Proposal for mangling template partial specializations.

[Mark Mitchell]

>>>>> "Coleen" == Coleen Phillimore <coleen at zko.dec.com> writes:

    Coleen> We have a #pragma do_not_instantiate which we use for that
    Coleen> purpose.  Supporting pre-instantiation libraries is very
    Coleen> important to us as well.  Unfortunately, other than a
    Coleen> pragma, there doesn't seem to be a cleaner way to do this.

My point is that a sizable existing body of code does this using the
specializatoin/instantiation trick.  I'll not argue that it's cleaner,
but it is important to take existing practice into account.

    >>  o I don't think the standard doesn't require a diagnostic.

    Coleen> No, it doesn't.  But does the standard prohibit linking a
    Coleen> specialization when an instantiation is required?  Is this
    Coleen> something that Perennial can write a test for?

I don't think the standard says anything.  It just says "undefined
behavior, no diangostic required", I think.  That means
implementations can do what they want.  Perennial can certainly write
a test -- but it's a quality-of-implementation issue, and not even one
where it's clear to me which implementation has a higher quality!

    >> From my perspective, it is broke.  It prevents conforming
    >> implementations from being ABI complaint.

I don't understand that sentence.  Can you explain why no conforming
implementation can mangle a specialization and an instantiation in the
same way?

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