signed vs unsigned char in coding.h

signed vs unsigned char in coding.h

[Dan Nicolaescu]

struct coding_system in coding.h contains this:

      char *safe_charsets;


and coding.c has this:

#define SAFE_CHARSET_P(coding, id)      \
  ((id) <= (coding)->max_charset_id     \
   && (coding)->safe_charsets[id] >= 0)

on some platforms "char" is unsigned by default, so the above >=0
comparison is always true.

What is the intention here, should safe_charsets be defined as "signed char" ? 

Re: signed vs unsigned char in coding.h

[Chetan Pandya]

--- On Fri, 2/20/09, Dan Nicolaescu <dann@ics.uci.edu> wrote:
> struct coding_system in coding.h contains this:
> 
>       char *safe_charsets;
> 
> 
> and coding.c has this:
> 
> #define SAFE_CHARSET_P(coding, id)      \
>   ((id) <= (coding)->max_charset_id     \
>    && (coding)->safe_charsets[id] >= 0)
> 
> on some platforms "char" is unsigned by default,
> so the above >=0
> comparison is always true.
> 
> What is the intention here, should safe_charsets be defined
> as "signed char" ?

It looks like there are more things needed here. Since the safe_charsets is part of a string, it makes sense to declare it as unsigned char and compare it to 255.

However, when in coding.c:9298 this is initialized with 
  safe_charsets = Fmake_string (make_number (max_charset_id + 1),
				make_number (255));
This also has problems since it tries to create a string with a character that is not ASCII_CHAR_P. Wouldn't it make sense to create with ascii char and then replace all with 255 instead?

Chetan