Re: automake/500: unnecessary -I options in DEFAULT_INCLUDES (make clutter)

[Ralf Wildenhues <Ralf dot Wildenhues at gmx dot de>]

The following reply was made to PR automake/500; it has been noted by GNATS.

From: Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
To: Paul Eggert <eggert@CS.UCLA.EDU>
Cc: automake-gnats@sources.redhat.com
Subject: Re: automake/500: unnecessary -I options in DEFAULT_INCLUDES (make clutter)
Date: Tue, 29 Aug 2006 11:43:37 +0200

 * Paul Eggert wrote on Tue, Aug 29, 2006 at 10:11:01AM CEST:
 >  Ralf Wildenhues <Ralf.Wildenhues@gmx.de> writes:
 >  
 >  > Why not just use `AUTOMAKE_OPTIONS = nostdinc'
 >  
 >  Yes, we could do that in every Makefile.am to override the default
 >  rules, but it's inconvenient to use nostdinc all over the place.  I'd
 >  like Automake's default to be more convenient.
 
 (You could also change that globally as argument to AM_INIT_AUTOMAKE;
 but you know that, and it isn't the point here.)
 
 >  Sorry, I must not have been clear.  I wasn't advocating removal of
 >  '-I$(srcdir)'.  I was advocating the removal of '-I.', in the special
 >  case where no .h files are put into the builddir.  In that case, -I.
 >  is not needed, since all the .h files are in $(srcdir).
 
 I misunderstood you, did not read carefully enough.  Sorry about that.
 
 >  > I think the BUILT_SOURCES rule is too strict.  You can reasonably
 >  > have a header that is generated by config.status
 >  
 >  Yes, but that case is caught by the CONFIG_HEADER part of the patch.
 
 Built header files can be generated by means other than CONFIG_HEADER,
 for example by CONFIG_FILES instantiation.  If Automake is to change
 behavior as you suggest, that would be an incompatibility, so at least
 it would need a NEWS entry.  
 
 Cheers,
 Ralf