CI: package-lint optional dependency issues (which-key)

[eshrh]

CI error:
Error: You should depend on (emacs "30.1") if you need `which-key--create-buffer-and-show'.

which-key was added to emacs 30.

Relevant (?):

• meow/meow-shims.el

  Lines 389 to 408 in e5817f3

  	(defvar which-key-mode)
  	(declare-function which-key--create-buffer-and-show "which-key"
  	(&optional prefix-keys from-keymap filter prefix-title))
  	(defvar meow--which-key-setup nil)
  	(defun meow--which-key-describe-keymap ()
  	"Use which-key for keypad popup."
  	(if which-key-mode
  	(setq
  	which-key-use-C-h-commands nil
  	meow-keypad-describe-keymap-function
  	(lambda (keymap)
  	(which-key--create-buffer-and-show nil keymap nil (concat meow-keypad-message-prefix (meow--keypad-format-keys))))
  	meow-keypad-clear-describe-keymap-function 'which-key--hide-popup)
  	(setq meow-keypad-describe-keymap-function 'meow-describe-keymap
  	meow-keypad-clear-describe-keymap-function nil
  	which-key-use-C-h-commands t)))

• Being told that transient should depend on a newer emacs or transient purcell/package-lint#281
• Allow ignoring of errors purcell/package-lint#125
  • purcell seems to be of the opinion that you should put each shim into a different, small package. I'm honestly inclined to agree, but it obviously involves a nonzero amount of more work.

Either way, something needs to be done about the package-lint error... recent commit (a159b74) was a naive attempt to quickly silence the problem but I quickly realized there's a bigger problem here. Warrants more discussion, hence this issue.

package-lint also doesn't like all the eval-after-loads in the shims but that's a different problem (or maybe not, idk)

[jcs090218]

Would you be open to considering tools like Cask or Eask? They provide more flexibility for tasks such as building, linting, and testing, which might give us more options to work around this issue—for example, temporarily disabling certain checks. 🤔

I'd be happy to help integrate the CI workflow into the project if that would be helpful. 😊

[DogLooksGood]

I have very little knowledge about CI workflow, I'm open to these tools.

[jcs090218]

I've opened a PR in #761 that reproduces the same error, but for now, I've moved it into its own step with continue-on-error: true enabled. What do you think of this workaround? 🤔