Re: inputting characters by hexadigit

[Stefan Monnier]

> I can't believe this. So now we're going to muck with existing
> completion mechanisms even more, trying out new features, just for
> this Unicode character input thing?

Looks like a sensitive subject, huh?

> How about finishing what you were doing to move the completion code to
> Lisp,

It's finished for now.

> and adding the `completing-read-function' variable we've been
> waiting for?

Talking like a king?  I'm not opposed to such a change, but I have other
things to do.  So if someone else wants to install such a thing, he
should go right ahead.  In the mean time, defadvice works well for
such things.

> The so-called translation to Lisp has already resulted in adding new
> "features" with no discussion here.

The purpose of the translation was to then facilitate the introduction
of new features.  So, yes, there are new features, of course.  Indeed,
I have no discussed them but instead imposed them by fiat, waiting for
the reactions.  Surprisingly enough, I basically haven't heard any
complaint about the new behavior(s), even though it enables
the equivalent of partial-completion-mode by default.

> The addition of `completing-read-function', unlike those changes, will
> not require users to change their code in any way.

The new completion code is about as compatible to the old as I could
make it.  I can't remember the problems it introduced, but IIRC they
were all unrelated to the translation itself but instead related to
other bugfixes that were applied at the same time.

BTW, I regret having forgotten about the incompatibilities.  It would be
good to keep track of them somewhere, to help other package developpers.


        Stefan