Re: bool-vector implementation in the Emacs core

[no-spam@cua.dk (Kim F. Storm)]

Ted Zlatanov <tzz@lifelogs.com> writes:

> 
> We're talking about implementations of ranges in the Gnus developer
> list right now, and our implementations of ranges is written in
> Lisp.  I can implement inversion lists in Lisp as well, but ranges
> are such an essential Gnus piece that it seems like seeking core
> support for them would be a good idea.

What kind of C-level support are you looking for?

If you implement inversion lists with lists, I don't really think
lookups or inserts will be significantly faster in C than in
byte-compiled Elisp.

I do think that it would be a good idea to add the bool-vector support
to the emacs lisp core files though, e.g. in subr.el.

What kind of API do you envision?

> 
> When dealing with vectors, we can preallocate a fixed number of slots
> at the end of the vector to allow for growth or shrinkage of the
> inversion list.  The empty slots can be contain the repetition of
> the last value, or 0 as an out-of-band value.  I think a list-based
> implementation is slightly better, though.

IMO, vectors are the wrong tool for this kind of job -- lists are both
faster and more flexible to use for inversion lists.

> I did a search on this topic, but could not find anything.  I hope my
> proposal is of interest to people other than Gnus developers (Unicode
> coders, for instance).

It's a good question -- will unicode benefit from this at the C-level? 

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.