Re: Simple algorithmic question I hope :-)

["William B. Clodius" <wclodius@lanl.gov>]

Robert Dewar wrote:
> <snip>
> 
> Well I was of course referring to the most recent version. Of
> course Fortran is really odd, both the 77 and 90 standards
> from ANSI are valid at the same time, a very peculiar situation
> reflecting an inability to get a real consensus that F90
> represented *the* direction in which Fortran development should
> go. Does someone know if the ISO standard suffers the same
> schizophrenia? Normally there can be only one ISO standard for
> a language (Ada 83 is no longer an ISO standardized language).
> <snip>

No, ISO recognized only one standard. The US appears to have been the
only country that retained both standards as applicable. I believe this
dual standard ended in 1996 or 1997, shortly before Fortran 95 became a
standard. Politically this was justified as driven by the NIST
conformance tests for Fortran 77 which was odd because

	1. Fortran 90 was to all intents and purposes a superset of Fortran 77
+ MIL-STD 1753. About the only problems the test code could have is if
the test code were not valid, or they used a name for a procedure that
was identical to one of the new intrinsics added in Fortran 90, a
conflict that is easy to fix using the EXTERNAL statement and still
leave the test code valid Fortran 77.

	2. Most compiler implementors that I have talked to, more than half a
dozen, talked about the conformance tests as if they were a joke.

	3. By 1995 NIST was talking about getting out of programming language
conformance testing activity.

In reality it was driven by the large number of Fortran 77 vendors, more
than half the total, that either could not afford to upgrade their
compilers to standard conformance or wanted to focus on other areas
(e.g. Watcom and C/C++) but still wanted to tout their compiler as
standard conforming.