On Sat, 2005-12-10 at 20:16 +0100, Guilhem Lavaux wrote:
> Hi,
>
> Here is a patch which (partially) fix a regression in kaffe. If you
> define a class like that:
>
> class A
> {
> private static final ObjectStreamField[] serialPersistentFields =
> {
> new ObjectStreamField("i", int.class),
> new ObjectStreamField("i", int.class)
> };
>
> /* ... */
> };
>
> then the JDK throws an InvalidClassException in ObjectStreamClass.lookup.
> This goes against the API. However it is true that we must take into
> account the presence of duplicate fields. So I propose we throw an
> InvalidClassException on writing the object if this sort of situation
> happens.Why not do similar and throw the InvalidClassException from ObjectStreamClass.lookup()? (But then document that clearly in our version of course.) If you do like in your patch then at least document this behavior. It is slightly confusing that ObjectStreamClass.lookup() returns a non-null result, but getFields() suddenly returns an empty ObjectStreamField array as if there are no Serializable fields. Now user code has no real way to check whether that is because the ObjectStreamClass is invalid (because of bogus serialPersistentFields) or because it just has no serializable fields. Cheers, Mark
signature.asc
Description: This is a digitally signed message part
_______________________________________________ Classpath-patches mailing list Classpath-patches@gnu.org http://lists.gnu.org/mailman/listinfo/classpath-patches
