Re: Fortran Namelist Package

[Dan Nagle <dnagle@erols.com>]

Hello,

"Howard W. LUDWIG" wrote:
> 
<snip original question>

> Was NAMELIST standardized in Fortran 90?  I'm not aware that it was.

It was, and the definition was further enhanced in Fortran 95.
(Comments in namelist files were added.)

> I know that many compilers provided NAMELIST as extensions to
> FORTRAN 66 and FORTRAN 77; however, as extensions and not part of
> the standard, the implementations among vendors varied quite a bit.
> (I believe DOD had NAMELIST in some MIL-STD defining Fortran for
> DOD purposes, but it did not match any ISO/IEC/ANSI standard--in fact,
> I suspect that DODs failure to get NAMELIST and ENCODE/DECODE approved
> as part of FORTRAN 77 was part of the driving force for DOD having a
> language they could exercise more control over--and they ended up
> with Ada--not too shabby.)

Well,

ENCODE/DECODE were replaced by internal read/write when Hollerith
was replaced by the character type (with f77).

Mil Std 1753 was universally implemented if not adopted by ANSI.

It is because, in part at least, the "details" of extensions often
differ from compiler to compiler that "industry practice" isn't
simply copied into the Fortran standard.  It's often the semantics
rather than the syntax that differs, and often only "slightly".

> 
> In other words, you need to be careful about using such NAMELISTs in
> Fortran because of the variety of syntax used by vendors.  Even if you
> do find existing Ada code (and I am not aware of any--have you checked
> the PAL?), it may not do what you need because it supports the "wrong"
> syntax.

If the Ada code is newer than f95, it's probably at least trying to
support
the right syntax; if it's newer than f90, it may not support comments in
the namelist file.  If it's older than f90, it's probably trying to
support
Mil Std 1753.  Of course, you're right to test it first ;-)

Any Fortran 90/95 compiler will have a "new namelist/old namelist"
switch if the vendor supplied a namelist which differed from what was
standardized.

> 
> Howard W. LUDWIG

-- 

Cheers!

Dan Nagle
Purple Sage Computing Solutions, Inc.