Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes: > | You mean wchar_t == uint32? UCS-4 is the encoding, no? > > yes. > (but wchar_t on linux also follow the ucs-4 encoding...)
???? There's no encoding in an int. You mean that wchar_t on linux will hold 4bytes I think whereas on Windows wchar_t will hold only 2bytes and therefore cannot hold all values that can be stored in a UCS-4 encoded document. The 2 byte wchar_t is sufficient to hold all values that are present in the Base Multilingual Plane. Values outside the BMP would need an encoding other than UCS-4 (like the multi-atom UTF-16 encoding) if the data were stored in a 2byte atom. At least, I *think* that's how it all fits together, but unicode tends to confuse me ;-) > | typedef std::basic_string<lyx::char_type, lyx::uchar_traits> lyx::ustring; > | What do I miss? > Perhaps nothing. So why do you prefer std::vector<lyx::char_type> as the container? You're a rationale bloke, so there must be a reason? Trying to get a feel for all this, Angus
