Virtual Calls: Make the ABI match the implementations

[Jason Merrill]

>>>>> Mark Mitchell <mark at codesourcery.com> writes:

>>>>> "Jason" == Jason Merrill <jason at redhat.com> writes:
 Jason> OK, what if the function is compiled with a compiler that
 Jason> emits the thunks on an ad-hoc basis, and a vtable that
 Jason> needs a thunk is compiled with a compiler that emits the
 Jason> thunks with the function?  Seems that it would lose.

 > I think that will work under our proposal because our proposal
 > requires all compilers to emit the thunks on an ad-hoc basis.  In
 > other words, all compilers must emit the thunks wherever they are
 > referenced, as COMDAT.

Ah, I see.  I'm strongly opposed to requiring all compilers to do work to
support their semi-compliant peers.

Jason