Re: inputting characters by hexadigit

[Stefan Monnier]

> The categories are what you get when you split all the candidate strings
> based on the word syntax.  partial-completion-mode does the same thing,
> the difference is that it expects you to know that you can use z-x-y
> because there's a zeta-xray-yogurt variable.  My proposal tries to
> display just the categories to condense the list of candidates, refining
> and digging deeper into the categories whenever possible until there's a
> small enough list of candidates.

So, it's like a substring search (like what you get with * A P L TAB),
except that the substring is anchored at the beginning of words, and
that *Completions* only lists the words rather than actual completion
candidates (on the assumption that there is a fair bit of redundancy in
the list of words).  It sounds interesting.

The "anchored substring search" should be fairly easy to implement as
a new completion-style.  The *Completions* change would require less
"modular/configurable" changes, I suspect, but it's worth trying it out.


        Stefan