https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89696
kargl at gcc dot gnu.org changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |kargl at gcc dot gnu.org
--- Comment #4 from kargl at gcc dot gnu.org ---
(In reply to Khang H. Nguyen from comment #3)
> Sorry, I just have one more quick question.
>
> Bug 1:
> Nonetheless, for read(), I was just wondering, if you read from a list input
> then it should be like that.
>
> However, if it is just a raw string and it acted like that, then how do you
> tell if a string is an invalid integer string?
Formatted input will tell you that you have a problem.
% cat a.f90
program foo
integer i
read(*,*) i
read(*,'(I4)') i
end program foo
% gfcx -o z a.f90
% ./z
% ./z
12 b
12 b
At line 4 of file a.f90 (unit = 5, file = 'stdin')
Fortran runtime error: Bad value during integer read
The first '12 b' is read a '12'. The second read says the 4 characters
are an integer, so '12 b' generates an error.
> Do you have to build a check for that then? Shouldn't read() be able to
> detect that because if you build a check on top of that, you would have to
> go through all the character one and then read would then goes through all
> the characters again.
>
> Bug 2:
> As you said that it is not a bug. I just suggested that for -2147483648, it
> is still assignable with range check on, you just have to assign -2147483647
> - 1
Yes, you can assign -2147483647 - 1. As Thomas suggested, the first
minus sign is a unary operator and the second minus sign is a binary
operator. So, there are two operators in the expression.
