Re: master 3ed79cd: Separate bytecode stack

[Eli Zaretskii <eliz@gnu.org>]

> From: Mattias Engdegård <mattiase@acm.org>
> Date: Sun, 13 Mar 2022 19:44:22 +0100
> Cc: Emacs-Devel <emacs-devel@gnu.org>
> 
> 13 mars 2022 kl. 18.39 skrev Eli Zaretskii <eliz@gnu.org>:
> > 
> > This changeset causes compilation warning in a 32-bit build in which
> > !LISP_WORDS_ARE_POINTERS and which was configured --with-wide-int:
> > 
> >  In file included from bytecode.c:22:
> >  bytecode.c: In function 'sf_set_ptr':
> >  bytecode.c:396:20: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
> >    396 |   fp[index] = XIL ((EMACS_INT)value);
> 
> Thank you for testing this configuration! Silly mistake of mine; please try with the pushed change.

The warning is gone, but I cannot say I'm happy with the code, see
below.

> > More generally, I'm quite nervous to see void * pointers and integers
> > being put into the same array.
> 
> Actually what we are doing is storing arbitrary pointers in a Lisp_Object, and this should always be possible.

I don't think I understand why you need to do this in the first place.
XLP was introduced for very specific and very rare situations, and I
don't see why we would need in this case to use something similar.
What am I missing?  What kind of pointers do you need to store in the
fp array, why, and for what purpose?  And if you do need to do that,
why not use a union?