Re: Contents of <confdefs.h> for C++

[Walter E. Brown]

I had raised (rather courteously, I thought) some technical issues in
my several recent notes to this mailing list.  Alas, instead of
technical responses dealing with these issues in a professional
manner, I have been largely lectured (sometimes nicely, sometimes
not -- witness such words as "condemned" "silly," "abusive") about
programming style by people who haven't seen my code, nor who have any
insight as to its objectives.

You, for example, first deny that certain practices are "arguable" (my
term) -- and then argue about them!  Isn't that "silly"?

I've now been accused of "dramatically overstating" -- but it could be
claimed that it is use of language such as "universally condemned"
that smacks of "dramatically overstating"!

I note that no one has asked for the basis of my assertion.  When I
claimed that "a substantial portion of the C++ community" takes a
viewpoint akin to mine, it seems to have been assumed that I was
grandstanding.  However, as a voting member of the C++ standards
committee, I have first-hand knowledge of the (sometimes extreme!)
lengths to which the committee goes to avoid forcing its community to
pay for C++ language and library features in programs that don't use
such features.  It's quite an important philosophical underpinning of
C++, and it's been there for quite a long time.  The autoconf
community doesn't have to accept that, of course, but I believe it
would be a very great shame if it did not, because it would
unnecessarily reduce the tool's applicability.

With all due respect, I'd like us to stop generating heat and start
generating light instead:  Programmers don't get to decide how other
programmers write code.  You can recommend, suggest, discuss,
persuade, teach, debate, etc., etc., but you don't get to decide style
issues for me, for my company, or for my community.  Spending any more
time on that is non-productive, so I won't.

What the maintainers on this list do get to decide, of course, is to
what extent autoconf will support its user community.  At the moment,
it is my opinion that autoconf is not well-supporting the C++
community.  I have previously offered rationale for my views, and have
even suggested alternatives for consideration.  I'd like someone to
make an authoritative decision on the need for implicit, undocumented
support for a "feature" that runs contrary to established C++
philosophy in general, and that has the potential for breaking (and
has broken) user code.

I regret that I will be away from my office for several weeks while my
wife recuperates from surgery (scheduled for early tomorrow morning --
your thoughts and prayers on her behalf would be welcomed), and will
therefore be unable to take further part in this mailing list until my
return.  I look forward to collegial discussion at that time.

Best,

    - WEB


----- Original Message -----
From: "Braden McDaniel" <address@hidden>
To: "Walter E. Brown" <address@hidden>
Cc: "Raja R Harinath" <address@hidden>; <address@hidden>
Sent: Monday, July 08, 2002 12:37 AM
Subject: Re: Contents of <confdefs.h> for C++


> On Sun, 2002-07-07 at 17:15, Walter E. Brown wrote:
> > Thank you for your reply.
> >
> > It appears we have different viewpoints of what needs a rationale.
I,
> > and a substantial portion of the C++ programming community, tend
to
> > take what may be termed a minimalist view:  why should I have to
pay
> > for a feature I don't want and have no need for?  Others, at least
in
> > the autoconf world, it appears, regard the use of exit() as so
> > fundamental that they are always willing to pay for it, and to
force
> > others to do so, as well.
>
> [snip]
>
> As a C++ user and an autoconf user, I believe you are dramatically
> overstating your case.
>
> > Suppose my program defines entities whose names clash with names
> > declared in <stdlib.h>.  While it is certainly arguable whether
this
> > constitutes good practice, it seems equally certain that such
programs
> > comply with the relevant language standard(s).
>
> I'm not sure what led you to believe that it's "arguable" as to
whether
> this is good practice. IME, it is universally condemned. But there
are
> plenty of silly and abusive things you can do that fall within "the
> relevant language standard(s)"; I'm not convinced that's an adequate
> acid test of what autoconf should support. It sounds to me as if you
(or
> the libraries you're using) are doing something foolish, and you
deserve
> to get burned. Your identifiers should be in a namespace.
>
> --
> Braden McDaniel                           e-mail:
<address@hidden>
> <http://endoframe.com>                    Jabber:
<address@hidden>
>
>