On Wed, Nov 01 2023 12:48, Philip Kaludercic wrote:
> Tony Zorman <tonyzorman@mailbox.org> writes:
>
>> On Wed, Nov 01 2023 09:09, Philip Kaludercic wrote:
>>>> diff --git a/lisp/use-package/use-package-core.el b/lisp/use-package/use-package-core.el
>>>> index 34c45b7aec..5d0d554baf 100644
>>>> --- a/lisp/use-package/use-package-core.el
>>>> +++ b/lisp/use-package/use-package-core.el
>>>> @@ -1654,7 +1654,8 @@ use-package-normalize--vc-arg
>>>>                               (t (ensure-string v))))
>>>>                   (:vc-backend (ensure-symbol v))
>>>>                   (_ (ensure-string v)))))
>>>> -    (pcase-let ((valid-kws '(:url :branch :lisp-dir :main-file :vc-backend :rev))
>>>> +    (pcase-let ((valid-kws '( :url :branch :lisp-dir :main-file :vc-backend :rev
>>>> +                              :shell-command :make))
>>>
>>> Why is use-package checking for valid keywords in the first place?
>>
>> Better error messages, mostly. Especially people switching from
>> quelpa/straight/vc-use-package might be surprised that :vc is not a
>> drop-in replacement for those packages. I feel like alerting them to
>> this fact sooner rather than later makes for a better experience.
>
> IIUC this would raise an error when an unknown keyword is encountered,
> right?

Yes, a declaration like

    (use-package foo
      :vc (:url "url" :blargh "123"))

would result in the following message

    ⛔ Error (use-package): Failed to parse package foo: use-package: Keyword :vc received unknown argument: :blargh. Supported keywords are: (:url :branch :lisp-dir :main-file :vc-backend :rev :shell-command :make :ignored-files)

Things get a bit muddier if ':blargh' would be passed down to
package-vc-install.

Now that you mention it, I noticed a tiny mistake (resulting in a worse
error message!) in the second of the original patches. I've attached a
corrected version.

>>>> I will cheekily bump this, and also Cc. Philip as the most likely
>>>> reviewer.
>>>
>>> I don't use use-package nor am I familiar with the code base, so I
>>> wouldn't value my input that much.
>>
>> Oh, fair enough. In either case, I couldn't think of anyone else—sorry
>> for the noise :)
>
> I think that Stefan Kangas would probably be the best to ask, since he
> was the one responsible for merging use-package into the core.

Thanks! I have Cc'd Stefan, hoping to not come across as too pushy :)

  Tony