Re: UNICODE

[Hans Aberg]

At 15:46 +0300 2002/02/12, pecherin wrote:
>Is it possible to use Bison with UNICODE ?

Formally, the Bison generated parser reads token numbers provided by the
lexer and knows nothing about character codes, so from that point of view,
there is nothing different form using one character encoding from another.

Note however that Bison reserves token numbers 1..255 for use with the
character codes that in the Bison .y file are enclosed by '...', like when
having a Flex rule
  . { return (unsigned char)yytext[0]; }
I figure that a future Bison Unicode option might extend this to say the
range 1..2^24-1 and make such characters available from within the .y file
(with the ability to have this file in Unicode as well), but there is
currently no way to access such characters.

But if you avoid that, you should have no problems with Bison I think,
because there is nothing else that depends on character codes (except error
and debugging output which assumes that char based streams are used).

  Hans Aberg