[RFC] "C" header options

[Mark Mitchell]

>>>>> "Gabriel" == Gabriel Dos Reis <gdr@codesourcery.com> writes:

    Gabriel> For functions whose siignatures are changed by C++, that
    Gabriel> technique won't work, and <xxx.h> won't define their
    Gabriel> names in std::.  Are you suggesting, we postpose that
    Gabriel> after gcc-3.0?

Yes.  See my subsequent post where I illustrate the KCC technique.
And claim that it is both simple enough and good enough to be our best
hope at this point.

    Gabriel> As said earlier, <string.h> should be implemented in
    Gabriel> terms of <cstring>, not the other way.

Oh, definitely.  I'm trying to get poor Benjamin off the hook here by
getting something for 3.0 that is plausible -- even if known to be
imperfect.  I think the issues are complex enough that another round
of discussion will be required before we know exactly how to get this
100% right -- if that is even possible without cooperation from the C
library.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com