Re: Allowing duplicate template specialisations

From:

=?ISO-8859-1?Q?Daniel_Kr=FCgler?= <daniel.kruegler@googlemail.com>

Newsgroups:

comp.lang.c++.moderated

Date:

Mon, 17 Jan 2011 13:56:38 CST

Message-ID:

<ih2381$gkn$1@news.eternal-september.org>

Am 17.01.2011 12:50, schrieb Alexander Lamaison:

[..]

I wish it were just a theoretical issue to do with the spec but I actually
get a compiler error:

remote.cpp(94) : error C2766: explicit specialization;
'comet::comtype<IObjectWithSite>' has already been defined
object_with_site.hpp(41) : see previous definition of
'comtype<IObjectWithSite>'

The specialisations both look like this (i.e. they're identical):

template<> struct ::comet::comtype<::IObjectWithSite>
{
    static const ::IID& uuid() throw() { return ::IID_IObjectWithSite; }
    typedef ::IUnknown base;
};

with the original template like so:

template<typename T> struct comet::comtype
{
    static const IID& uuid() throw()
    { return comtype<typename T::interface_is>::uuid(); }

    typedef typename comtype<typename T::interface_is>::base base;
};

I understand that in the same compilation unit, the compiler has received
two separate definition but they're identical; can it not just choose one?

OK, so if I understand your use-case correctly you want support multiple
definitions of such entities in the *same* translation unit. This looks
odd to me, because there is a longstanding and general principle that
multiple definitions of entities (including class types, primary
templates or partial specializations as well as inline functions) shall
have only a single definition per translation unit. I wonder why you
cannot prevent this compiler error: Are these really two independent
inclusion headers with the same specialization definition which don't
share the same "inclusion guard"? How does this come?

Greetings from Bremen,

Daniel Kr?gler

--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]