Eli Zaretskii <eliz@gnu.org> writes:

>> From: Pengji Zhang <me@pengjiz.com>
>> Cc: 77336@debbugs.gnu.org
>> Date: Mon, 31 Mar 2025 11:31:36 +0800
>> 
>> Eli Zaretskii <eliz@gnu.org> writes:
>> 
>> >> Is casting the 'Lisp_Object' to a number acceptable here?
>> >
>> > No, because Lisp_Object could be a struct (if Emacs was configured with
>> > "--enable-check-lisp-object-type").
>> 
>> Sorry I did not make it clear. I meant 'make_fixnum(XLI(elt))'. I
>> suppose 'XLI' will handle that case correctly?
>
> I'd prefer not to do that unless there's no other reasonable solution.
>
>> I am actually concerned that there could be a collision between two
>> different 'elt's, e.g., a symbol and a string, because 'make_fixnum'
>> alters the tag of 'elt'. I think it might be fine because 'elt' should
>> always be a 'Lisp_String' when this is called.
>
> Sorry, I don't understand: what do you mean by "alters the tag of
> 'elt'", and how could make_fixnum affect 'elt'?

It does not affect 'elt'. I meant that 'make_fixnum' may change the
value of the number from the tagged pointer:

    INLINE Lisp_Object
    make_fixnum (EMACS_INT n)
    {
      ...
        {
          n &= INTMASK;
          n += (int0 << VALBITS);
        }
      return XIL (n);
    }

So if I understand it correctly, that means 'make_fixnum(XLI(x))' could
give the same result for, e.g., a fixnum and a string.

>> > A global variable is what I had in mind.  I don't see why it would be
>> > less clean.
>> 
>> It is just that maintaining a global state is not as tasteful to me. If
>> 'make_fixnum(XLI(elt))' is not good here, I am happy to go this way.
>
> Yes, a global variable is cleaner than using XLI for this purpose.

Thanks. Please see the attached patch.