Fortran Versions (was JB Passes)

[Chuck Guzis]

On 27 Mar 2007 at 12:59, woodelf wrote:

> I rather liked full blown FORTRAN IV. I have been looking at
> the PDP8 Fortran IV version and all things considered that was
> a powerful machine for its price range in the 1970's.
> All this talk of Fortran 77, 90 make fortran a more modern
> langage but it seems to me they lost something in that the old
> programs have to be re-written since they don't support things
> like sence switches for example.

The problem with "full blown" FORTRAN IV was the same as "full blown" 
FORTRAN II and "full blown" FORTRAN--everyone had extensions--and 
what the selfsame extensions did could vary wildly between vendors.  
Some used the logical operators as bitwise booleans; e.g.,  C = 
A.OR.B; there were optional arguments in I/O statement syntax,  
additional FORMAT specifiers, etc. FWIW, I think that IBM FORTRAN IV 
supported sense switches only through function calls.  CDC 6000-
series machines didn't have sense switches (or blinkinlights) --
rather a location in  low memory within the user's RA could be set 
via operator command.

I seem to recall (this was a LONG time ago) that 7090 FORTRAN had 
extensions that keyed off of what was punched in column 1.  "D" meant 
that the operators were double precision; "B" meant boolean, "I" 
meant complex (IIRC).

The big problem with permissive-by-default vendor-unique extensions 
is that it's just about impossible to keep programmers from using 
them--and so make migration to a different vendor's platform more 
difficult than it needs to be.

You haven't lived until you've worked on an RFP response that 
addresses how you're going to handle another vendor's extension to a 
supposedly "standard" language.  

Didn't COBOL 75 mandate that the "standard" language contain no 
vendor extensions unless specifically enabled?  Again, it's been a 
long time, but I seem to recall a something in that spirit. (COBOL 
was even worse than FORTRAN when it came to vendor-unique features).

Cheers,
Chuck