Jerry DeLisle wrote:
> Regression tested on x86-64-linux-gnu.
> OK for trunk?
OK. Thanks for the patch.
+ && mpz_sgn (fp->unit->value.integer) < 0)
+ gfc_error ("UNIT number in CLOSE statement at %L must be positive",
+ gfc_error ("UNIT number in statement at %L must be positive",
I'm wondering whether is should be "must be nonnegative"? The reason is
that 0 is not uncommon for stderr. The check is OK, however.
"Nonnegative" also matches the Fortran standard: "a file-unit-number
whose value is nonnegative or equal to one of the named constants INPUT
UNIT, OUTPUT UNIT, or ERROR UNIT of the ISO FORTRAN ENV module
(13.8.2)." (F2003, 9.4). Fortran 2008 just added (in 9.5.1): "...or a