Re: warning C4250 (method inherited via dominance) vs error c2250 (ambiguous inheritance)

From:

"Tom Widmer [VC++ MVP]" <tom_usenet@hotmail.com>

Newsgroups:

microsoft.public.vc.language

Date:

Tue, 18 Sep 2007 15:07:24 +0100

Message-ID:

<OCjM#0f#HHA.1164@TK2MSFTNGP02.phx.gbl>

j.j.a.perks@googlemail.com wrote:

I have a classic virtual inheritance diamond in my class structure,
and I was wondering to what extent C4250 ('class1' : inherits
'class2::member' via dominance) was safe to ignore.

It looks like just an informative comment in case you don't know how
multiple and virtual inheritence works.

More specifically,

are there any (not too obscure) cases of ambiguity that wouldn't get
flagged as an error, typically C2250, where behaviour is undefined or
at least counter-intuitive?

I don't think so. A qualified call: myderived.OverrideF::g(); will cause
a link error I think, since it is a direct call to Root::g().
Unqualified calls will call the correct virtual function.

In this example:

struct Root { virtual void f() = 0; virtual void g() = 0; };

struct OverrideF : virtual Root { virtual void f(); };

struct OverrideG : virtual Root { virtual void g(); };

struct Derived : OverrideF, OverrideG { };

the definition of Derived causes warning C4250 (and does so whether or
not the Root methods are declared as abstract), and yet I can't see
how this would be anything other than as expected.

So is there any reason not to punt this warning down to Level 4,
knowing that any real ambiguities will be flagged as an error?

It seems like a bit of a nannying warning, and can be ignored I think.

Tom