GCJ problem on RS/6000

Tom Tromey tromey@redhat.com
Fri Feb 8 15:10:00 GMT 2002

>>>>> "David" == David Edelsohn <dje@watson.ibm.com> writes:

David> Other than convenience, I do not see any reason that GCJ and
David> G++ need to generate $ in symbol names.  GCJ could translate
David> everything rot13 or EBCDIC because the spec cannot impose
David> requirements at that level.

I agree.

David> 	Allowing $ in identifiers is one thing, but blindly copying
David> them through to assembly files when NO_DOLLAR_IN_LABEL is
David> defined creates big problems for portability of GCC, G++ and
David> GCJ.

Please see Bryce's post on the subject:


I think Bryce's approach here sounds good.  The corresponding change
to g++ would have to be made, as we require compatible mangling.

See also the discussion that occurred when we last changed this part
of gcj:



More information about the Java mailing list