bit sets and their uses

[Jose E. Marchesi]

Hello Chris.

> Hi everyone,
>
> van Vliet defines a mode CharBag in his transput routines as:
>
> mode {ℵ₀} CharBag =
>   PSEUDO56 { Some mode which allows efficient retrieval of
>     information as to the presence or absence of a given character in a
>     given set (for example, a bit string). };
>
> He goes on to use CharBag to define sets of characters that are interesting
> for formatting purposes as well as supporting terminator characters.
>
> This probably seemed pretty simple to implement at the time where "all char
> values fit in 8 bits".  But with Unicode...
>
> So I did some looking around and came upon this:
>
> https://roaringbitmap.org
>
> which seems quite interesting, at least from the perspective of handling
> bit sets efficiently.
>
> I'm wondering if anyone else has looked at this problem and found a
> preferred solution?

If it was me I would just use a string for CharBag at this point (making
STRINGTOBAG pretty trivial) then worry about performance later..

Unicode has _a lot_ of codepoints, so a sophisticated implementation of
sets like the one you mention is probably what we need.