Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!swrinde!ucsd!ucsdhub!hp-sdd!apollo!ross
From: ross@apollo.HP.COM (Mike Ross)
Newsgroups: comp.lang.fortran
Subject: FORTRAN 90 survey
Message-ID: <490e8d3b.20b6d@apollo.HP.COM>
Date: 7 Mar 90 20:04:00 GMT
Sender: root@apollo.HP.COM
Lines: 103

A few weeks ago, I posted 3 survey questions
concerning the use of ANSI FORTRAN 90. 

I got 13 responses, which is hardly
enough to make a representative sample,
but since there seemed to be interest
in the results, here they are.

First, several people corrected my
use of the term FORTRAN 8x, since the
ANSI committee has recently renamed it
FORTRAN 90. I appreciate the information,
butI already knew that, but used the more 
common FORTRAN 8x term. Just figured more people 
would understand what I was asking for. After all, 
how many people call the current standard 
ANSI X3.9-1978 FORTRAN 77, its official name? 
(rhetorical question)

1) Assuming a reasonable implementation of FORTRAN 8x
were available on your machine/OS, would you be likely
to use it, say, in preference to your present FORTRAN-77
compiler?
         12 Yes, 1 no, unless compiler was cheap or free

2) Which of the new features in FORTRAN 8x do you feel
you are likely to make the most use of? Which features
are most important to your work? *

MODULES     - 7         Dynamic memory allocation   - 5

User defined
types/structures - 7    IMPLICIT NONE - 1

Bit processes - 1       Array syntax - 9

Free format - 3         Long names/underscores - 3

DO..END DO/DO WHILE - 3 Subroutine argument interface,
                         including argument checking   - 3

Pointers - 3            Stream/non-advancing I/O - 4

Ability to pass
arbitrary sized arrays - 1  

Position independent actual arguments - 1 

Ability to specify precision portably - 1

Recursion - 2

array and structure valued functions - 1

User defined operators, overloading - 2

"Ratforish stuff(e.g. CASE, etc) "  - 1

*no representation made here of the accuracy or characterization
of these features -- I'm just reporting what was said.

3) If a translation tool were available to convert your
FORTRAN programs from their present state ( presumably
FORTRAN-77 with extensions) to FORTRAN 8x, would you be
interested? Would you use it? How much would you be
willing to pay for such a tool?

Yes - 5    No - 8

Of those who said they would use it, 3 said
they would only be willing to do so if it were
cheap ( such as via educational discount ) or
free.

There were several answers to the effect of
" Why do I need a translator, since FORTRAN-77 
is a subset of FORTRAN-90, and my present code
will just compile?"

People had different ideas as to what I meant
when I said a translator, but these ideas were
interesting in and of themselves. *My* concept
of a translator was something that would go through
and convert really musty constructs like Hollerith
and ENCODE DECODE to something a bit more modern,
and that would also find things like ASSIGNED GOTO,
& make it into an internal subroutine, as well as
finding DO loops that could readily be converted to
the FORTRAN 90 array syntax and performing that function.
A couple of people voiced concern over the ability of
a tool to perform such functions reliably, and produce
readable output. 

The consensus appears to be that there isn't much
need/use for such a tool. Interesting.

There must, of course, be more regular FORTRAN
users out there on USENET than just 13. If you
didn't respond to the survey, and would still
like to, I'm still interested in hearing your
opinions.
                     Mike Ross
                     HP/Apollo Languages