Current state of multibyte character support

Current state of multibyte character support

[Matt Hiller]

Configuring gcc with --c-enable-mbchar allows gcc to process input files
where strings, comments, and #include filenames are encoded with S-JIS,
JIS and EUCJP instead of plain ASCII.

According to a thread I found on gcc@gcc.gnu.org from December 1998
(http://gcc.gnu.org/ml/gcc/1998-12/msg00171.html), shift-jis can cause
compilers trouble in that it uses '\' to encode Japanese characters. The
work that I've done recently leads me to say that problems like this are
cropping up again, especially in cpplex.c.

The most recent work on cpplex.c seems to be Neil Booth's. If I read 
aright (http://gcc.gnu.org/ml/gcc/2000-09/msg00268.html), Neil tried 
to do this work such that multibyte support could eventually be 
added. Anyone know how successful that ultimately was, or what other 
issues may present themselves?

Thanks much,
Matt

p.s.: I found bugs in c-lex.c:lex_string, but I believe I have them worked
out.

Re: Current state of multibyte character support

[Zack Weinberg]

On Sun, Apr 07, 2002 at 09:47:38PM -0700, Matt Hiller wrote:
> 
> The most recent work on cpplex.c seems to be Neil Booth's. If I read 
> aright (http://gcc.gnu.org/ml/gcc/2000-09/msg00268.html), Neil tried 
> to do this work such that multibyte support could eventually be 
> added. Anyone know how successful that ultimately was, or what other 
> issues may present themselves?

We are working on it.  There is a chance of full multibyte support in
3.2, but don't hold your breath.

> p.s.: I found bugs in c-lex.c:lex_string, but I believe I have them worked
> out.

We're aware of the bugs.  Most of that code has to get torn out and
replaced.

zw