Re: test for the MOD() intrinsic

[Richard E Maine <Richard dot Maine at nasa dot gov>]

On Nov 28, 2005, at 1:35 PM, Uttam Pawar wrote:

> According to f95, the second parameter can be 0 (zero), and if that is
> the case then the result is processor dependent. While, f2003 says, the
> second parameter can not be zero.

Intriguing. And puzzling. A quick glance verifies that you are correct 
about the difference, but I find this surprising in that it looks to me 
like an incompatibility - that is valid f95 code is not necessarily 
valid f2003 code. I would have expected an incompatibility like that to 
be listed in 1.6.1 of f2003 as an f95 incompatibility. Makes me wonder 
whether perhaps f95 was changed by an interp.

Ah... Additional data. Though this isn't listed in 1.6.1, it *IS* 
listed in 1.6.2 as an f90 incompatibility. This implies to me that f95 
and f2003 are the same, but f90 (and earlier) were different. I take 
that as additional evidence that this must have been changed in an f95 
interp. Hmm. I've got to run pretty quickly, but lets see if I can 
track down the interp. That would have to be one with edits, which 
would mean it would show up in a corrigendum, which makes it a lot 
simpler to find as there are only 2 or 3 f95 corrigenda...

Yep, there it is in f95 corrigendum 2, resulting from f95 interp 87. 
And yes, that same corrigendum also adds it as an f95/f90 
incompatibility.

So my conclusion is that if you want to support both variants depending 
on the -std= setting, f95 and f2003 are the same, with f90 being 
different.

--
Richard Maine                |  Good judgment comes from experience;
Richard.Maine@nasa.gov       |  experience comes from bad judgment.
                            |        -- Mark Twain