bug#65386: [PATCH] ; Refine some 'package-vc' docstrings

[Philip Kaludercic <philipk@posteo.net>]

Eshel Yaron <me@eshelyaron.com> writes:

> Philip Kaludercic <philipk@posteo.net> writes:
>
>> Eshel Yaron <me@eshelyaron.com> writes:
>>
>>>> Also, would it make sense to determine this at compile-time?  On the
>>>> other hand, if a VC backend is installed later on from ELPA, we would
>>>> want the custom type to reflect this.
>>>
>>> Yes, I couldn't find a way to defer computing the set of candidates to
>>> "customization type", I'm not sure if that even makes total sense.  I
>>> think it's not that crucial since someone adding a VC backend and
>>> immediately trying to customize these options seems to me like a very
>>> minor edge case, and we had the same issue prior to my patch anyhow.
>>
>> Likely yes, but if we have two mostly equivalent solutions, I think that
>> flexibility without having to restart Emacs is preferable to performance.
>>
>
> Hmm I'm not sure I completely follow.  Which two solutions do you have
> in mind?  I agree that it would be nice have the `:type` updated if the
> user adds a relevant VC backend, but I don't think `defcustom` supports
> something like that.  How do you propose to achieve that flexibility?

Sorry, in the end it doesn't matter, I can't quickly think of a way to
define a type flexible enough to automatically update to the
environment, nor does computing a constant value at compile time seem
either worth it.

>> [...]
>>
>>> From 8a339d3bffe81f80e7e968505c40855d6598b8ac Mon Sep 17 00:00:00 2001
>>> From: Eshel Yaron <me@eshelyaron.com>
>>> Date: Sun, 20 Aug 2023 16:20:54 +0200
>>> Subject: [PATCH] ; Refine 'defcustom' types in 'package-vc'
>>>
>>> Only include VC backends that support cloning in the ':type' of
>>> 'package-vc-heuristic-alist' and 'package-vc-default-backend'.
>>>
>>> * lisp/emacs-lisp/package-vc.el (package-vc--cloning-backend-p)
>>> (package-vc--backends): New functions.
>>> (package-vc-heuristic-alist, package-vc-default-backend): Use it.
>>> ---
>>>  lisp/emacs-lisp/package-vc.el | 13 +++++++++++--
>>>  1 file changed, 11 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/lisp/emacs-lisp/package-vc.el b/lisp/emacs-lisp/package-vc.el
>>> index a3762d252b0..44a4624e49f 100644
>>> --- a/lisp/emacs-lisp/package-vc.el
>>> +++ b/lisp/emacs-lisp/package-vc.el
>>> @@ -62,6 +62,15 @@ package-vc
>>>  (defconst package-vc--elpa-packages-version 1
>>>    "Version number of the package specification format understood by package-vc.")
>>>  
>>> +(defun package-vc--cloning-backend-p (be)
>>> +  "Return non-nil if the VC backend BE supports cloning repositories."
>>> +  (or (vc-find-backend-function be 'clone)
>>> +      (alist-get 'clone (get be 'vc-functions))))
>>> +
>>> +(defun package-vc--backends ()
>>> +  "Return a list of VC backends suitable for cloning package VCS repositories."
>>> +  (seq-filter #'package-vc--cloning-backend-p vc-handled-backends))
>>
>> Do you think that these utility functions are useful enough to have in
>> general, or would it be better to just define a defconst once that
>> generates the `(choice ...)' object?
>
> I think these are nice to have, but I don't mind removing them in favor
> of a `defconst` (or a `define-widget` for that matter) if you think
> that's better.

I think a defconst would be a simpler solution, having functions like
these always bear the danger of being misused, despite being marked as
internal.