PATCH: invoke.texi (was: Patch for bug with new-initializers)

[Jason Merrill]

>>>>> Mark Mitchell <mmitchell@usa.net> writes:

> Well, I will jump in here since my words are re-appearing.  Jason has
> worked very hard to fix implicit typename, and it does have a purpose:
> backwards-compatibility with old code.  I still have reservations
> about it, but Jason has worked very hard to improve it, and I have
> personally witnessed no recent implicit typename crashes.  Of course,
> I've also been changing code to use typename in all the places the
> standard requires.  In any case, I believe that Jason is "in favor of"
> the implicit typename extension.

Yep.  Of course, that particular extension is a special case, as it
provides a measure of compatibility with code that was correct up until
very recently.

Jason