[cp-patches] Stupid spec constant values

[Stuart Ballard]

(I'm not subscribed to classpath-patches so please cc me on responses)

Here are a couple of patches to deal with japi errors relating to
constant field values. In both cases they reflect stupidity in the
spec but since binary compatibility *does* require that constant field
values should match I think the japi errors are valid. Sun even
include these values in their documentation these days, ridiculous as
they are.

I'd prefer if someone else would commit these because I haven't
touched the classpath code in ages and I'm afraid of breaking things
;) I'd also like someone who understands the rmi code to check that
patch, because even though I changed the only code in Classpath itself
that uses the packagePrefix field, I'm not sure whether there are any
implications for code outside of Classpath that refers to it. (Note
that any such code would already be broken if compiled against the JDK
- my concern is whether *more* work is needed here to cover all the
cases).

Here's a changelog for the RMI one (hope I got the format right):

2005-10-04  Stuart Ballard  <address@hidden>

        * java/rmi/LoaderHandler.java (packagePrefix): Match Sun's value.
        * java/rmi/RemoteRef.java (packagePrefix): Likewise.
        * java/rmi/RemoteObject.java (readObject): Hardcode the GNU package 
prefix.

And for the swing one:

2005-10-04  Stuart Ballard  <address@hidden>

        * javax/swing/undo/StateEdit.java (RCSID): Match Sun's value.
        * javax/swing/undo/StateEditable.java (RCSID): Likewise.

Thanks,
Stuart.
--
http://sab39.dev.netreach.com/

classpath-rmi.patch
Description: Text Data

classpath-swing-undo.patch
Description: Text Data