Re: java.rmi.server patch

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Frank" == Frank Ch Eigler <fche@redhat.com> writes:

Frank> Assuming the "-" side of the diff is the new version, looks
Frank> okay.

Thanks.  (Your assumption is correct in this case, the `old' part of
the diff is classpath, but for this class they are ahead of us.)

Frank> How strict are you (classpath + libjava) trying to be with matching
Frank> the signature of Sun java.* functions, with respect to "throws"
Frank> clauses?

We have to match them.  However, we don't need to declare unchecked
exceptions (and our coding standard is to leave them out), and the
order doesn't matter.  RMI might not conform to our standard here,
since it was just a big code drop from kaffe, which had its own coding
standard.

>> We've been a bit lax on merging Classpath changes back into libgcj
>> recently.

Frank> If they are the upstream source for this neighbourhood, would it have
Frank> been more appropriate for me to submit my patch to classpath
Frank> first/instead, and then merge?

There's not really a Classpath patch submission process.  You can put
them in the Classpath patch database, but review there seems to be
more spotty than even this list.

Tom