bug#13399: 24.3.50; Word-wrap can't wrap at zero-width space U-200B

[Eli Zaretskii]

> Date: Sun, 03 Feb 2013 19:57:31 +0100
> From: martin rudalics <rudalics@gmx.at>
> CC: 13399@debbugs.gnu.org
> 
> I now rewrote IT_DISPLAYING_WHITESPACE as
> 
> #define IT_DISPLAYING_WHITESPACE(it)                                  \
>    ((it->what == IT_CHARACTER                                         \
>      && !NILP (CHAR_TABLE_REF (Vword_wrap_chars, it->c)))             \
>     || ((STRINGP (it->string)                                         \
>       && !NILP (CHAR_TABLE_REF                                        \
>                  (Vword_wrap_chars,                                   \
>                     SREF (it->string, IT_STRING_BYTEPOS (*it)))))     \
>         || (it->s && !NILP (CHAR_TABLE_REF                            \
>                           (Vword_wrap_chars,                          \
>                              it->s[IT_BYTEPOS (*it)])))               \
>         || (IT_BYTEPOS (*it) < ZV_BYTE                                        
> \
>          && !NILP (CHAR_TABLE_REF                                     \
>                     (Vword_wrap_chars,                                \
>                        (*BYTE_POS_ADDR (IT_BYTEPOS (*it))))))))       \
> 
> 
> and have a character table called `word-wrap-chars' such that
> (aref word-wrap-chars ?​) returns t, but it doesn't wrap at a
> U-200B character.  Is there some additional wrinkle like some
> hardcoded space/tab in the word-wrap code I have to observe?
> Or is my code wrong?

Does CHAR_TABLE_REF return the right value?

Also, the code is wrong here:

   SREF (it->string, IT_STRING_BYTEPOS (*it))

and here:

   it->s[IT_BYTEPOS (*it)]

and here:

   *BYTE_POS_ADDR (IT_BYTEPOS (*it))

in that that it assumes the character is always one byte, which is
clearly wrong with non-ASCII characters.  You should instead use
FETCH_CHAR and STRING_CHAR.