psmith@convex.UUCP (02/10/88)
There has been a lot of "religion" expressed on the net over FORTRAN 8x in
the last few days. In answer to some of the questions, I have included a
summary of the balloting in this paper.
The key issue right now is not what your religion is on the subject, but
the fact that the ballot period is about to close. IF YOU DON'T EXPRESS
YOUR OPINION NOW...YOU WILL NOT GET AN OPPORTUNITY TO EXPRESS IT LATER...
Letters MUST be at this address no later than February 23, 1988.
Send your comments to:
X3 Secretariat
CBEMA
311 First Street NW, Suite 500
Washington, DC 20001-2178
Now, a few notes on some of the comments from the net...
Kurt Hirchert:
I count about 9 government agencies in the list...You are
certainly welcome to comment again, but if you do, please give us the
correct information in each case... My counts are later in this paper.
Also, there is NO subset defined for this proposed standard...Maybe the
vendors are free to "implement a subset, just don't call it standard
Fortran" but I don't think that would be a wise move for any rational
vendor to take. The bottom line is that FORTRAN 8x is a much larger
language than FORTRAN 77. We can argue about the magnitude...but it
is much larger. I tend to believe Richard Weaver from IBM when he
compares it to both Ada and Pascal in his ballot comments. We did
not include the Pascal comparison...but it makes 8x look even larger.
The ballot comments are public record...anyone can get a copy. I
believe that Dick is right when he states: "The difficulty with 8x
lies not in the quality of the work that X3J3 has done, but rather in
the somewhat surprising observation that the result of X3J3's work
is a new language and not a revision of Fortran." Maybe it corresponds
more to Pascal...but the term on the street is "Adatran."
In the comments that 8x warns that Decremented statements will become
obsolete over a "very, very long period of time", the official rules
of ANSI state that a standard must be re-certified every 5 years or
a revision made at that time. With the current rules that
Decremented statements are moved to Obsolescent statements and then
may be removed...this COULD be done within a 10 year time period...
COULD...
It also seems odd that the proposed standard is so complex that someone
had to write a book..."Fortran 8x Explained" to explain it?? That
flags me that we have a problem if we can read the proposed standard
and understand it...without having an explanation.
Kent, the man from xanth
Maybe you are not interested in the performance of the compilers...just
being a user, but that is the number 2 issue with the customers we have...
That is customer feedback...not the "desire" of the compiler group.
Sometimes, major surgery on very ill patients results in them croaking
on the operating table...
On the issue of "every vendors favorite construct..." the project
proposal for the FORTRAN 8x effort states: "it is likely that
augmentations (new features) will mainly be drawn from functionality
that exists in advanced implementations of existing processors..."
Of the 4 major areas of new functionality, only part of one area
exists in any advanced implementation. Did the committee violate
the charter??
On the issue of dusty decks...there has already been other responses
on the net. How about some more opinions in this area???
On the issue of carrying along deprecated features...and cost
effectiveness... Let me quote again from the project proposal:
"One of FORTRAN's most important characteristics is that
efficient processors can be implemented at a reasonable cost. One
of the most important goals during the next revision will be to
retain this characteristic." When the major vendors and compiler
writers are telling the public it going to take a long time to
implement and is going to be expensive (and YOU will bear the cost
because the vendor will pass it on to you) you might want to listen...
Again a quote from Dick Weaver's IBM ballot: "Permits inexpensive
implementations? NO. 8x impacts implementation costs in several
ways:
- The size of the language; there is simply lots more work to do
- Dependent compilation. The USE statement requires information from
other (prior) compilations. This requires new library servies
(an "environment"), both to store and control this information.
- Other operating system services are impacted; Link editing must
now handle scopes for external names, for example. "
I hope you are right that the FORTRAN community will be willing to
bear the cost of moving from "the old to the new compilers." It is
NOT going to be cheap.
I think it is interesting that you find exception with some of the
overview material that was copied from directly from the FORWARD of
the draft proposal, pages i to iii. I left NOTHING out unless I made
a typo... You had comments on Derived Data Types, Modular Definitions,
and Language Evolution... I would suggest that you expland on your
comments and send them to the committee.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Having made more comments, I shall sit back and wait for the FLAMES...
Time is running short. SEND YOUR COMMENTS TO THE COMMITTEE...and
let them know your feelings.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
The following is a summary of the X3J3 Ballots. The Letter Ballot was held
in December, 1986 and the Role Call in May, 1987. The addition of the
affiliations for each group is mine. I count
Universities - 4
Vendors - 15
Government - 9
Businesses - 6
??? - 4
Summary of X3J3 Ballots...
Letter Role
Company Name A Ballot Call
1. Advanced Computer Techniques Adamezyk, J. Stephen ? Aff Neg
2. National CTR for Atmospheric Res. Adams, Jeanne C. G Aff Aff
3. Gould, Inc. Barber, Graham V Aff Resigned
Phillamore, David Aff, New
4. UNISYS Bowe, Valerie V Neg Neg
5. UNICOMP (University of New Mexico) Brainerd, Walter U Aff Aff
6. Hewlett-Packard Burch, Carl D. V Aff Aff
7. Aberdeen Proving Ground Campbell, Lloyd G Aff Aff
8. CSP Inc. Crowley, Ted V Aff Absent
9. Harris Allison, Robert V Aff Neg
10. Oxford University Ellis, T. Miles U Aff Aff
11. Bell Communications Research Freeman, Murray B Aff Absent
12. Masscomp Gridley, Kurt V * Aff,New
13. DEC Harris, Kevin V Neg Neg
14. CRAY Hendrickson, Richart V Neg Aff
15. Natl Ctr Supercomputting Appl Hirchert, Kurt W. U Aff Aff
16. Data General Hoover, Tracy Ann V Aff Aff
17. Prime Johnson, Andrew V Aff Aff
18. Peritus International, Inc. Lakhwara, Anil ? Neg Absent
19. EG&G Idaho Incorporated Marshall, Neldon G Aff Aff
20. Grumman Aircraft Syst. Martin, Bruce B Aff Aff
21. Lawrence Livermore Lab Martin, Jeanne G Aff Aff
22. Los Alamos National Lab Marusak, Alex G Aff Neg
23. Computer Sciences Corp. Matheny, James B Aff Aff
24. DD CERN Metcalf, Michael G Aff Aff
25. Edinburgh Portable Compilers Millard, G.G. B Aff Aff
26. SLAC Moss, Leonard G Neg Neg
27. Boeing Phillips, Ivor B Neg Neg
28. Control Data Corp Ragan, Richard V Aff Aff
29. Aere Harwell Reid, J.K. G Aff Aff
30. Data-Term Schenk, Werner ? Aff Aff
31. University of Liverpool Schonfelder, J.L. U Aff Aff
32. Argonne National Lab Smith, Brian G Aff Aff
33. Alliant Computer Corp Swift, Richard C V Aff Aff
34. Concurrent Computer Corp Thompson, Brian V Aff Aff
35. Amoco Production Co Wagener, Jerrold B Aff Aff
36. NCC, Manchester, UK Wearing, Allison ? * Neg
37. IBM Weaver, Richard V Neg Neg
38. International Computers, Inc. Wilson, Alan V? Aff Aff