Re: RFI: Trailing blanks

[Gary V. Vaughan]

On Fri, Oct 19, 2001 at 05:00:52PM +0200, Akim Demaille wrote:
> >>>>> "Gary" == Gary V Vaughan <address@hidden> writes:
> 
> Gary> But then, I wonder why you bother to `smash' anyway?  
> 
> Because I frequently read configure, and I like to have it readable.
> My Autoconf programming style is also based on the hypothetic idea
> that some day M4 could be able to expand
> 
> define(foo, bar
> baz)
>     foo
> 
> info
> 
>     bar
>     baz
> 
> instead of
> 
>     bar
> baz

If it was easy, then it would certainly be a nice enhancement, however
(there's always a `however':-)!), I think what you are really asking for
is an bourne shell `indent' ccmmand... I would rather have m4 be lean
and mean, rather than waste cycles on this stuff -- especially
considering the vanishingly small percentage of autoconf users who want
to read their configure files.

> [[snip]]
> But I understand your point.

`kay.. I guess I am labouring that point in the previous paragraph then ;-)

> Gary> Who cares if the machine readable parts are ugly?  Well,
> Gary> obviously you and I do, and perhaps a few dozen others who read
> Gary> through it while we develop complicated code generation macros,
> Gary> but I don't think we should make the user interface for the
> Gary> thousands of casual users less friendly for the sake of
> Gary> ourselves.
> 
> I'm not sure I understand where you think it is less friendly.  We are
> not requiring that people dnl their AC_REQUIRE, we advice that they do
> so :)

It is an unfortunate consequence of the way that m4 works that you need
to be at least vaguely aware of the target language (bourne shell) as
well as the macro API (autoconf.m4f say) you are using... but advocating
the additional dnls exposes an internal detail of the macro expansion
mechanics to the user.  It is easier not to explain why dnl is "needed"
here, and let the user be blissfully unaware of all the harmless blank
lines in the generated configure script.

> Gary> I really think that leaving the blanks in is the right solution.
> 
> The last argument in favor was also normalizing configure as much as
> possible, to avoid useless diff's.  But now that we use LINENO, it
> should be less of a problem.  So if everybody agrees --smash is
> useless, let's remove it.

You have my vote!  (as if there was ever any doubt) :-)

Cheers,
        Gary.
-- 
  ())_. Gary V. Vaughan     gary@(oranda.demon.co.uk|gnu.org)
  ( '/  Research Scientist  http://www.oranda.demon.co.uk       ,_())____
  / )=  GNU Hacker          http://www.gnu.org/software/libtool  \'      `&
`(_~)_  Tech' Author        http://sources.redhat.com/autobook   =`---d__/