https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93727
--- Comment #30 from Jerry DeLisle <jvdelisle at gcc dot gnu.org> ---
A minor adjustment:
diff --git a/libgfortran/io/write.c b/libgfortran/io/write.c
index 1bad5c49531..af75738c436 100644
--- a/libgfortran/io/write.c
+++ b/libgfortran/io/write.c
@@ -1976,7 +1976,7 @@ write_ex (st_parameter_dt *dtp, const fnode *f, const
char *p, int kind)
/* Format the exponent field with specified width 'e'. The 'e' parameter
is the total exponent width INCLUDING 'P' and the sign. */
- int exp_digits = e - 2; /* Subtract 'P' and sign character */
+ int exp_digits = e;
if (exp_digits < 1)
exp_digits = 1; /* Minimum 1 digit */
I was thinking e is a width. No, it is the number of exponent digits requested
by the user.
I have the two test cases in the patch corrected now.
I need to review the standard regarding rounding of hex digits if a user
specifies a width too narrow.
