X3J3/96-135 Date: August 12, 1996 To: X3J3 From: Jerrold Wagener Subject: POSIX binding to Fortran ----from Michael Hanna, 21 May 1996: ++ From jwagener@ionet.net Mon May 20 11:43:07 1996 ++ Date: Mon, 20 May 1996 12:42:06 -0600 ++ From: "Jerrold L. Wagener" ++ To: Craig Dedo , mjhanna@sandia.gov (Michael J Hannah) ++ Subject: Re: POSIX Binding to Fortran ++ Content-Disposition: Inline ++ ++ ++ > >since the completion of the standard. X3J3 should note that this ++ > >Fortran binding standard is coming up on its 5 year sunset date that ++ > >will require it to either be reconfirmed by IEEE, or to be expired as ++ > >a standard. X3J3 may wish to communicate a position on this issue ++ > >to IEEE for their consideration. I firmly believe that the ++ > >continued viability of the Fortran language itself is in serious ++ > >jeapordy, and that retaining at least some kind of binding to POSIX ++ > >is essential to its continued viability, but then I am prejudiced. ++ > ++ ++ > Thanks for the advice on the IEEE sunset date on POSIX.9. I ++ > have taken the liberty of sending a copy of this to Jerrold Wagener. ++ > ++ ++ Thanks Craig, and Mike, for this info. The current Fortran binding to POSIX ++ is a Fortran 77 binding. That also makes it a Fortran 90 binding, even though ++ it does not make use of some of the "nice" Fortran 90 features that a "full" ++ binding to Fortran 90 would employ. WG5 curently has a Fortran 90 binding to ++ POSIX as a Fortran 2000 requirement (presumably the "90" would change to at ++ least "95" and perhaps "2000"). If that requirement is reaffirmed (and ++ "firmed") at the WG5 meeting in July, then it would make sense to let the ++ current binding drop; otherwise it makes sense to keep it. I disagree. Considering the amount of time that it is likely to take to get a new binding developed and approved, I would recommend retaining the current standard until that occurs. Perhaps it could be retained as an "archived" standard, e.g. no changes, but still defined. The drop date is 1997. The existence of Fortran 2000 is likely to be at the earliest 2000. Keeping the current standard for an additional 5 year term would mean it would drop in 2002, just in time for the new standard to be out. And since the current standard specifically references binding to Fortran 77, having both in existence for a year overlap is no problem because they would be binding to essentially different languages. In other words, even if you plan to have a POSIX binding in Fortran 2000, I think the existing standard needs to be retained for at least one more five year period. ++ I'll make sure that WG5 has this information, and understands these options, ++ at its July meeting. Thanks again for the info. (And good to hear from you ++ again, Mike, after so long.) ++ ++ Jerry ++ ++ (PS - and sorry to hear that Sandia is wasting federal funds chasing C++ ++ rather than using a good language like F90. :-) ++ ----from Michael Hanna, 22 May 1996: ++ > In other words, even if you plan to have a POSIX binding in ++ > Fortran 2000, I think the existing standard needs to be retained for ++ > at least one more five year period. ++ ++ Good point. I agree. ++ What exactly needs to be done to cause this to happen? ++ ++ Jerry I recommend two separate actions. 1) A coordination letter from X3J3 to the IEEE POSIX steering committee requesting that IEEE renew the standard for a further 5 years. 2) A coordination letter frow WG5 (is that the languages number?) to WG15 as an ANSI position to renew the ANSI standard for a further 5 years. I believe that if X3J3 requests it in this manner it will be a pro forma approval, especially if the letter mentions that X3J3 plans a Fortran 90 replacement and is only requesting retention of the F77-based standard until X3J3 can complete the F90 replacement, which is proposed for F-2000. If it is clear that the Fortran community is asking for it, and yet they make clear their schedule for replacing it, I see no problems arising. Michael ----from Michael Hanna, 23 May 1996: ++ Date: Wed, 22 May 1996 10:30:19 -0600 ++ From: "Jerrold L. Wagener" ++ ++ Sounds good, Michael; I will plan to pursue this approach with both X3J3 and ++ WG5. Any chance you can give me the exact numbers of the documents, both ++ nationally and internationally that we wish to renew? IEEE Std 1003.9-1992 IEEE Standard for Information Technology-- POSIX FORTRAN 77 Language Interfaces-- Part 1: Binding for System Application Program Interface (API) ISBN 1-55937-230-3 I do not know the ANSI number, but was informed that it was approved as an ANSI standard. The IEEE office may have that information. It never went forward to ISO since it references FORTRAN 77 which is now only an ANSI archived standard and does not exist in the ISO arena. There was some talk at one time of registering it for a "fast-track" ballot as an ISO Technical Report, Type 3, but do not believe that ever happened. Michael ----from Michael Hanna, 26 Jun 1996: Jerry, Lowell Johnson of the POSIX IEEE committee just left me voice mail about the sunset date of the POSIX binding to Fortran. I called him back and left voice mail that I had alerted you to this issue and that X3J3 was interested in retaining the existing binding while X3J3 continues to work on a replacement in the syntax of Fortran 90/2000. I gave him your e-mail address and phone number for further follow up. His phone is: (612)635-7305 at Unisys the last e-mail address I have for him is: 3lgj@rsvl.unisys.com Michael ----from Andrew Tait, 18 Jul 1996 To: Shane McCarron From: Andrew Tait Cc: Jerry Wagener (Chairman X3J3), Miles Ellis (Convenor WG5) Subject: Open Systems Standards I have just finished browsing through the "Open Systems Standards" guide recently published by Uniforum and of which you are the author. I would like to comment on the information about Fortran contained in this guide. ISO/IEC 1539:1980 has not been the international Fortran standard for several years. It was replaced by ISO/IEC 1539:1991, "Information technology - Programming languages - Fortran" which defines the language known as Fortran 90. This standard will soon be updated to Fortran 95. Furthermore, X3 has recently initiated a letter ballot of its members to withdraw ANSI X3.9 - 1978 Programming Language FORTRAN (known as FORTRAN 77). This ballot closes on 96/08/06. If FORTRAN 77 is withdrawn Fortran 90 will be the only US Fortran standard. At the time IEEE Std 1003.9-1992 was being developed the committees working on the definition of Fortran 90 (i.e. X3J3 and WG5 - I participated in both) questioned the wisdom of binding to an obsolescent standard. Since FORTRAN 77 may soon be history one can only question the value of a binding to a defunct standard. Work on a replacement for IEEE Std 1003.9-1992 based on Fortran 90 was started but, to my knowledge, there has been no activity on it for several years. Since Fortran 90 compilers are now widely available, adhering to IEEE Std 1003.9-1992 may not be a good idea as it probably limits the effective use of the advanced features of Fortran 90. Andrew D. Tait Director, UTS Development Amdahl Corporation 1250 East Arques Avenue (m/s 278) P.O. Box 3470 Sunnyvale, California 94088-3470 USA phone: (408) 746-7370 fax: (408) 737-5338 email: adt10@oes.amdahl.com