Re: maintainer-check fallout: C++ and type checks

[Ralf Wildenhues]

Hi Paul,

* Paul Eggert wrote on Thu, Mar 09, 2006 at 11:59:48PM CET:
> Ralf Wildenhues <address@hidden> writes:
> 
> > (actually, rcs-5.7 is >10 years old; its Changelog suggests this code
> > to be 15.  amazed..)
> 
> I wrote that code many years ago, as part of RCS's 'configure'
> procedure, before Autoconf even existed.

Yes, it amazed me just as much to see familiar names attached to the
changes.

*snip*
> These changes all look good; please install.  And thanks again.

Installed.  Thanks for the review!

Another issue I was blissfully ignorant of here: of course fixed-size
arrays
  AC_CHECK_TYPE([[int [10]]])

don't work, but that's not a regression either.  It would be a
nice (post-2.60) exercise to fix this in one of these ways:
- let the user specify the right place where the type name would need
  to be placed:
     AC_CHECK_TYPE([[int ]AC_TYPE_PLACEMENT[[10]]])
     AC_CHECK_TYPE([[int (*]AC_TYPE_PLACEMENT[) (int, double)]])
- use some heuristic to find the right position (yuck!)
- use a declaration parser (i.e., CPP) to find the right position.
  (writing one in M4 would still leave you to shoot yourself in the
  foot with macros and typedefs found in prerequisite headers).

Just thinking out loud..

Cheers,
Ralf