bug-gnu-emacs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

bug#66908: Exposing more public nadvice API


From: Visuwesh
Subject: bug#66908: Exposing more public nadvice API
Date: Sat, 04 Nov 2023 08:18:59 +0530
User-agent: Gnus/5.13 (Gnus v5.13)

[வெள்ளி நவம்பர் 03, 2023] Stefan Monnier wrote:

>>> It's clearly the internal function most frequently used outside of
>>> `nadvice.el`, indeed.
>>> I think it'd be OK to promote that function to a non-internal name.
>> It would be nice if you could since it would also provide some guarantee
>> of it being a fairly stable interface.
>
> Could you describe the circumstance where you need it?

We need to get the func-arity of the original function and not its
advice.  I have xref-find-definitions adviced but

    (func-arity 'xref-find-definitions);; ⇒ (0 . many)

returns the func-arity of the advice meanwhile, what we are really
interested in is the func-arity of xref-find-definitions as shown in the
*Help* buffer i.e.,

    (func-arity (advice--cd*r (indirect-function 'xref-find-definitions))) ;; ⇒ 
(1 . 1)

which is the right return value.  It might be nice to not have to call
`indirect-function' here for the "global" function but you can be a
better judge of that.

>> The other function that I saw get usage outside of nadvice and bytecomp
>> was advice--symbol-function.  Should we use its return value, or is
>> indirect-function's return value good enough for advice--cd*r?  IIRC,
>> help-function-arglist uses indirect-function and it also works for our
>> use case.
>
> I don't think `advice--symbol-function` is a good candidate because its
> semantics is not very clearly defined.  E.g. I'd be hard pressed to give
> a comprehensible documentation of it without either being too vague,
> or promising things I can't always provide, or getting too much into the
> the nitty gritty details of the various possible situations.
> That's why I haven't promoted the comment in its body
> to an actual docstring :-(
>
> Most callers are in only one of the many different situations, in which
> case they usually don't need that functionality.

In our case, the functions that will be checked for its arity should be
defined at the time of func-arity call.  Or at least auto-loaded AFAIU.
Philip can correct me if I'm wrong.





reply via email to

[Prev in Thread] Current Thread [Next in Thread]