Re: Compiling Elisp to a native code with a GCC plugin

[Stefan Monnier]

>> However, AFAIK Emacs always makes a unibyte string if all the
>> characters are pure ASCII.  So this does not matter in practice.
> That's true in Stefan's personal Emacs, AIUI, but otherwise I bet aset
> on a multibyte string can turn it into pure ASCII.

Actually, it's not true in my personal branch because I use there some
(failed-)experimental code which distinguishes between
"unibyte/multibyte/anybyte" where "anybyte" means "could be either" and
is used for purely ascii strings (or more specifically, it's used for
those strings where the unibyte and multibyte representation are the
same).
It might have been a good idea, but it seems this is not worth the
trouble changing now.  So all that's left from this experiment is that
when I see problem with multibyte/unibyte I absolutely need to try it
with the trunk code, because it may just be a bug in my branch.


        Stefan