[comp.sys.sun] Sun f77 v 1.3.1: MAJOR Bug???

rhoward@msd.gatech.edu (Robert L. Howard) (08/17/90)

I just installed Sun FORTRAN v 1.3.1 on my Sparcstation (SunOS4.1) and
began reading the notes when I discovered:

Complex arithmetic with -fast fails.
	If you specify inlining and a floating point option other than
	-fsoft, then complex arithetic fails.
	Example 1:  f77 -fast any.f
	Example 2:  f77 -cg89 -libmil any.f

Well this is not a minor bug.  I have a couple of questions though...

1.	On my SS1, I can't invoke -fsoft as an option. It doesn't
	recognise it.  f77 -help does not list it.  Is this the
	default??

2.	Does anyone have some code that produces any errors (specifically
	in Ex. 2 as that is how I compile my code)?  I have not noticed
	any gross errors in the output, but I wonder about the subtle
	ones that might bite me later.

This really stinks!  If you compile without -libmil everything takes
*twice* as long to run.

I really need some confirmation about the existence of this bug....

Thanks all,
Robert

| Robert L. Howard             |    Georgia Tech Research Institute     |
| rhoward@msd.gatech.edu       |    MATD Laboratory                     |
| (404) 528-7165               |    Atlanta, Georgia  30332             |
|     UUCP:   ...!{allegra,amd,hplabs,ut-ngp}!gatech!msd!rhoward        |

atul@decwrl.dec.com (atul) (10/08/90)

I reported this bug to Sun Hotline. Here is the reply I got from them.

***

X-Subject: RE: SO#542715 - problems with complex in Fortran

I am following up on a call that you were working on with Eric.  Eric has
asked me to look into this problem.  You are right, there is a bug here in
the inline math libraries.

I have filed a bug against the Fortran compiler(bug#1043400) and we will
get it fixed in a future release.

Thank you for the information, this really does help us in fixing up our
products.