On 03/24/2010 05:21 PM, Norihiro Tanaka wrote:
> However, maybe we have
> to provide instead a !MBS_SUPPORT version of mbtolower, and use it
> unconditionally?
Hmm, We mayn't also need to convert a text into the lower case for
MB_CUR_MAX == 1, but I don't make sure that.
Yes, dfa.c takes care of that right now. But actually the entire
conversion is useless now that we use upstream glibc and that one has
RE_ICASE. I prefer to leave it in place for now, but once 2.6.x
stabilizes it will go away.
> diff5 is ok too, but I made some changes there.
>
> Would you mind reviewing the patches I posted to the mailing list?
Ok. But I don't like cur_mb_len not to be modified with `const',
though it's always equal to 1 in !MBS_SUPPORT.
True.
Paolo
