Everyone:
I am having problems getting CVF 6.6C to compile a simple program using a
command options file. When I compile any program, I get the message:
df: error: Negation of '/NoF66' switch not allowed

This occurs even on a simple "Hello, World" program.

The command I am using is:
df @cvf_debug.opt hello_F95.f90 /link

Unfortunately, the compiler does not give any clue as to what the conflict
is, nor how to resolve the conflict. I have tried looking through the on-line
documentation without success. Could someone please tell me how to resolve this
conflict?

Here is a copy of the command options file.
[Begin command options file]
/NoAlignment
/NoAltParam
/Architecture=P6
/Assume=(ByteRecl,NoDummy_Aliases,Minus0)
/NoAutomatic
/Browser
/Check=(Bounds,Flawed_Pentium...

Craig Dedo wrote:

...

...

...

"dpb" non.net> wrote in message news:feeqe7$4i3$1@aioe.org...

dpb non.net> wrote:

As an aside, note that in spite of its name and description, the /f66
switch probably has nothing do do with f66 compatibility. At least most
switches that go by names like that don't.

An appendix in the f77 standard has a long list (27 items, IIRC) of
incompatibilities between f66 and f77 (and thus also later standards).
Most so-called f66 switches don't do anything about any of these.
Instead, they usually specify that zero-trip-count DO loops execute the
body of the loop one time. It is a widely held misconception that this
behavior is an f66 feature. The switch names encourage this
misconception. It ain't so. If this were an f66 feature, it would have
been in the list of incompatibities in the above-mentioned appendix.

The f66 standard did not specify that such DO loops executed one pass of
the loop. Instead, the f66 standard explicitly and clearly said that
such DO loops were invalid. It happens that the one-trip behavior was a
consequence of the implementation choices of some particular f66
compilers. It was not so in all f66 compilers; I personally worked with
counterexamples.

Richard Maine wrote:

...snip...

Having never looked at what it does, in CVF it

/[no]f66
Syntax:

/f66 or /nof66

dpb non.net> wrote:

That's the one I was talking about as not accurately described as an f66
feature. I hadn't checked cvf in particular, for there are several
compilerts where this is the only thing that the similarly-named switch
does.

I'd forgotten that one, but I do see it in f77's list of
incompatibilities. That's one out of 24 (I think I said 27 before
without checking; it is actually 24).

On Oct 9, 12:28 pm, nos...@see.signature (Richard Maine) wrote:

For the DEC/Compaq/Intel compilers, interpret the "f66" (or similar)
switch as meaning "do what DEC FORTRAN-IV did". As you point out,
most of these differences involve behaviors that were left unstated by
the earlier standard and then specified in the later standard. If the
new standard's behavior conflicts with the old compiler behavior, then
a switch is needed. We have also added switches to adjust behavior
specified by F90 (left unstated in F77) and even F2003 compared to F90/
F95.

For many of our customers, compatibility with F66-era compilers is
what they mean when they ask if we support FORTRAN-66, even though
most of their programs are not strict F66-compliant. The naming of
the switch makes it easy for them to understand.

Steve

Richard Maine wrote:

dpb non.net> wrote:

Me too, of course. It might not have come across quite right in the way
I said it, but I did not mean that as an indictment of other people in
contrast to myself. I was also part of the era, and acted much like
those other people. I like to think that I eventually learned from the
experiences. Thus the quote in my signature.

Richard Maine wrote:

Oh, I wasn't responding to the tone, simply got me thinking of how I got
to the point...or the other case where the same guy sold a job to
evaluate the algorithms/implementation of one of the new codes written
in response to the TMI incident that turned out, when we got the source
to be about 100k lines of the worst mishmash of cr^hode I've _ever_
seen. The customer PM, when informed it would be a much bigger job than
initially foreseen, said "Why? It's just Fortran, isn't it?" :)

I don't know just why, but it seems I'm in "geezer mode" today, sorry...