implicit dom_id for morph selectors

[leastbad]

Type of PR (feature, enhancement, bug fix, etc.)

Enhancement

Description

This proposes to allow morph to accept a resource in lieu of a CSS selector. Any resource will be converted to a string with dom_id.

Before:

morph dom_id(post), "I am a teapot"

After:

morph post, "I am a teapot"

Before:

morph {dom_id(post1) => render(post1), dom_id(post2) => render(post2)}

After:

morph {post1 => render(post1), post2 => render(post2)}

I've also made the Reflex#dom_id helper more powerful than its Rails counterpart, so that it now accepts ActiveRecord relations and arbitrary objects (that get converted to strings):

dom_id(post1) # "#post_1" (the current behaviour)
dom_id(Post.all) # "#posts"
dom_id("teapots", "search") # "#teapots_search"

That's all pretty cool, but with the above in place, we can dial it up.

morph User.first

This is equivalent to:

morph dom_id(User.first), render(User.first)

And, thanks to a quirk in the way that morphdom works, this is now possible:

morph Post.all

This is equivalent to:

morph dom_id(Post.all), "<div id=\"#{dom_id("posts")}\">#{ApplicationController.render(Post.all)}</div>"

morphdom DGAF so long as the outermost container element has the same id as the target. If you happen to be rendering a collection with a partial that lines up with render defaults, we can now infer everything but the collection itself.

Finally, I added a render_collection method to Reflex that developers can use for more complex morphs:

render_collection(render(posts, partial: "home/posts", locals: {pagy: pagy}), "#my_content_div")

<div id="my_content_div">post1post2</div>

Thanks to @RolandStuder for the initial inspiration.

Why should this be added

Developer happiness and the principle of least surprise.

Checklist

• My code follows the style guidelines of this project
• Checks (StandardRB & Prettier-Standard) are passing
• This is not a documentation update

[hopsoft]

I wonder if there is any additional syntactic sugar we could include by leveraging to_partial_path. Too magical?

[leastbad]

I think that we're implicitly leveraging to_partial_path when we support the morph @posts syntax, even though it's not something we have to do directly.

What do you have in mind? Tell me like I'm not in your head! 🤓

[leastbad]

I reworked the new dom_id method so that it was a case structure, but standardrb kicked my ass and turned it into an if/elsif and I'm no longer sure it's still cute.

Either way, I had to allow the wrap method to squash the # being prepended to prevent a dreaded double-hash. I also added a sub to the fall-through dom_id case to prevent people from accidentally sending an extra # character into the method.

All of this is open to iteration. 💯