Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes: > | So why do you prefer std::vector<lyx::char_type> as the container? > | You're a rationale bloke, so there must be a reason?
> I just picked something that would work. After all we are not going to > use (any?) string operations on such strings. Ok, *that's* the info I was looking for. Thanks. Even if I'm not sure I follow your "After all..." comment. What about our (limited) search and replace functions? AFAICS, here's the list of std::string member functions that don't exist in std::vector: operator+ operator+= append compare copy data find find_first_not_of find_first_of find_last_not_of find_last_of replace rfind substr We use most of them, no? However, I don't suppose any of them are particularly hard to implement as free standing tools. The Boost.StringAlgo library probably covers them all already. FWIW, I seem to remember that CJK-LyX used std::wstring and the Qt version at least didn't need much alteration to the standard LyX at all. That would suggest that, whatever container you end up using, this project looks entirely feasible. Angus
