This belongs to #community-development but the answer is probably "no, as long as there's a free tier or a non-free tier that's been gifted by Slack-the-company".

reduced? is used when you want to exit early. The last item isn't guaranteed to have this value.

roger,thanks

postwalk, though i remember prewalk having some protection against the thing you just stated

it does

@U0K064KQV i shared a thread on this topic

it depends on if your transformation reaches a fixed point

user=> (clojure.walk/prewalk
 (fn [x]
   (prn "x" x)
   (if (= x :b)
     [:d]
     x))
 [:a [:b :c]])
"x" [:a [:b :c]]
"x" :a
"x" [:b :c]
"x" :b
"x" :d
"x" :c
[:a [[:d] :c]]
user=>

Hum, when I encounter a symbol I return (deref sym) and it seems like it then goes into this and encounters the symbol again return (deref sym) again, and it stackoverflows doing that

well, for starters you can't deref a symbol, you must mean resolving the symbol to a var and derefing the var

(walk/prewalk
 (fn[e]
   (if (= e 'i)
     `(deref ~e)
     e))
 '(i))

there is guarantee that such a thing is a tree, it can be a cyclic graph

ah, you aren't derefing it, you are producing a form that when evaluated would deref it

sure, that will cycle, no fixed point

Ya, sorry about that. So I want to be like, hey ignore the form I just returned, don't prewalk over it basically.

It be nice if I could like wrap it in a reduced to indicate that.

I think if you put in reduced it will stop at that branch Then postwalk and deref reduced

Hum, ya I guess I could do that with a second pass. But then I worry there could be an actual reduced in the code that's not from my macro.

And I can't wrap it in a list or vector to indicate termination either, so maybe I need my own custom type.

(defn prewalk
  "Like prewalk, but skips forms that are reduced."
  [f form]
  (walk/walk
   (partial prewalk f)
   #(if (reduced? %) (deref %) %)
   (f form)))

You can't have reduced? values in macroexpansions and expect them to compile, so introducing them yourself should probably be fine, especially if your prewalk is occurring before other macroexpansions.

So such an implementation of a "prewalk-substitute" would probably be fine.