Re: [PATCH] Fix error: array subscript has type 'char'

[Manuel Bouyer <bouyer@antioche.eu.org>]

On Thu, Jan 14, 2021 at 11:53:20AM +0100, Jan Beulich wrote:
> On 12.01.2021 19:12, Manuel Bouyer wrote:
> > From: Manuel Bouyer <bouyer@netbsd.org>
> > 
> > Use unsigned char variable, or cast to (unsigned char), for
> > tolower()/islower() and friends. Fix compiler error
> > array subscript has type 'char' [-Werror=char-subscripts]
> 
> Isn't this something that wants changing in your ctype.h instead?
> the functions (or macros), as per the C standard, ought to accept
> plain char aiui, and if they use the input as an array subscript,
> it should be their implementation suitably converting type first.

I asked for inputs from NetBSD developers familiar with this part.

Although the parameter is an int, only a subset of values is valid,
as stated in ISO C 2018 (Section 7.4 paragrah 1):
> In all cases the argument is an int, the value of which shall be
> representable as an unsigned char or shall equal the value of the
> macro EOF.  If the argument has any other value, the behavior is 
> undefined.                               


As stated by NetBSD's ctype(3) manual page, NetBSD and glibc took different
approach. NetBSD emits a compile-time warning if the input may lead to
undefined behavior. quoting the man page:
     Some implementations of libc, such as glibc as of 2018, attempt to avoid
     the worst of the undefined behavior by defining the functions to work for
     all integer inputs representable by either unsigned char or char, and
     suppress the warning.  However, this is not an excuse for avoiding
     conversion to unsigned char: if EOF coincides with any such value, as it
     does when it is -1 on platforms with signed char, programs that pass char
     will still necessarily confuse the classification and mapping of EOF with
     the classification and mapping of some non-EOF inputs.


So, although no warning is emmited on linux, it looks like to me that the
cast to unsigned char is needed anyway, and relying on glibc's behavior
is not portable.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--