[comp.lang.c] \"major adjustments\" in ANS C

lcc.marks@SEAS.UCLA.EDU (Mark Seecof) (01/08/88)

Doug Gwyn, replying to another, writes:

> Actually, the time for major adjustments was over a year ago;
> several years ago if you include participating on the committee
> as a way to have even more influence on the process.
> Now is the time to repair serious deficiencies, only.

If the Committee would refrain from making novel adjustments of its
own AFTER the first review period (I am particularly concerned by the
proposed change to force "fortran-style" parentheses interpretation,
but "noalias" may be a better example (though I probably won't object
to *it* ;-)) then the rest of us wouldn't be in the position of
submitting comments on proposed changes during the second-review-
period that should have been handled in the first period.

Changes in the second draft should address deficiencies in the
first.  Proposed changes from the previous standard (in this case
K&R) should be in the first draft, so that people can comment on
them early.

The informal information we have suggests that the parentheses
change results from the Committee feeling pressured by a threat
from ISO not to adopt the ANS standard as an ISO standard unless
the change was made.  I wish the Committee would remember that
their task is to create an American National Standard, not an
international standard.  If ISO, due to inferior comprehension
of the issues involved, can't stand the K&R parentheses rule
they can just make a little change in the ANS standard and
adopt that.  To phrase it another way... the ANSI Committee should
do things the right way even if others want to do things the wrong
way.  This kind of blackmail (do it my way or I won't adopt your
standard) should be given exactly the respect it deserves--none.

The compiler manufacturers will adhere to the better (hopefully
ANS) standard anyway, so there's no danger involved.  Just for
fun, what d'you suppose the Committee would do if CCITT applied
pressure the opposite way?  (I know CCITT doesn't do programming
languages, but they often interact with other ANSI committees,
and routinely adopt standards incompatible with ISO's; pretty
funny--two incompatible official international standards!)


Mark Seecof

gwyn@brl-smoke.ARPA (Doug Gwyn ) (01/08/88)

In article <11179@brl-adm.ARPA> lcc.marks@SEAS.UCLA.EDU (Mark Seecof) writes:
>Changes in the second draft should address deficiencies in the
>first.  Proposed changes from the previous standard (in this case
>K&R) should be in the first draft, so that people can comment on
>them early.

Point well taken, except please note that all the changes I can
recall have been prompted by the first set of public comments,
or at least were asked for there.  Even "noalias" resulted from
trying to deal with the aliasing issue which was raised by some
public comment (I forget which), even if the committee actually
would have tackled this anyway on their own initiative.

Certainly there were many requests to "honor parentheses".  We
kept turning them down until finally someone (I recall who, but
don't want to name him) pointed out that it wasn't as serious a
change as we had been thinking.  The ISO demand for it was the
last push it took to get us to accede to this common request.

As to ignoring ISO requests/demands, few people think it would
be a good idea for the ANSI and ISO standards to differ.  Like
it or not, most major computer vendors these days are dealing
with an international marketplace, so they will have to provide
ISO-compliant compilers.  I sympathize with what you say, but
that's not the way the committee chose to handle this.

barmar@think.COM (Barry Margolin) (01/09/88)

In article <11179@brl-adm.ARPA> lcc.marks@SEAS.UCLA.EDU (Mark Seecof) writes:
>  I wish the Committee would remember that
>their task is to create an American National Standard, not an
>international standard.

There is an enormous amount of pressure for American standards to
conform to ISO standards.  This didn't happen with Pascal, and it
really confuses things, much the same as the problems caused by the
two current major Unix versions (Berkeley and AT&T).  X3J11 will have
to do quite a bit of convincing to get X3 to approve a standard that
is known to differ from the ISO version.

>  If ISO, due to inferior comprehension
>of the issues involved, can't stand the K&R parentheses rule
>they can just make a little change in the ANS standard and
>adopt that.  To phrase it another way... the ANSI Committee should
>do things the right way even if others want to do things the wrong
>way.  This kind of blackmail (do it my way or I won't adopt your
>standard) should be given exactly the respect it deserves--none.

You talk as if ISO is some type of dictatorial organization.  ISO
technical committees are made up of representatives from all the
national standards committees, with each country having one vote.

A similar comment goes to all the people blasting X3J11 for screwing
up the language, making it unusable and unimplementable.  Who do you
think makes up X3J11?  It's just a collection of compiler developers
and users.  "Noalias" presumably got added because a number of users
felt they needed a way to advise the compiler about an optimization
and the developers felt they could implement this.  Standards
committee members do not come from an ivory tower, they are mostly
representatives from the trenches, people just like us, who feel they
understand the issues.  Of course, they are fallible, which is why
ANSI and ISO procedures include the public review requirements.

>The compiler manufacturers will adhere to the better (hopefully
>ANS) standard anyway, so there's no danger involved.

More likely, they will adhere to the one that is more popular among
customers.  Of course, there is a chicken/egg problem, since customers
are influenced by vendor marketing.  But if a vendor has a sizeable
non-US market, this will probably be the ISO standard.
Barry Margolin
Thinking Machines Corp.

barmar@think.com
uunet!think!barmar

franka@mmintl.UUCP (Frank Adams) (01/13/88)

In article <11179@brl-adm.ARPA> lcc.marks@SEAS.UCLA.EDU (Mark Seecof) writes:
>pretty funny--two incompatible official international standards!)

"The nice thing about standards is that there are so many of them to choose
from!"

(Sorry, I don't know who said it.)
-- 

Frank Adams                           ihnp4!philabs!pwa-b!mmintl!franka
Ashton-Tate          52 Oakland Ave North         E. Hartford, CT 06108