Pretty-printing expressions yields nil *1, etc.

[cemerick]

My understanding is that cider implements pretty-printing in the REPL and elsewhere by wrapping the expression to be evaluated in an explicit clojure.pprint/pprint form (e.g. https://github.com/clojure-emacs/cider/blob/master/cider-repl.el#L554). Naturally, this leaves *1 holding nil, something I presume no one likes.

I'd like to suggest two options:

1. Evaluate an interpolated expression like (let [x %s] (clojure.pprint/pprint x) x), which will leave *1 set properly. cider-repl-handler can then simply not do anything with :value, which will ensure that only one printed representation of the evaluated value is seen, e.g.:

(lambda (buffer value)
  (if (not cider-repl-use-pretty-printing)
    (cider-repl-emit-result buffer value t)))

1. Alternatively (and perhaps as a more long-term solution), the standard pr-values middleware can be modified to provide a set!-able session-local configuration var so that clients/users can (a) set which function should generate the printed representation of :value, and (b) set which slot in the response message that representation should put in (since certain printed representations are likely not read-able, something that :value strings really should be).

I've tweaked my cider locally to do (1), and it works well. Would you like a patch for that?

Further, thoughts on the utility of (2) are welcome.

[bbatsov]

I'm aware of the current problem and I've been meaning to fix it for a while, but I've been pretty busy at work lately. I'd gladly accept a patch for (1).

Option (2) sounds great as well and something like your suggestion is definitely the way to go long-term. I'll have to think a bit more about the details, though.

[cemerick]

PR sent for (1).

I'm happy to revisit this once (2) is done. That's the only (reasonable) way pretty-printing in ClojureScript REPLs is ever going to work AFAICT.

[bbatsov]

Btw, why can't pretty printing be implemented for ClojureScript? (I don't track its development)