This is the mail archive of the gcc-patches@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]
Other format: [Raw text]

Re: [Dodji Seketeli] Patch PR c++/45200

On 08/12/2010 09:16 AM, Dodji Seketeli wrote:
Jason Merrill<jason@redhat.com> writes:

The change is OK, but the testcase looks over-reduced; apply1 isn't
actually used at all.


It's true that apply1 is not used but it is useful to trigger the bug
because it has a different number of template parameters than struct
apply. E.g, removing apply1 or removing its last template parameter
makes the crash to vanish.

OK, then.

I'd like to see about moving back to using TYPE_CANONICAL to compare these types; it seems to me that might work if we just use different TYPE_CANONICAL for template parameters from parameter lists of different lengths. So TYPE_CANONICAL would be the same for template type parameter 1 of 3 in all templates, but different from template type parameter 1 of 2.

Does that make sense to you?

Jason

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