Re: gawk: length return incorrect value when MB_CUR_MAX > 1

[Paul Eggert]

Aharon Robbins <address@hidden> writes:

>> I found mbrtowc() store L'\0' to &wc when detect '\0'.
>
> Is this portable behavior? I.e., defined by the ISO C standard?

Yes.  You can find this in
<http://www.opengroup.org/onlinepubs/009695399/functions/mbtowc.html>
The wording there differs from the C standard, but it boils down to
the same thing once you remove the mentions of POSIX notions like
LC_ALL.

The first paragraph of the DESCRIPTION section says that the value of
the wide-character code corresponding to the null byte is 0, and that
mbrtowc must store the wide-character code in the object pointed to by
the first argument.

I don't know of any mbrtowc implementations that disagree with the
standard here.