Re: [External] : Re: [NonGNU ELPA] New package: Denote-Refs

[Philip Kaludercic]

Drew Adams <drew.adams@oracle.com> writes:

>> > Yes, () would do as well - in fact better, since it
>> > doesn't suggest any possibility that '() represents
>> > something different from ().
>> >
>> > You don't quote nil or t, do you?  Or 42 or "foo" or
>> > :bar.  Why quote () then?
>> 
>> In this case it doesn't matter, because () is self-evaluating, but (foo)
>> isn't, so to me it seems more consistent to treat () as a list where I
>> want to make it explicit that I am inhibiting evaluation.
>
> Why do you want to make it explicit that you're
> inhibiting evaluation?
>
> As you say, in the case of () "it doesn't matter".
> But that's the only case we're (I'm) talking about.
>
> Apparently it does matter to you.  It matters to
> make it clear that you're inhibiting evaluation.
> Why?  What are you trying to say to a (human)
> reader by doing that?
>
> Not a rhetorical question; I really wonder what
> your thinking is about this.  You're not the first
> to do this on purpose, but I've never understood
> why someone would (in Elisp).

While I understand why () doesn't need to be quoted -- in the sense that
the reader sees no difference between the empty list literal and the
empty list notation -- I just don't like the inconsistency between the
requirement to quote a non-empty list if I want to inhibit evaluation,
while this is not the case for an empty list.

        '(1 2 3)

has to be quoted,

        '(1 2)

has to be quoted,

        '(1)

has to be quoted, but

        '()

suddenly doesn't?  If it weren't necessary, I'd consider the `eq`ality
of '(), () and nil this an implementation detail.

> It's not a big deal, admittedly.  But to me doing
> that could suggest to a human reader that there's
> some difference between evaluating '() and ().
>
> To me, including in terms of communicating with a
> human, this case is no different from t/nil/42/:foo.
> Why "inhibit evaluation" of something that evaluates
> to itself, and why draw attention to the fact that
> you're doing that?  What's the human message there?
>
> I mean, if the point of using () instead of nil
> is to pass a certain connation to a human reader,
> then we're (both) trying to give some importance
> to that human message - saying a bit more than
> just that the value is nil.  In that context (of
> admittedly minor importance), avoiding suggesting
> something untrue is relevant.

Another point is that in the case of deftheme, the () might look like a
empty parameter list, sort of like how (defun foo () ...) might look
like.  By explicitly quoting the list, we "clarify" that this is not the
same kind of thing.