2java code could be enhanced such that it would check if the columnDefinition matched the default columnDefinition for the field, and if not it would explicitly define it.

Right now, I need to edit the generated java by hand (which is particularly painful when you want to run seam generate-entities over and over again because it keeps overwriting my edits). I have only seen this problem for CHAR fields. Also note that I had this problem using SQL Server too, so it isn't related to Oracle.

Thanks!

Two issues here:

1) hbm2java could possibly (in most cases) include the column definition if it is different from the default mapping.

2) schema validator should be adjusted so it does not complain when not needed ( p.s. I don't understand why you are running shcemavalidator when you are using reveng ;)

I would say #1 is the easiest thing to do, but it would also be wrong since the char(1) is actually a perfect valid mapping for a char (same is varchar2(1))

#2 is what is needed to be fixed since it should check the compability of the type and not the "exactness" of the type.

As max noted in his reply, schema validator needed to be fixed. Until now, April 2008, this is not yet done, isn't it?

I experienced same problem. Should I report an bug in JIRA?