"Tobias C. Rittweiler" <tcr@...> writes:
> It would make more sense to me if the Xref functions return NIL for
> _symbols_ which do not designate the things in the function's
> domain. And they should signal a type-error for everything else
> (non-symbol atoms, conses.)
It's not clear to me that non-symbol atoms and conses are always out of
the domain of introspection functions. I can easily imagine function or
macro names as conses, for example, and it's only a tiny leap of the
imagination to contemplate who-calls functionality on
funcallable-instances, and only another small leap to contemplate
funcallable arrays or funcallable numbers. Of course, existing
sb-introspect functions may not work with such exotica, but...
> Is it ok for you to change the xref functions accordingly?
I would prefer it if the new sb-introspect functions were made
consistent with the existing ones for now; that is, to return nil on
unexpected input, but I'm not too stressed if the contrib "owners"
(where we define that as "anyone who actually does the work") has a