Kyle Meyer writes:

> Christopher Lemmer Webber <cwebber@dustycloud.org> writes:
>
>> Kyle Meyer writes:
>
>>> It looks like emacs-magit uses a revision that follows the popup to
>>> transient switch that happened after the 2.90.1 release.  Magit-annex's
>>> master branch has been updated to work with transient and will be
>>> released along with the next release of Magit.
>>
>> That may explain why I had to manually insert the @ keybinding it
>> supposedly should have done automatically.
>
> Correct.
>
>> I'm a little confused since I haven't been following magit though.
>> "popup to transient switch"?
>
> Sorry, here's more context: Magit's popup interface used to be defined
> with Magit-Popup.  Jonas Bernoulli wrote Transient to replace
> Magit-Popup [^1].  As of 1ed3febf (Merge branch 'transient' [#3728],
> 2019-02-14), all of Magit's popups have been rewritten with Transient.
> That change hasn't made it into a release yet, but it happened before
> the revision that emacs-magit is built from.
>
> [^1]: https://emacsair.me/2019/02/14/transient-0.1/

Ok, revised patch.  It works fine, though I am getting these errors:

  Compiling /gnu/store/3lc8vna2n2nhcdkfiqldg426ngjmzjnm-emacs-magit-annex-1.7.1-1.0bc9673/share/emacs/site-lisp/magit-annex-autoloads.el...

  In toplevel form:
  magit-annex-autoloads.el:10:45:Warning: reference to free variable
      ‘magit-mode-map’
  
  In end of data:
  magit-annex-autoloads.el:60:1:Warning: the function ‘transient-append-suffix’
      is not known to be defined.
  Compiling /gnu/store/3lc8vna2n2nhcdkfiqldg426ngjmzjnm-emacs-magit-annex-1.7.1-1.0bc9673/share/emacs/site-lisp/magit-annex.el...
  Loading /gnu/store/h9ihmqaf15vw70p6hxfqb4s48wwl5cn1-emacs-magit-2.90.1-3.c761d28/share/emacs/site-lisp/magit-version.el (source)...
  
  In end of data:
  magit-annex.el:834:1:Warning: the function ‘magit-log-setup-buffer’ is not
      known to be defined.
  Done (Total of 2 files compiled)

That seems strange/alarming.

Nonetheless, it seems to work fine... so not sure why we're getting that
error then.

One more question... emacs-xyz.scm seems inconsistent.  We have both of
these input styles:

      (propagated-inputs
       `(("magit" ,emacs-magit)
         ("transient" ,emacs-transient)))

      (propagated-inputs
       `(("emacs-magit" ,emacs-magit)
         ("emacs-transient" ,emacs-transient)))

The latter is more common amongst the rest of Guix... anyone have a
guess as to why we've got this variant of package input styles across
Guix's Emacs packages tho?