Conference turris::languages

A company can offer an unvalidated Ada compiler.  It simply isn't useful
for bidding on contracts that require a validated one.  However, to use
the name Ada, they must be intending to validate in the not-to-distant
future.

There is also a validation process for Fortran, and I think there will be
one for Pascal.

/Bevin

Bevin:

	Who is going to validate the Fortran compiler? What about the Pascal
compiler? Is the documentation included in the validation?

Jack

There is an ANSI standard for something called Minimal BASIC (BTW, you
might want to post this question in the BASIC notes file -- their information
is probably much more current from mine).  That standard (mine is dated 1978)
says in section 1.4.2 "Conformance by an Implementation" paragraph (7):

  It is accompanied by documentation that describes the actions taken in
  regard to features referred to in this standard as "undefined" or 
  "implementation-defined."

Since any salable implementation of BASIC is a superset of Minimal BASIC,
I would guess that most of the documentation would have to pass the test
of the Federal Software Testing Center.  The only people who care about
this stuff (in the context of BASIC) are the Feds -- in order to sell a
BASIC product to a Federal agency, the product must be standard-conforming,
even though most of the product will be "implementation-defined," that is,
not covered by the Minimal standard.

JP

There is no ANSI PL/I validation suite.   Until recently there was only
one implementation that even claimed to be a full language implementation.
There is an ANSI defined PL/I subset (called G subset for general purpose)
and that is what most implementations try to conform to.   There is no
validation test for subset implementations either.
                   
	db

There is no threat of lawsuit for people who sell compilers for "Fortran"
which are not validated (unlike the people who use the name Ada).  You just
can't foist it on people who require a validated compiler (like the government).

Of course, you better not lie about it being validated when it isn't. . .
				/AHM

What's the Federal Software Testing Center? What do they do, and are
they limited to testing products the Government buys?
			---Walter

The new PASCAL validation offered by the federal government is the same
as the ISO validation offered by the BSI (British Standards Institute).

One of the requirements for validation is that the compiler must detect 
certain errors, or explain in the compiler documentation which errors 
it doesn't detect.

Maybe this is what your wife is running into on the BASIC compiler?

			Bb

And COBOL implementations may be validated by the fed, too.

RE: .6

The Federal Software Testing Center (FSTC) is the group responsible for
validating FORTRAN and (I think) COBOL.  A Federal Information Processing
Standard (FIPS) exists for both languages.  U. S. Government departments
are required to use FIPS-compliant language implementations.  FSTC is the
group that runs test suites to determine whether a particular implementation
is compliant.  VAX FORTRAN (both VMS and Ultrix) and PDP-11 FORTRAN-77 have
been validated by the FSTC.

--PSW

Re .6,.9:

And Fortran-10/20 has also been validated (at the "Full" langauge level).

I believe that the "FIPS" Fortran requirements are that the compiler
be ANSI standard, and that it have the capability to flag non-standard
conforming constructs.

There is also a document called MIL-STD-1753, which calls for a bunch
of extensions to the standard like DO WHILE (), END DO.  I think that
Fortran-10/20 contains all the features, and Vax-11 Fortran contains
virtually all of them.  Many people don't realize that some of the DEC
extensions come from this document.
				/AHM