Fix: Revert de0b62b604e87ce43e2adc9e09bcd5174e8b877a (evil-with-delay change)

[gagbo]

EDIT: This PR now fully reverts the problematic commit, while adding some glue to make CI green.

Old version

This solution is incomplete, and might still fail because of the
impossibility to expand the hook-name dynamically, see the
discussion/screenshots in the PR

Brief summary of what the PR does

Try to address the issues in #750

Checklist

Assume you're working on mpc mode:

• byte-compiles cleanly
• M-x checkdoc is happy. Don't manually write (provide 'evil-collection-mpc), M-x checkdoc can do it automatically for you
• define evil-collection-mpc-setup with defun
• define evil-collection-mpc-mode-maps with defconst
• All functions should start with evil-collection-mpc-

[gagbo]

Edit

This comment is no longer relevant to the PR here, but mentions the former tries to use properly the new evil-with-delay macro to migrate (tl;dr: no way to have a dynamic hook-name without writing yet another macro layer on top of everything)

Old comment

I mostly tried to solve it by trying different quoting strategies and then using macroexpand-step to look at the result.

Before the PR (current state on master):

Current state as the PR is made:

Important notes

• I still didn't test the fixes locally, it's a bit painful to do so yet, I mostly wanted to start a conversation about what we should expect the macro to expand to.
• Note that in the before screenshot, the name of the hook is unique (evil-delay-in-{backquote}), which might hint that the issue is indeed that the quoting is off with the master version of evil-collection-translate calls.
• Also note that I couldn't find a way to properly expand the value of hook-name in the resulting expansion, so I don't really know how to unlock this part. In a scratch buffer, calling macroexp-quote with a dynamic var seems to work:

So I'm not sure how I can go further and fix that. Is the change of evil-mode actually breaking after all? Or is there a quoting pattern that would fix this issue?

[jojojames]

I think we should have a reverse commit of the original commit that caused the regression and then we add your fixes on top of the reverted commit. This way, we can evaluate the necessary changes from the original (deprecated) method and your new changes.

[gagbo]

Probably. I just made the revert commit this way as a compromise to keep the CI green. If it's not important, just reverting is probably going to be enough.

I'm not well-versed in macro magic, but I had a lot of issues trying to get the custom evil-collection-translate-in-%s name in the new evil-with-delay macro.

[condy0919]

If there is no further discussion, I will merge 2 days later. :-)