This is the mail archive of the
mailing list for the GCC project.
Re: [tree-profiling-branch PATCH IPCP extensions + Function cloning
> On Jan 17, 2005 11:02 AM, Razya Ladelsky <RAZYA@il.ibm.com> wrote:
> > > What makes templates so special so you want to avoid the versioning
> > > here?
> > Many libstdc++ tests failed when we versioned templates.
> > I'll continue to investigate this, in the meantime I want to avoid it.
> Strange. What do we know about templates at this point?
> I would expect that templates do not exist as far as the call
> graph is concerned... Templates are front-end business, and
> cloning is part of the middle-end. If you end up attempting
> to clone a template, something somewhere is badly broken.
Templates are often tricky busyness just because they comdat nature and
other nueances. But we definitly need to investigate it.
(one "obvious" posibility is forgotten comdat flag when clonning I
wonder if we want to invent name mangling for specialization so
specialized templates can be unified)
> What does the backtrace of the failure look like?
> Gr. exactly