normalize the class prop

[lucywang000]

To support specifying :class with vectors/lists of symbols. This could make it easier to construct class names conditionally.

{:class (cond-> '[foo bar]
     (some-test?)
     (conj 'barz)}

And normalize the class string:

{:class "foo bar
baz"}

;; becomes
{:class "foo bar baz"}

This could be handy when writing multiline string and still control the 80-chars line limit when there are lots of classes. It's very common when using functional frameworks like tachyons.

This happens at macro expansion, so no runtime overhead is added.

[tekacs]

This would be super helpful -- I've currently customized my $ macro to support an arglist-style argument before props that's a class vector.

The rules that I've set up are:

• Quoted symbols, strings and keywords are all valid inside the vector.
• Dynamic evaluation is permitted in the value (because I often want to use (when ...))
• Nil values are stripped out after evaluation.

So that these work:

;; ns: namespaced
(def vars-too '[a b c])

($ :div ["some-class" '[things go "here"] (if condition "one two" :three) (when something namespaced/vars-too)])

Keywords are probably unnecessary, but other than that the functionality is helpful.

[lucywang000]

The behavior looks great now, thank you! Only thing I ask now is a bit of restructuring to colocate all the branches for this inside normalize-class for readability.

@Lokeh that would improve the readability, but that would lose the benefits for compile time optimization, e.g. 1) for a string value, there is no need to emit a string? check at runtime 2) constant lists like '[foo bar] could be turned into a string at compile time.

[lilactown]

@lucywang000 that should still happen; at macro time, it will take the :clj branch of normalize-class which should only emit a runtime normalize-class if it's not a string? or quoted vector.

At least, that was my intention with my suggested edits.

[lilactown]

I think I forgot to add the conditions to the :clj normalize-class. 😅

[lilactown]

I'll accept this PR and play with it a bit. I'll tag you in the PR if I do reorganize the normalize-class logic.

[lucywang000]

Thanks @Lokeh !