HELP w/bug on.... mangling? new-api? something else?

[Martin v. Loewis]

> > > 	_ZTCSt10istrstream0_Si
> > > 	_ZTCSt10ostrstream0_So
> > > 	_ZTCSt9strstream0_Sd
> > > 	_ZTCSt9strstream0_Si
> > > 	_ZTCSt9strstream8_So
[...]
> Yes.  So the question is ... why aren't those vtables being defined
> anywhere?

They should have been defined when the first virtual non-inline
function was defined, together with the other vtables as well. In
fact, the only place where this symbol is mangled (in
class.c:build_ctor_vtbl_group), it is also defined right-away.

I can't answer your question, so you may either want to investigate it
yourself, or produce a bug report (preprocessor output etc).

Regards,
Martin