Hello Marina,
The fix was so that entities using byte[] and char[]'s as pks can be cached
corretly to maintain object identity, and has nothing to do with platforms.
The test I checked in to prove it works though requires an entity with
byte[] as the pk. Oracle does not allow blobs or long raws as primary keys,
so I had to write the table creator to create the table using a RAW type -
which may/may not work on other databases.
Using byte[]/char[] will work using java2db on any database if the type it
converts to is allowed to be a pk on that database. Oracle is one that it
won't work on because byte[] is mapped to long raw, hence I created the
field with a Raw instead of letting the platform choose a type.
Regards,
Chris
----- Original Message -----
From: "Marina Vatkina" <Marina.Vatkina_at_Sun.COM>
To: <persistence_at_glassfish.dev.java.net>
Sent: Wednesday, January 10, 2007 1:31 PM
Subject: Re: fix for 1193
> Chris,
>
> I'm not sure I understand the comment.
> Does this change enables byte[] and char[] pk support for Oracle only?
> Will java2db work for the above types? Will it work on Oracle only?
>
> thanks,
> -marina
>
> Christopher Delahunt wrote:
>> Short Description: bug fix 1193 allowing byte[] and char[] types to be
>> used as pks
>> Code reviewed by: Tom Ware
>> Tests added: PrimitiveArrayPKCachingJUnitTestCase
>> Comment: Fix as suggested by Adam Leftik. The tests only configured to
>> run on Oracle, since the table creator for the test uses a raw type for
>> the pk.
>