>[JBDP] I don't know how many people have compilers and libraries
(including
>legacy libraries, whether home-grown or thirds party) that support
>character-style and string-style operations on unsigned shorts so I don't
>know how "few" the people are who will be inconvenienced.
It is our belief that most platforms have available to them the wcsxxx()
type APIs these days. These accept wchar_t types, which is what XMLCh is
supposed to be anyway, as you read later in my previous message.
>> The original intent when I designed the system was that XMLCh would
float
>> to wchar_t.
>[JBDP] This would obviously be a big improvement. Firstly because
wchar_t
>is distinct from unsigned short and secondly because it is much more
likely
>that the libraries that people have support string-like operations on
>wchar_t -- even if legacy code doesn't.
Don't assume, unfortunately, that wchar_t is a distinct type. Its
*supposed* to be according to the spec, but our friends MS don't support
that yet in VC++. Since that is obviously the major compiler we have to
support, we cannot do anything in the code at this time that really depends
upon this difference, because it wouldn't work on VC++ :-(
