This is the mail archive of the
mailing list for the GCC project.
Re: [patch,libgfortran] Fix NIST fm906.for
- From: Tobias Schlüter <tobias dot schlueter at physik dot uni-muenchen dot de>
- To: Brooks Moses <bmoses at stanford dot edu>
- Cc: fortran at gcc dot gnu dot org, gcc-patches at gcc dot gnu dot org
- Date: Sun, 10 Jul 2005 22:19:40 +0200
- Subject: Re: [patch,libgfortran] Fix NIST fm906.for
- References: <42D17B2D.email@example.com> <42D18037.firstname.lastname@example.org>
Brooks Moses wrote:
> You're reading the wrong part of the standard (assuming that's the F95
> standard, which it appears to be); that's talking about list-directed
> _output_. On list-directed _input_ of complex values, it has this to say:
> 10.8.1: "Each of the numeric input fields may be preceeded or followed
> by blanks. The end of a record may occur between the real part and the
> comma or between the comma and the imaginary part."
> Thus, the NIST test case in question is indeed within the standard's
> constraints for a file that Fortran 95 should be able to read, and the
> problem you fix is a case of non-compliance with the standard.
> Also, given those requirements, I think it also may be worth including a
> test case that has the comma on the second line to make sure that works
> correctly as well.
Thanks for pointing this out. This means that EOR may not occur both before
and after the comma. Probably we should test this as well.