"char" is signed by default on many systems, so when using a "char"
variable as input to isspace(), which takes an "int" parameter, the
parameter gets sign-extended. Now in some libc implementations, the
isspace() function is implemented as a macro that directly indexes
an array for looking up the result - which might go wrong, of course,
if the byte values has the highest bit set. So when compiling Hatari
with Cygwin, there are the following compiler warnings:
.../build68k.c: In function ‘main’:
.../build68k.c:314:18: warning: array subscript has type ‘char’ [-Wchar-subscripts]
314 | while (isspace(*opstrp))
| ^~~~~~~
.../build68k.c:319:18: warning: array subscript has type ‘char’ [-Wchar-subscripts]
319 | if (!isspace (*osendp))
| ^~~~~~~
.../build68k.c:331:19: warning: array subscript has type ‘char’ [-Wchar-subscripts]
331 | while (!isspace(*p++));
| ^~~~
Thus let's make sure to cast the "char" to unsigned first before using
it as parameter to the isspace() function.
char tmp[100], *p;
int slen = 0;
- while (isspace(*opstrp))
+ while (isspace((unsigned char)*opstrp))
opstrp++;
osendp = opstrp;
while (*osendp) {
- if (!isspace (*osendp))
+ if (!isspace ((unsigned char)*osendp))
slen = osendp - opstrp + 1;
osendp++;
}
strcpy (tmp, opstrp);
strcat (tmp, " ");
p = tmp;
- while (!isspace(*p++));
+ while (!isspace((unsigned char)*p++));
*p = 0;
printf("/* %s */\n", tmp);
printf("{0x%04X,%2d,{", bitpattern, n_variable);