This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]

Re: V3 problems

To: jbuck at racerx dot synopsys dot com
Subject: Re: V3 problems
From: Mark Mitchell <mark at codesourcery dot com>
Date: Mon, 05 Feb 2001 18:52:53 -0800
Cc: dje at watson dot ibm dot com, bkoz at nabi dot net, gcc at gcc dot gnu dot org
Organization: CodeSourcery, LLC
References: <20010205111100J.mitchell@codesourcery.com><200102052059.MAA07351@racerx.synopsys.com>

>>>>> "Joe" == Joe Buck <jbuck@racerx.synopsys.com> writes:

    Joe> I think that the proper way to handle this is to use the same
heuristic
    Joe> that is used elsewhere: the definition for the vtbl and
type_info stuff
    Joe> goes in the same .o file that defines the first noninline, nonpure virtual
    Joe> function.  If an .o file gets an definition of that function due to
    Joe> explicit template expansion, it should also get a definition
of the vtbl
    Joe> and type_info bits.

I agree -- and I *think* that is what we already do.  Anyhow, I'm
pretty sure that explicit instantiation of a whole class certainly
dumps out the vtable.

Bottom line: we should fix the compiler, if there are problems there.
(And that won't be hard.)  So far, though, my testing indicates there
are no problems...

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

References:
- Re: V3 problems
  - From: Mark Mitchell
- Re: V3 problems
  - From: Joe Buck

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]