Re: IBM/Absoft bug on Mac OS-X

From: Richard E Maine (nospam_at_see.signature)
Date: 03/09/05 

Date: Wed, 09 Mar 2005 09:07:48 -0800

In article <nospam-9E44E3.08115409032005@news.supernews.com>,
 Richard E Maine <nospam@see.signature> wrote:

> In article <1110268687.750677.93830@o13g2000cwo.googlegroups.com>,
> robert.corbett@sun.com wrote:

> > Here, the reference to the variable MY would be enough to cause
> > the object file test_mod.o to be extracted in most implementations.
>
> Yes, I'd think that also. But then, the OP didn't try most
> implementations, so maybe it does work ok in most of them. Or I might
> be wrong in my thinking of how implementations would probably handle
> this. Hmm. Maybe I should throw it at one or two others to see.
> Perhaps a little later....

Ok. Now is a little later. I didn't bother to do a very exhaustive
check (in particular, I didn't bother to copy the files off of my mac),
but I tried it with all 3 f90 compilers that I have installed on this
machine at the moment: nag, g95, and gfortran. All 3 worked fine with
the library (with minor changes as to whether you do or do not put a
blank after the -l on the command line, but that's a known oddity
unrelated to the current question; uncommenting the call in the source
code, as mentioned by the OP, would not have fixed a problem like that).

So it does appear to work ok in a library with "most" (3 out of 5)
compilers tried so far (all on Macs). I suppose one might debate whether
g95 and gfortran were sufficiently distinct to count as 2. 

-- 
Richard Maine                       |  Good judgment comes from experience;
email: my first.last at org.domain  |  experience comes from bad judgment.
org: nasa, domain: gov              |        -- Mark Twain