On Thu, Jun 9, 2016 at 9:43 PM, Paolo Carlini <paolo.carl...@oracle.com> wrote: > ... and this version passes testing. For real.
This seems reasonable, but I still wonder why b1 has error_mark_node type, since it has no initializer at all. Jason
On Thu, Jun 9, 2016 at 9:43 PM, Paolo Carlini <paolo.carl...@oracle.com> wrote: > ... and this version passes testing. For real.
This seems reasonable, but I still wonder why b1 has error_mark_node type, since it has no initializer at all. Jason