Re: C++ FE question: When is CLASSTYPE_VBASECLASSES valid?

1 msg	Re: Patch: delete treelang
2 msg	Seg fault in call_gmon_start

C++ FE question: When is CLASSTYPE_VBASECLASSES...
\ Doug Kwan (關振德) (7 Mar 2008)
. \ Mark Mitchell (20 Mar 2008)
. . \ Doug Kwan (關振德) (20 Mar 2008)

1 msg	gcc-4.3-20080306 is now available
1 msg	Some regression numbers
2 msg	GCC 4.3.0 Status Report (2008-03-06, 2nd editio...
2 msg	Injecting data declarations?
1 msg	[tuples] Merged trunk->branch @132948
4 msg	GCC 4.3 - Error: Link tests are not allowed aft...
1 msg	Enum in namespace
6 msg	GCC 4.3.0 Status Report (2008-03-06)
1 msg	gcc-4.2-20080305 is now available
2 msg	GCC build problem
6 msg	Re: [PATCH] Make alias_sets_conflict_p less con...
1 msg	Soliciting help upgrading our bugzilla install
2 msg	Lots of new test failures on trunk due to unrec...
100 msg	Linux doesn't follow x86/x86-64 ABI wrt directi...
3 msg	gcc 3.3.2 Wind River VxWorks PowerPC
4 msg	[PATCH] BIT_FIELD_REF_UNSIGNED considered harmful
4 msg	[tuples] Branch frozen (again)

Subject:	Re: C++ FE question: When is CLASSTYPE_VBASECLASSES valid?
Group:	Gcc
From:	Doug Kwan (關振德)
Date:	20 Mar 2008

Thanks Mark. I am using classes after they are defined. So it should
be okay for me.

-Doug

2008/3/19 Mark Mitchell <mark>:
>
> Doug Kwan (關振德) wrote:
>
> > I have a question about the validity of CLASSTYPE_VBASECLASSES.
> > Due to templates, it is not possible to know if a class has virtual
> > bases or not until the class is fully instantiated. Is checking
> > processing_template_decl sufficient to guarantee that
> > CLASSTYPE_VBASECLASSES is valid?
>
> No; if the class is presently being defined, that will not be set.
> However, it should be safe to check that for a complete class when
> !processing_template_decl.
>
> --
> Mark Mitchell
> CodeSourcery
> mark
> (650) 331-3385 x713
>
>