Re: [3.3] Patch ping

[Geoff Keating <geoffk at geoffk dot org>]

> Cc: rmathew@hotmail.com, gcc@gcc.gnu.org
> From: Tom Tromey <tromey@redhat.com>
> Reply-To: tromey@redhat.com

> >>>>> "Geoff" == Geoff Keating <geoffk@geoffk.org> writes:
> 
> Geoff> Please complete the unchecked items in the checklist:
> 
> Ranjit, could you do these things?
> 
> Geoff> [ ] The patch either fixes an existing testcase, or adds a new
> Geoff> testcase.
> 
> This is a curious requirement in this case.  I've been waiting to
> check in Ranjit's changes to libgcj until this patch is approved.  As
> long as the compiler and the runtime are consistent, JNI code will
> work if it is compiled against our runtime.  If I check in his runtime
> changes first (which are correct and bring us in line with the JNI
> ABI), then no JNI code will work.  Doing this would let us satisfy the
> "fix an existing testcase" requirement.

Yes, that'll do; the aim is just that there is some test for the new
code.  These are generic instructions, supposed to be suitable for any
patch, but there are always weird cases.

> I think in this case we ought to recognize that the situation is
> unusual.  Ranjit, please run your before- and after- tests against a
> libgcj with your stdcall patch applied.

Yes, that's fine.  The bootstrapping requirement is just to ensure
that some minor patch doesn't break the tree for everyone else.

-- 
- Geoffrey Keating <geoffk@geoffk.org>