> I can see how that might be useful. The end user can already get this > behavior by adding about 9+(the identifier length) characters to their > program. Since Length() is inline, I suppose it would be just about as > efficient for the user to do the computation as for BString to do it. I > would favor moving a change such as this to R2, where we will also > hopefully have a string class that properly handles UTF8. :-) Uhm, what's all that fuss around BString not handling UTF8 ? You can store UTF8 strings there, just don't use any functions but CountChars(), or String(), and you're done :)) Seriously.... Usually APIs have two string types, one for single byte characters and one for multi-byte characters. It even makes sense, since a full UTF8 string class would have serious performance issues if compared to a single byte one.