Absoft vs. NKR Fortran... the latest

[Jim Jagielski]

As you may recall, I few days ago I posted a comparison of NKR Research's
FORTRAN and Absoft's MacFortranII for A/UX. After working with MacFortranII
for a little bit longer, I really feel the need to repost.

As mentioned before, compiling with Absoft takes longer that using NKR...
however my statement of 2x longer was conservative. It usually takes a LOT
longer (3-5 in some cases!!). It also takes a lot of memory and disk
space (in /tmp). In many cases, this time is simply unacceptable, and I'm
running on a 50MHz Mac! I have also noticed some strange things when reopening
connect units, for example unit 6, stdout. 1st of all, when you reOPEN, Mac-
FortranII closes your previous connection, which it shouldn't do. My point
can best be served by a simple example.

Vax FORTRAN uses Carriage Control code by default. Under NKR FORTRAN, it's
not the default, but all you have to do is add the CARRIAGECONTROL="FORTRAN"
option in OPEN and all is well. When you do this to UNIT 6, it enables
carriage control on your terminal. However, Absoft, due to it closing the
preconnect, makes you ADD FILE="/dev/tty" which is totally non-needed if it
did what it was supposed to do. It also uses a different option called ACTION
instead of CARRIAGECONTROL (and it's action='print' not carriagecontrol=
'fortran')... since the carriage control properties of VAX FORTRAN are VERY
well known, you would think that they would at least use the "normal" option
name.

I do enjoy having all the VAX library functions available as well as complete
sdb support, but these aren't worth the excess time as picky little idiosyncra-
sies of MacFortranII... and I was NEVER able to get it to perform any more
that 2% faster than NKR compiled code!

So I'm sticking with NKR... if Absoft straightens these "features" out, maybe
I'll try it again (I hope they provide a free upgrade, considering the input
and suggestions I've provided via their BBS), but as for now, I'll only use it
when I need to sdb some code... but even then I may not, since the latest
version/update of NKR FORTRAN promises better sdb support...

I'll keep you posted!

--
=======================================================================
#include <std/disclaimer.h>
                                 =:^)
           Jim Jagielski                    NASA/GSFC, Code 711.1
     jim at jagubox.gsfc.nasa.gov               Greenbelt, MD 20771

"Kilimanjaro is a pretty tricky climb. Most of it's up, until you reach
 the very, very top, and then it tends to slope away rather sharply."