Another mangling loose end

[Matt Austern]

Martin von Loewis writes:
 > > Probably the simple solution: add linkage as a qualifier in function
 > > types.
 > 
 > Yes, that sounds simple.
 > 
 > > Issue: do we introduce mangling for C and C++ linkage only, or do
 > > we handle other kinds of language linkage as well?
 > 
 > Definitely not. Please have a look at Core Issue 13
 > (http://www.informatik.hu-berlin.de/~loewis/corer9.html#13), where one
 > of the suggested resolutions is to deprecate that feature (using
 > linkage specifiers on types).

But Bill isn't suggesting that the committee should revisit the
decision to make linkage part of the type.  All he's suggesting is
that we change the syntax for attaching linkage to function types.

Bill's suggestion may be a good one; linkage specification for
function types is one of the ugliest parts of C++ declaration syntax,
which is saying a lot.  I don't think it affects our decisions,
though.  It's a source language issue, and the ABI issues will remain
the same.

                        --Matt