maine@elxsi.dfrf.nasa.gov (Richard Maine) (12/20/90)
On 20 Dec 90 16:21:43 GMT, zctsfisher@qut.edu.au said:
zctsfisher> The standard for Fortran 77 is of course ANSI X3.9-1978
zctsfisher> and this standard is easily accessible and clear. It takes
zctsfisher> only a small effort to learn your way around it. There is
zctsfisher> no need to debate, as this group often does, what is in
zctsfisher> the standard. It is only necessary to read it.
...
zctsfisher> I only go to compiler supplier's handbooks when I
zctsfisher> specifically wish to do something which cannot be done
zctsfisher> efficiently in proper Fortran or ...
Amen! In the case of one of our vendors, who I'll refrain from naming,
I tried unsucessfully to get them to throw out 90+% of their manual
and instead sell copies of the standard, letting their manual document
only deviations, extensions, and implementation specifics. I found the
vendor manual to be poorly organized, poorly written, imprecise, and
occasionally blatantly wrong on subjects that were well covered in the
standard and were not system-dependent. The manual was obviously a
low-priority effort that did not get the attention needed to do a
good job. I tried to convince them that if they refrained from
trying to paraphrase the standard in their own words, but put their
effort in documenting the things that the standard doesn't, they could
produce a more useful document for less work. I never did succeed.
Although this was a particularly bad case, I'd offer the same advice
to other vendors also. Although not tutorial, the standard is not
really hard to read for someone used to reading technical documentation
(the only kind of people that bother reading vendor manuals anyway).
It is clearer than many vendor manuals and this isn't surprising.
A lot of work went into the standard, and if you want to write
something that is clearer, while still complete and accurate, it is
going to take a lot of work. Now if you want to skip the complicated
questions and present the basics in a tutorial fashion, that's
another matter, but a good vendor manual can't get by with skipping
the complicated questions.
This isn't to say the standard is "perfect" (whatever that word means).
Just that supplemental documentation should address the areas of
imperfection instead of trying to paraphrase the large body of it that
is fine as it stands.
--
Richard Maine
maine@elxsi.dfrf.nasa.gov [130.134.64.6]zctsfisher@qut.edu.au (12/21/90)
During the recent discussion on "Intrinsics" I found the number of
postings quoting other than the actual Standard very disappointing.
The standard for Fortran 77 is of course ANSI X3.9-1978 and this
standard is easily accessible and clear. It takes only a small effort to
learn your way around it. There is no need to debate, as this group
often does, what is in the standard. It is only necessary to read it.
(There are of course parts of the standard where the meaning of the
standard is debatable but that is another matter. There is no other
authority.)
Given the interest in this group in maintaining standards may I urge
readers to get a copy of the standard, learn your way around it and
do as I do and use it as your reference manual. (It doesn't take long to
learn to ignore every second page - the little used Subset Language.)
I only go to compiler supplier's handbooks when I specifically wish to
do something which cannot be done efficiently in proper Fortran or when
the compiler appears to be wrong. In the latter case if the different
interpretation of the compiler is not listed by its supplier as a known
variation from the standard it should be reported as an error.
The standard may not be the ideal as a learning text so other books
designed for this purpose have their place but they should be abandoned
once the basics are learnt.
We can all learn too by occasionally browsing right through the
standard. It is amazing to discover that facilities exist we have not
used which overcome supposed deficiencies in the language.
In brief, if you have not read it in the Standard you cannot say
authoritatively what is Standard no matter what other source you have
used and we should all strive to be authoritative.
W.J.G.(Bill) FISHER,
Director, Computing Services (Gardens Point),
Queensland University of Technology
Mail P.O. Box 2434 Ph (Inatl) (61+7) 223 2123 (w)
Brisbane (61+7) 378 2810 (h)
Queensland 4001 Fax (61+7) 221 1461
Australia Internet w.fisher@qut.edu.au
Telex AA44699 VAX PSI 505272223015::W.FISHERbernhold@qtp.ufl.edu (David E. Bernholdt) (12/21/90)
In article <MAINE.90Dec19195753@altair.dfrf.nasa.gov> maine@elxsi.dfrf.nasa.gov (Richard Maine) writes: >Amen! In the case of one of our vendors, who I'll refrain from naming, >I tried unsucessfully to get them to throw out 90+% of their manual >and instead sell copies of the standard, letting their manual document >only deviations, extensions, and implementation specifics. FPS (Floating Point Systems) used to do exactly this with their FPS-x64 product line. In fact when we retired our '164, I grabbed the copy of the standard from the manual set for myself. Interestingly, they don't do this with their FPS-500 product line -- I suspect most of that came from Celerity. IMHO, the _best_ vendor Fortran manual/reference guide comes from DEC for VAX/VMS Fortran. The one I'm thinking of covers both the standard and the extensions, putting extensions in a different color of ink, so they are readily apparent. I can't comment on its treatment of the standard, because I rarely used it for such questions. However, I think it is quite nice to have the whole language (standard and extensions) set out in a single manual -- helps avoid getting the "runaround". >On 20 Dec 90 16:21:43 GMT, zctsfisher@qut.edu.au said: >zctsfisher> The standard for Fortran 77 is of course ANSI X3.9-1978 >zctsfisher> and this standard is easily accessible and clear. I quite agree with this, though there is one book I use more than the standard itself: Harry Katzan, Jr., Fortran77, Van Nostrand Reinhold, 1978, ISBN 0-442-24278-6. This is the book from which I _learned_ Fortran. Its comprehensive and inexpensive ($11 in paperback in 1982). I usually check it before going to the standard, and rarely do I need to go any further. I highly recommend it. -- David Bernholdt bernhold@qtp.ufl.edu Quantum Theory Project bernhold@ufpine.bitnet University of Florida Gainesville, FL 32611 904/392 6365
oles@kelvin.uio.no (Ole Swang) (12/21/90)
Trivial question: How do I get a copy of the standard? -- ----------------------------------------------------------------------- Ole Swang assistant lecturer, Dept. of Chemistry, U .of Oslo -----------------------------------------------------------------------
maine@elxsi.dfrf.nasa.gov (Richard Maine) (12/22/90)
On 21 Dec 90 09:35:24 GMT, oles@kelvin.uio.no (Ole Swang) said: Ole> Trivial question: How do I get a copy of the standard? Write to Sales dept. American National Standards Institute, Inc. 1430 Broadway New York, NY 10018 U.S.A. Ask for ANSI X3.9-1978, "Programming Language FORTRAN". When I ordered mine, it was $19.95, plus I think a few bucks for shipping and handling. That was quite a while ago. I'm sure its gone up since, but it should still be reasonable as technical books go. (The actual standards are cheaper than the drafts, which are not so reasonable; I sort of understand the reasons, but it still grates). I can't find a phone number handy. If you can find one from information or somebody else on the net, it might save you sending a first letter just to inquire about the price. They took my personal check; I don't recall what other payment forms they might accept. There may be a better way in Europe. I don't have any information on that. -- Richard Maine maine@elxsi.dfrf.nasa.gov [130.134.64.6]
bernhold@qtp.ufl.edu (David E. Bernholdt) (12/22/90)
In article <1214@orange19.qtp.ufl.edu> I wrrote: >In article <MAINE.90Dec19195753@altair.dfrf.nasa.gov> maine@elxsi.dfrf.nasa.gov (Richard Maine) writes: >>Amen! In the case of one of our vendors, who I'll refrain from naming, >>I tried unsucessfully to get them to throw out 90+% of their manual >>and instead sell copies of the standard, letting their manual document >>only deviations, extensions, and implementation specifics. > >FPS (Floating Point Systems) used to do exactly this with their >FPS-x64 product line. I got an interesting response from someone at FPS about this. It seems that FPS had to stop producing the manual because someone said it was too much like the standard (it _was_ the standard, in a binder with an additional section of FPS's extensions) and wanted FPS to pay royalties. [No flames about the legalities, please -- I'm just relating the story!] Seems like the Fortran user community would be better served if vendors wouldn't have to worry about such things -- maybe we would have more people familiar with the standard then! -- David Bernholdt bernhold@qtp.ufl.edu Quantum Theory Project bernhold@ufpine.bitnet University of Florida Gainesville, FL 32611 904/392 6365
dik@cwi.nl (Dik T. Winter) (12/22/90)
In article <1218@orange19.qtp.ufl.edu> bernhold@qtp.ufl.edu (David E. Bernholdt) writes: > I got an interesting response from someone at FPS about this. It > seems that FPS had to stop producing the manual because someone said > it was too much like the standard (it _was_ the standard, in a binder > with an additional section of FPS's extensions) and wanted FPS to pay > royalties. More interesting would by how much was expected in the form of royalties. Would the price indeed be much too high? > > Seems like the Fortran user community would be better served if > vendors wouldn't have to worry about such things -- maybe we would > have more people familiar with the standard then! Of course. But who then will pay the standardization process (which is not cheap)? -- dik t. winter, cwi, amsterdam, nederland dik@cwi.nl
jgd@convex.csd.uwm.edu (John G Dobnick) (12/22/90)
From article <1218@orange19.qtp.ufl.edu>, by bernhold@qtp.ufl.edu (David E. Bernholdt): > > I got an interesting response from someone at FPS about this. It > seems that FPS had to stop producing the manual because someone said > it was too much like the standard (it _was_ the standard, in a binder > with an additional section of FPS's extensions) and wanted FPS to pay > royalties. Perhaps if the vendor arranged a "bulk purchase" of the standard, and then shipped it? Seems like that should, among others things, keep the standards groups happy. (And solvent.) Just a random thought. -- John G Dobnick (JGD2) Computing Services Division @ University of Wisconsin - Milwaukee INTERNET: jgd@csd4.csd.uwm.edu ATTnet: (414) 229-5727 UUCP: uunet!uwm!csd4.csd.uwm.edu!jgd "Knowing how things work is the basis for appreciation, and is thus a source of civilized delight." -- William Safire
levine@liege.ics.uci.edu (David Levine) (12/27/90)
Richard Maine writes: >Ole> Trivial question: How do I get a copy of the standard? >Write to > Sales dept. > American National Standards Institute, Inc. > 1430 Broadway > New York, NY 10018 > U.S.A. >Ask for ANSI X3.9-1978, "Programming Language FORTRAN". phone 212-354-3300 price is now $35.00 + shipping