Well, by strings I meant 8-bit/16-bit chars (std::string/wstring). One "problem" with these string patterns is that they let the engine do more than mere comparison -- they let the engine actually enumerate all the characters in between. If I remember correctly, I've seen linear-ish-time regex engines (i.e.: I'm not talking about the backtracking ones you see in most languages, although they might sometimes do this too) take advantage of this by making lookup tables for the characters in an interval, and this clearly doesn't scale when you go to 32-bit or 64-bit "characters". They really have to implement efficient set intersection for large sets, which is possible, but I'm not I've seen them do this.
If you're looking for specific examples of "large alphabets", there's a very easy one in my mind, which is UTF-32. If you're looking for data that isn't a plain sequence of integers in any shape or form... I have to think back, as I don't remember what the use case I ran into was, sadly. I'll post here if I remember.
If you're looking for specific examples of "large alphabets", there's a very easy one in my mind, which is UTF-32. If you're looking for data that isn't a plain sequence of integers in any shape or form... I have to think back, as I don't remember what the use case I ran into was, sadly. I'll post here if I remember.