From: Richard Maine <maine@altair.dfrc.nasa.gov>
Newsgroups: comp.lang.fortran
Subject: Re: fun-to-read textbook recommendation
Date: 04 Jan 2000 12:23:29 -0800

horationelson <defaultuserNOdeSPAM@bromhead.free-online.co.uk.invalid> writes:

> If you want FORTRAN to look like C, why not program in C?

I don't want that.  I personally find C abominable.  If I liked it,
then I'd be programming in it.  Tried that when I had concluded that
f77 was inadequate for me, but f90 compilers were not yet out; decided
I'd wait for f90.  I'll not (re)start the usual C vs Fortran flamewar
by even mentioning the particular things of C that I find abominable.
That doesn't seem relevant.  What is relevant (to me) is that I
personally do like (on the whole) f90, and I don't figure that anyone
else is in a position to make the judgement for me that, because they
think f90 and C seem simillar, that this means I should like C just as
well as f90.  I am the world's leading expert in terms of knowing what
I like.  My choices may well be poor, but I know what they are.  I find
it presumptuous for others to tell me what I ought to like.

Isn't this the same poster that just said

 "If you don't like it, don't use it, but don't stop the rest of us who do like it
  from using it!"

Perhaps the antecedant that we have in mind for the "it" is different, but
I can't think of more appropriate words.  I certainly agree with them.

> But, it [f90/f95]
> has various things in it which are "deprecated" and hence due to be
> dropped in a future revision of the language.

I think you haven't actually looked at the standards in question, but
are relying on rumor and hearsay.  The list of obsolete features in
f90/f95 is *VERY* small.  It certainly does not include GOTO or any
other such widely used features.  There has never even been a
suggestion to add GOTO to the list.  I think someone once did try to
add COMMON/EQUIVALENCE, but if so, that suggestion got nowhere at all.
The list is really quite small.  (Well, it does include the assigned goto,
but that's nothing at all like the "plain" gotos).

I might add that the current f2k plan adds *NOTHING* to the obsolete
list, and does not delete anything that was formerly on the obsolete
list.  I've also never actually seen an f95 compiler that failed to
support the very few deleted features of the language.  (There were
no deleted features in f90, so it doesn't come up there).

Do not confuse the personal preferences of authors (or of myself)
with the dictates of the standard.  You will no doubt read several
books in which authors label many things as deprecated in their
judgement.  This has *NOTHING* to do with the standard; it is merely
the personal opinion of the author in question.

There is nothing at all in f90, very little in f95, and the same
amount in the current draft of f2k, that prevents legal f77 code from
running.  Of the few features that f95 allows compilers to delete, no
actual compilers have.

There are f90 compilers that don't implement some fairly widespread
f77 extensions.  But thats a different matter.  There are also f77
compilers that don't implement those same extensions.

I think you worry too much about the wrong aspects on this matter.  To
me, the biggest reason to avoid the deprecated features (even those
that are not formally obsolete in the language) has more to do with
things like them being error prone, and not nearly so much to do with
the worry that some future compiler might not support them.  Both
things (errors or compiler non-support) could possibly happen.  Many
things could possibly happen.  I know which of those 2 I'd worry about
most though.

--
Richard Maine
maine@altair.dfrc.nasa.gov

From: Richard Maine <maine@altair.dfrc.nasa.gov>
Newsgroups: comp.lang.fortran
Subject: Re: fun-to-read textbook recommendation
Date: 04 Jan 2000 14:08:49 -0800

jbs@watson.ibm.com writes:
> In article <ue3dsdio1q.fsf@linux.dfrc.nasa.gov>,
>  on 04 Jan 2000 12:23:29 -0800,
>  Richard Maine <maine@altair.dfrc.nasa.gov> writes (in part):
> >Isn't this the same poster that just said
> >
> > "If you don't like it, don't use it, but don't stop the rest of us who do like it
> >  from using it!"
> >
> >Perhaps the antecedant that we have in mind for the "it" is different, but
> >I can't think of more appropriate words.  I certainly agree with them.
>
>          Do you agree then that it was wrong to withdraw the fortran 77
> standard in an attempt to force people to move to fortran 90?

No, I wouldn't.  But I see that as a different kind of issue, and I consider
the "in an attempt" part to be unfounded speculation.  I don't recall any
anouncement that the f77 standard was withdrawn in attempt to force people
to move to f90.  I just recall an announcement that the f77 standard was
withdrawn.  I was there for at least some of the relevant votes.  And
whereas that certainly doesn't mean that I necessarily understood the
motivation of every person in the room (I'm quite sure I didn't), I don't
think those not there have any better claim to understand it.

So instead of asking me whether I'm still beating my wife, lets separate
it into the two questions that it really is

1. Do I agree with the decision to withdraw the f77 standard?  Yes.

2. Do I think people should be forced to move from f77 to f90?  No.
   (I do try to convince people of the benefits, but I'm not "into"
   forcing people that disagree with me.  And there are even occasions
   where I agree that the cost/benefit of such a move isn't there for
   existing codes).

Withdrawing the f77 does not force anyone to stop coding in f77.  That
same f77 code works with f90 compilers.  Nor does it stop anyone from
continuing to offer f77 compilers, or even from developing new such
compilers.  (I doubt the market will justify development of much in
the way of new f77 compilers, but the status of the f77 standard is
scarcely the driving factor there).  Heck, there is nothing to prevent
people from developing new f66 compilers - I just don't expect to see
it happen.  All in all, I see very little concrete direct effect of
withdrawing the f77 standard.

As far as I can see, the main point of withdrawing the f77 standard
was just the acknowledgement that the standard was no longer being
maintained.  You can't expect to get interpretation questions formally
answered or bugs in the standard fixed.  Not that any of this had
actually been done for f77 for a long time.  There was one
interpretation document (Fortran Information Bulletin 2 - no I don't
know what number 1 was), but that was long ago, I think shortly after
f77 came out.  Keeping the standard would have at least implied that
it was still actively maintained - not that I suspect that would have
been any more "real" than it had been for the last 2 decades.

P.S. There may be organizations that have policies that they will
require conformance to the current standard, but neither I (nor
the committee) force those policies on organizations.  It is the
choice of the organization to have such a policy.

--
Richard Maine
maine@altair.dfrc.nasa.gov