Re: SLATEC-FN fixes for numerical edge cases

[Mike Miller]

On Thu, Feb 23, 2012 at 4:14 AM,  Michael Goffioul <michael.goffioul
at gmail.com> wrote:
> On Fri, Feb 17, 2012 at 1:12 PM, Michael Goffioul <michael.goffioul at 
> gmail.com> wrote:
>> I'd like to propose the attached patch to solve a set of errors
>> occuring in SLATEC-FN functions for edge cases input. This
>> significantly reduces the number of test failures under MSVC (from 20
>> down to 6).
>>
>> Is there any objection?
>
> No reaction so far, so I committed it:
> http://hg.savannah.gnu.org/hgweb/octave/rev/d17237256856
>
> Michael.

Hi, a bit late to the party, but... objection.

When I switched from stable to dev this morning to test a build on
RHEL5 I noticed gfortran stopping at the ISNAN call.  I saw the change
was recent and found this post.

ISNAN doesn't show up in gfortran until the 4.3 series.  I commented
out the new check and all Fortran still compiles with gfortran 4.1.2,
and even g77 3.4.6.  The rest of Octave still supports gcc 4.1.2, at
least since this fix that went into 3.6.1:
http://hg.savannah.gnu.org/hgweb/octave/rev/1ecfba4427ff

-- 
mike