https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99950
Bug ID: 99950
Summary: Option -Wchar-subscripts leads to wrong fixes
Product: gcc
Version: 11.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: roland.illig at gmx dot de
Target Milestone: ---
When code calls the functions from <ctype.h> with an expression of type 'char',
GCC and Clang warn:
> array subscript has type char
Since the function prototypes expect an 'int', inexperienced programmers cast
the argument to 'int', which silences the warning:
https://github.com/search?q=%22isspace+int%22&type=code
More experienced programmers who heard something about 'unsigned' cast the
argument to 'unsigned int', which silences the warning as well:
https://github.com/search?q=%22isspace+unsigned+int%22&type=code
To get these programmers to fix their code properly, GCC should provide more
guidance in this area. As a quick prototype I implemented a check for these
wrong "fixes" in NetBSD's lint, which results in warnings like these:
warning: argument to 'function from <ctype.h>' must be cast to 'unsigned char',
not to 'int' [342]
The implementation is here, including a test:
https://github.com/NetBSD/src/blob/fdd2c09ca474/usr.bin/xlint/lint1/ckctype.c
https://github.com/NetBSD/src/blob/fdd2c09ca474/tests/usr.bin/xlint/lint1/msg_341.c
https://github.com/NetBSD/src/blob/fdd2c09ca474/tests/usr.bin/xlint/lint1/msg_341.exp
Maybe GCC can implement a similar check to prevent these wrong fixes in the
future.
