This is the mail archive of the
mailing list for the GCC project.
Re: C++ PATCH: PR 12815
On Mon, 2004-01-05 at 17:31, Jason Merrill wrote:
> On Mon, 5 Jan 2004 16:53:35 -0800, Mark Mitchell <firstname.lastname@example.org> wrote:
> > This patch fixes a code-gen problem in the C++ front end. Jason (some
> > while back) marked as all vtable references as TREE_CONSTANT, which
> > doesn't make sense. If the object isn't TREE_CONSTANT, and its
> > dynamic type is not known, then it's vtable entries are not
> > TREE_CONSTANT either, although they are TREE_READONLY. Jason, am I
> > missing something here?
> What I was trying to express is that the vtable for a particular object can
> be assumed to remain the same once it's been initialized; a vtable entry is
> a pure function of the object and the slot. But TREE_CONSTANT says it's a
> compile-time constant, which of course it isn't.
> I'd probably just remove the offending line rather than weaken it.
> Likewise the line in build_base_path that went in at the same time.
Would you be so kind as to take care of that, as the author of the
Mark Mitchell <email@example.com>