Fortran - On legitimacy (was: Q: F2k3 Vote Result)

Author

On legitimacy (was: Q: F2k3 Vote Result)

Michael Metcalf

2005-02-19, 3:58 pm

While I'm going on about this, here is the Editorial I wrote on the topic
for Fortran Forum in April, 2003:

"Editorial: On legitimacy

"I have always felt that the mechanism for producing international
programming language standards is somehow absurd. Whereas differences
between counties on standards for industrial products are to be expected,
how can it be assumed that Italian Fortran programmers might feel,
collectively, differently about the syntax of the PARAMETER statement to
Spanish ones? This feeling of unease grew during the final stages of the
preparation of Fortran 90 when two countries threatened, essentially, to
veto the whole standard for the sake of each of two single features! A huge
effort was required, by others, to head off those crises, and the two
features concerned, non-default character kinds and varying-length strings,
are, in the event, little used.

"Of course, the international mechanism is valid if it increases the amount
of consultation but, to the extent that consultation has slowly withered,
this justification has grown weaker. In the 1980s, most members of J3 and
WG5 seemed to be honest representatives of their respective communities, in
the case of users, or companies, in the case of vendors. Many of them
undertook a significant effort to ascertain the views of those they
purported to represent, by polling opinion or holding user meetings. In
addition, a considerable collective effort was expended in attending user
meetings at which the evolving Fortran 90 standard was explained to wider
groups (DECUS, SHARE, BCS, etc.) or to those gathered at conferences (for
instance, SIAM or CHEP). It could truly be said that the committees
faithfully undertook to explain their work to the whole community and, in
turn, took account of the community's (often conflicting) views. Within some
countries, the local WG5 organization also polled opinion and organized
special workshops. All in all, it can be claimed that, individually and
collectively, every reasonable effort was undertaken to publicize the
activities of the committees and to s

feedback. Thus, most members had a
certain legitimacy as representatives of a wider grouping.

"What is the situation today? The world has shrunk with the pervasive use of
the World Wide Web, and it is easy for anyone to find out what's going on in
the arena of Fortran standardization and to make suggestions and criticisms
via the Internet. So the process is very open. However, few of the large
user communities are still formally represented on the committees. Moreover,
there has been only a limited attempt to draw users' attention to the plans
for Fortran 2000 in structured fora such as conferences and workshops, as
used to be the case. (This is not a criticism of the few remaining
hard-working committee members, just a statement of fact. The community
itself is at fault for not providing sufficient support to standardization.)
Thus, one can seriously pose the question as to the legitimacy (as opposed
to the openness) of the process. Only a tiny number of the members of J3 are
active programmers within a larger scientific community. Countries can be
represented at WG5 by individuals who are devoted, but where consultation
within their respective countries apparently no longer takes place. The
process becomes subject to the whim of individuals. Is this what lies behind
the negative vote of Germany on the Committee Draft for Fortran 2000? As a
resident of that country, I have certainly been unaware of any effort to
canvass views on the draft standard, nor have I ever seen invitations to
discussion meetings, public or otherwise. Yet now, in the name of all German
Fortran programmers, a negative vote has been registered by DIN. With what
moral justification?

"I have no magic solution for improving representation, but perhaps a
systematic search for new WG5 members could be undertaken by noting, over a
reasonable period, all intelligent interventions to comp.lang.fortran and
comp-fortran-90, and approaching the contributors directly (rather than
issuing the blanket invitations that have so conspicuously failed in the
past). It would not be unreasonable to suppose that participants to these
electronic groups are representative of active programmers. Could their
views be formally polled? Could it be recognized that the Fortran community
has no national boundaries and that national votes are senseless?

"A future generation of Fortran programmers will feel thankful that a
faithful band has soldiered on until now but, unless new blood is found, new
mechanisms for consultation instituted, and legitimacy reestablished, the
auguries for the next cycle are dire. Will Fortran 2010 be a one-man show?"

2005-02-21, 3:59 am

Michael Metcalf wrote:
>
> Thus, one can seriously pose the question as to the legitimacy (as opposed
> to the openness) of the process. Only a tiny number of the members of J3 are
> active programmers within a larger scientific community. Countries can be
> represented at WG5 by individuals who are devoted, but where consultation
> within their respective countries apparently no longer takes place. The
> process becomes subject to the whim of individuals.

This is worse than I imagined, and it needs to change!
Imagine a member of Ferrari's design team where his only "experience"
was a one time attendance at the F1 race!? It would be unthinkable, just
as it should be unthinkable that anyone w/o advanced degree in hard
sciences and a minimum of at least 15 years of scientific programming
should do as much as serve coffee at the committee's meetings.

How do we change the system?
I'd suggest that an arrangement be made with netlib admin to solicit
feedback which would serve to determine what *real* users need or want,
and a *qualified* committee in conjunction with vendors would work only
on evaluating and adopting the top requests.

Gary L. Scott

2005-02-21, 3:59 am

bv wrote:
> Michael Metcalf wrote:
>
>
>
> This is worse than I imagined, and it needs to change!
> Imagine a member of Ferrari's design team where his only "experience"
> was a one time attendance at the F1 race!? It would be unthinkable, just
> as it should be unthinkable that anyone w/o advanced degree in hard
> sciences and a minimum of at least 15 years of scientific programming
> should do as much as serve coffee at the committee's meetings.
>
> How do we change the system?
> I'd suggest that an arrangement be made with netlib admin to solicit
> feedback which would serve to determine what *real* users need or want,
> and a *qualified* committee in conjunction with vendors would work only
> on evaluating and adopting the top requests.
>
I don't really think the quality of the work is bad. In fact, I think
it fairly high in quality. My only real issue would be with
prioritization of features to be worked. The fact that my own desired
features are not being considered (I've never submitted a proposal)
would be substantially my own fault I would assume. I think the current
committee is technically capable of quite a good design for any proposed
feature (notwithstanding workload issues).

--

Gary Scott
mailto:garyscott@ev1.net

Fortran Library: http://www.fortranlib.com

Support the Original G95 Project: http://www.g95.org
-OR-
Support the GNU GFortran Project: http://gcc.gnu.org/fortran/index.html

Why are there two? God only knows.

Democracy is two wolves and a sheep, voting on what to eat for dinner...
Liberty is a well armed sheep contesting the vote. - Thomas Jefferson

Gerald F. Thomas

2005-02-21, 3:59 am

"bv" <bvoh@Xsdynamix.com> wrote in message news:421932EA.4F18777@Xsd
ynamix.com...
>
> This is worse than I imagined, and it needs to change!

Sure, that's why we have Matlab and Mathematica to the rescue, courtesy of
Moler and Wolfram, respectively. You bet, these giants of NA harvest netlib
to the advantage of 'naive' Fortran amateurs like you and some. BION,
Fortran has just about outlived its usefulness, the fantasies (so don't go
soiling yourself Aardpig@hotmail.com and sidekicked Georgy porgy
hesterloli@hotmail.com
who has undoubtedly already been propositioned by the former deviant to
have his stool pushed in, but hey, you like that, eh?, B(T)oy George) of
diehard wonks and wankers excepted. For the next 15 - 20 years it's (sic
Fortran, assuming it doesn't morph to Fondue, Frites, Fcuk, or whatever) a
taxpayers (no apostrophe required Aardpig@hotmail.com and toadying
hesterloli@hotmail.comf BTW) funded junket, or meal ticket in the case of
messers metcalf et al, whose parasites can fraudently mislead their HR
personnel into believing that because they're participating in a useless
Voltarian charade of global standardization committee 'work', they're not
only entitled to free trips to nowhere in particular (eg, Los Vegas,
whoa!), but damit, they also deserve a salary increase to top up their
superannuation cache. Happily, the pot (or potty, figure it dolts) is
subject to honest-to-goodness audit. Well now, isn't that grand, even
special! -:))))))))))))))))))).

--
You're Welcome,
Gerry T.
______
"...as I further could have told you as brisk as your D.B.C.
behaviouristically paillet, with a coat of homoid icing which is in reality
only a done by chance ridiculisation of the whoo-whoo and where's hairs
theorics of Winestain." -- Joyce, 'The Wake'.

Dan Nagle

2005-02-21, 4:01 pm

Hello,

On Mon, 21 Feb 2005 01:05:00 GMT, bv <bvoh@Xsdynamix.com> wrote:

>Michael Metcalf wrote:
opposed[color=darkred]
J3 are[color=darkred]
be[color=darkred]
consultation[color=darkred]
The[color=darkred]
>=20
>This is worse than I imagined, and it needs to change!
>Imagine a member of Ferrari's design team where his only "experience"
>was a one time attendance at the F1 race!? It would be unthinkable, just
>as it should be unthinkable that anyone w/o advanced degree in hard
>sciences and a minimum of at least 15 years of scientific programming
>should do as much as serve coffee at the committee's meetings.
>
>How do we change the system?

You can change the language by volunteering and participating
in the process, however flawed the process is.

>I'd suggest that an arrangement be made with netlib admin to solicit
>feedback which would serve to determine what *real* users need or want,
>and a *qualified* committee in conjunction with vendors would work only
>on evaluating and adopting the top requests.

And you'd appoint yourself in charge of determining what "real" means,
and what the "qualifications" are and who meets them?

--=20
Cheers!

Dan Nagle
Purple Sage Computing Solutions, Inc.

Dan Nagle

2005-02-21, 4:01 pm

On Sun, 20 Feb 2005 19:14:59 -0600, "Gary L. Scott"
<garyscott@ev1.net> wrote:

<snip requoted>

>I don't really think the quality of the work is bad. In fact, I think=20
>it fairly high in quality. My only real issue would be with=20
>prioritization of features to be worked. The fact that my own desired=20
>features are not being considered (I've never submitted a proposal)=20
>would be substantially my own fault I would assume. I think the current=
=20
>committee is technically capable of quite a good design for any proposed=
=20
>feature (notwithstanding workload issues).

We probably won't have bit-strings because:

Nobody on WG5 or J3 really wants them,

there's no great clamor to motivate the vendors to produce them
as an extension, and

they're hard to do (bit addresses are a new class of address
on most current hardware).

If you volunteer for J3, you could at least solve the first issue.
And you could advocate solutions for the second and third.

--=20
Cheers!

Dan Nagle
Purple Sage Computing Solutions, Inc.

Dick Hendrickson

2005-02-21, 4:01 pm

bv wrote:

> Michael Metcalf wrote:
>
>
>
> This is worse than I imagined, and it needs to change!
> Imagine a member of Ferrari's design team where his only "experience"
> was a one time attendance at the F1 race!? It would be unthinkable, just
> as it should be unthinkable that anyone w/o advanced degree in hard
> sciences and a minimum of at least 15 years of scientific programming
> should do as much as serve coffee at the committee's meetings.
>
> How do we change the system?
> I'd suggest that an arrangement be made with netlib admin to solicit
> feedback which would serve to determine what *real* users need or want,
> and a *qualified* committee in conjunction with vendors would work only
> on evaluating and adopting the top requests.
>

I've seen a couple of comments in this thread along the
lines of "the committee doesn't do what the vendors know
needs to be done." Sure, there's some truth to that.
But, but, but.

For good or bad, there are about a dozen people who
regularily attend J3 meetings. Six are vendor
representatives. The secretary, the editor, the heads of
the interpretations sub-group, the "JoR" subgroup and the
High Performance Computing subgroup are all vendor
representatives. (Some people wear more than one hat).
Only the Data Concepts subgroup is headed by a non
vendor.

"JoR" is the Journal of Requirements subgroups and is the
SG that makes preliminary recomendations on new feature
proposals. Attendance fluctuates from meeting to
meeting, but at the last meeting JOR was half vendors
and half people who used to work for vendors.

J3 is hardly a perfect committee. Mike Metcalf made
many good and valid points in his editorial. But it
really isn't a case of pointy headed academics forcing
things down the throats of people who aren't represented.

Dick Hendrickson

Gary L. Scott

2005-02-22, 8:58 am

Dan Nagle

2005-02-22, 8:58 am

Hello,

On Mon, 21 Feb 2005 21:12:35 -0600, "Gary L. Scott"
<garyscott@ev1.net> wrote:

<snip>

>I'm presently taking graduate courses, I have a real estate business on=20
>the side, I maintain 3 web sites, I run every morning and work out at=20
>the gym every evening, I have a normal job during the day (or as needs=20
>dictate other shifts)...when oh when would I find the time? Maybe when=20
>I retire (no sooner than 15 years).

I'm a grad student, I run a business on the side. I only maintain
two web sites. I ride a bicycle for fitness, take tai kwon dow,
exercise for fitness, and had a round of neck and back therapy
last term. I've been notified that I have jury duty soon.
I'm working on a proposal for a grant.

It's a choice whether it's J3, or Habitat for Humanity or any other
volunteer activity.

--=20
Cheers!

Dan Nagle
Purple Sage Computing Solutions, Inc.

Richard E Maine

2005-02-22, 4:01 pm

In article <plnj11tfsrs1775jbhtpcpfahst991ql9o@4ax.com>,
Dan Nagle <dannagle@verizon.net> wrote:

> And you'd appoint yourself in charge of determining what "real" means,
> and what the "qualifications" are and who meets them?

Aye. There's the rub.

In the real world that I'm familiar with, you don't usually get the
chance to be emperor/king/whatever. You have to work with other people
rather than just command them from above.

The current ISO standard system has many flaws; I don't think I know a
single member of any of the committees who would argue otherwise, though
they might disagree about which specific things are flaws. An old adage
comes to mind - about having the courage to change the things that you
can, the serenity to accept the things that you can't change, and the
wisdom to know the difference. Committee members at various times try to
change things about the system or work with it in spite of its flaws
(often bitching a lot over dinner, and occasionally in print :-)).

But a person thinks that the fundamental problems come from paying too
much attention to (insert your favorite stereotype group here), and that
all would be great if everyone else would just do what that person says,
well... good luck in the real world outside of comp.lang.fortran. In my
view, that kind of attitude is part of a problem rather than part of a
solution. If one ever actually gets something real done with an attitude
like that, I'll be duly impressed.

I think that debates here about technical matters are fine. If one
thinks that something in the language is done wrong and should be fixed,
have at it. But when the debates turn into things like gripes about the
kinds of people "allowed" on the committee (usually, it seems, based on
completely uninformed characterizations)... Well, one is of course free
to debate that also.... or one could contribute to real-world solutions
(and if the real-world solution one wants to contribute to is something
more along the line of habitat for humanity, I certainly wouldn't
criticize that choice)..

--
Richard Maine | Good judgment comes from experience;
email: my first.last at org.domain | experience comes from bad judgment.
org: nasa, domain: gov | -- Mark Twain

Gary L. Scott

2005-02-23, 3:59 am

Dan Nagle wrote:
> Hello,
>
> On Mon, 21 Feb 2005 21:12:35 -0600, "Gary L. Scott"
> <garyscott@ev1.net> wrote:
>
> <snip>
>
>
>
> I'm a grad student, I run a business on the side. I only maintain
> two web sites. I ride a bicycle for fitness, take tai kwon dow,
> exercise for fitness, and had a round of neck and back therapy
> last term. I've been notified that I have jury duty soon.
> I'm working on a proposal for a grant.
>
> It's a choice whether it's J3, or Habitat for Humanity or any other
> volunteer activity.
>
It certainly is...but my plate is full until I decide otherwise.

--

Gary Scott
mailto:garyscott@ev1.net

Fortran Library: http://www.fortranlib.com

Support the Original G95 Project: http://www.g95.org
-OR-
Support the GNU GFortran Project: http://gcc.gnu.org/fortran/index.html

Why are there two? God only knows.

Democracy is two wolves and a sheep, voting on what to eat for dinner...
Liberty is a well armed sheep contesting the vote. - Thomas Jefferson

Gary L. Scott

2005-02-23, 4:00 pm

Dan Nagle wrote:

> On Sun, 20 Feb 2005 19:14:59 -0600, "Gary L. Scott"
> <garyscott@ev1.net> wrote:
>
> <snip requoted>
>
>
>
> We probably won't have bit-strings because:
>
> Nobody on WG5 or J3 really wants them,
>
> there's no great clamor to motivate the vendors to produce them
> as an extension, and
>
> they're hard to do (bit addresses are a new class of address
> on most current hardware).

I envision "bit string" as a simple packed array of bits. The most
efficient packing arrangement (bytes, integers) would have almost no
impact to basic functionality. To get the bit number, you simply
multiply/divide by the natural storage unit (and perhaps add +/-1).
What are you referring to as a "bit string" that would require something
other than an integer array and the already present bit intrinsics.
I've used the above method to draw bi-level bitmap images for decades on
several different platforms (same algorithm, just adjust the basic
storage unit size).

Now a second part to my proposal would include a "bit object" which
would propose many additional useful features like scale factors,
MSB/LSB, any number of bits, different encodings (BCD, 1's complement,
etc.). This would be somewhat like a derived type but with bit (string)
definitions as well as intrisic types.

I think the lack of clamor is because those interested moved on to other
languages out of frustration with Fortran's lack of ability. We need to
study other languages used for system programming, see what they're good
at and what they're not, and consider what would improve Fortran for a
particularly important programming domain. We need to bring people back
to Fortran. We need to provide a reason(s).
>
> If you volunteer for J3, you could at least solve the first issue.
> And you could advocate solutions for the second and third.
>

--

Gary Scott
mailto:garyscott@ev1.net

Fortran Library: http://www.fortranlib.com

Support the Original G95 Project: http://www.g95.org
-OR-
Support the GNU GFortran Project: http://gcc.gnu.org/fortran/index.html

Why are there two? God only knows.

Democracy is two wolves and a sheep, voting on what to eat for dinner...
Liberty is a well armed sheep contesting the vote. - Thomas Jefferson

Michael Metcalf

2005-02-23, 4:00 pm

"Gary L. Scott" <garyscott@ev1.net> wrote in message
news:111p09kp845b5ad@corp.supernews.com...
>
> I think the lack of clamor [for bits] is because those interested moved on
to other
> languages out of frustration with Fortran's lack of ability.

Exactly. This battle was lost in the late '80s. The bit intrinsics were the
compromise solution.

Regards,

Mike Metcalf

2005-02-24, 8:58 pm

Dan Nagle wrote:
>
>
> And you'd appoint yourself in charge of determining what "real" means,
> and what the "qualifications" are and who meets them?

You're not paying attention! - suggestion implies what it says: *real*
users, using "real* compilers, doing "real* problems are the ones who
would have a say what the *real* language needs or how it should evolve.
Likewise for committee *qualifications* - language philosophers need not
apply, only those with years of *reals* (listed above) under their
belts.

Richard E Maine

2005-02-24, 8:58 pm

In article <421E468C.A8F6ADF7@Xsdynamix.com>, bv <bvoh@Xsdynamix.com>
wrote:

> Dan Nagle wrote:

>
> You're not paying attention! - suggestion implies what it says: *real*
> users, using "real* compilers, doing "real* problems are the ones who
> would have a say what the *real* language needs or how it should evolve.
> Likewise for committee *qualifications* - language philosophers need not
> apply, only those with years of *reals* (listed above) under their
> belts.

Then I guess by those definitions, we must be in good shape now, because
every single member of the committee has years of experience with real
users, real compilers, and real applications. I know everyone who is or
has been a member for a decade and a half. I can and do testify to their
real experience. Do you claim to know any members who don't? If so which
ones? Can you name names?

I have 35 years of such real experience. Is that sufficient? If more is
needed, a few members would still qualify. I don't think there are any
with less than a decade; one of the newer members is young enough that
he might possibly still be slightly under that, but I doubt it.

While their aren't formal requirements for such experience, it just
happens that everyone who has ever bothered to apply for membership
(certainly in the 15 years I've been involved) does have such
experience. I don't find this incredibly surprising.

If you going to start saying that some users, compilers, or problems
don't count because they aren't "real", then I'll claim that you didn't
listen to Dan. He asked who was in charge of defining "real". I see
nothing that even approximates a definition above. All I see is
repetition rather than anything that approximates a definition. I
conclude that either you don't really know anything about the committee
members (which seems likely, as I've never seen *YOU* at a meeting or
heard, even indirectly, of you having serious correspondence with any
members) or, if you do know what you are talking about, then your
definition of "real" is different from mine.

If your definition of "real" is different from mine, then we are back to
Dan's question. Who gets to define "real"? Who is it that candidates
apply to, since you specifically mention applications. Yes, I know the
answer to that. Obviously, you get to define it. Somehow, that's not a
surprise either. :-) There is probably a version of reality in which
that makes sense, but it isn't my reality.

--
Richard Maine | Good judgment comes from experience;
email: my first.last at org.domain | experience comes from bad judgment.
org: nasa, domain: gov | -- Mark Twain

Ken Fairfield

2005-02-24, 8:58 pm

bv wrote:
> Dan Nagle wrote:
>
>
>
> You're not paying attention! - suggestion implies what it says: *real*
> users, using "real* compilers, doing "real* problems are the ones who
> would have a say what the *real* language needs or how it should evolve.
> Likewise for committee *qualifications* - language philosophers need not
> apply, only those with years of *reals* (listed above) under their
> belts.

*ploink*

This response is bull-headed single-interest group politics
at best, stupidity at worst. Users, compiler writers, and
language designers all need to be involved and represented
to have a reasonable chance of a good, usable language that
compiler writers can implement.

-Ken
--
I don't speak for Intel, Intel doesn't speak for me...

Ken Fairfield
D1C Automation VMS System Support
who: kenneth dot h dot fairfield
where: intel dot com

Sponsored Links