Having played with Clojure enough last year to be mildly dangerous (as I am now with Elixir!), the biggest thing which jumps out at me is more tool ergonomics than fundamentals. The Clojure REPL (similarly to other LISPS) is typically linked directly to your editor, such that you can hit a keystroke and instantly eval any changes into the running application, with output into the editor buffer (either as a transient overlay for info, or to a comment for further use). It’s hard to explain how smoothly interactive this makes the development process feel.

The up arrow should remember multiline input for goodness sake.

Go use Julia, and type this:

julia> function add(x, y)
   x - y
end
julia> add(4, 3)

Now, go back and edit the add function to fix it. It’s straightforward.

It is not so in elixir (or python, or ruby, so we’re not alone but that is no excuse to not do better).

Also, I kind of want naked functions in the repl context. That’s probably too much to ask but it could be done by having a “IEx.Repl” module that is “imported” into the context and the def macro in the IEx scope triggers a recompilation of this module

That would be so awesome. I work around it by using clipboard history/manager.

I’m super glad they recently allowed pipelining without requiring parens.

If we’re making requests, I would like a dynamic prompt. Seems easy to allow an anonymous function as a prompt, rather than an interpolated string. Use case is we have a multi-tenanted app and it would be nice to see what tenant you’re set to in the prompt:

iex(1)> Tenant.set("foo")
iex(1)[foo]> Tenant.set("bar")
iex(1)[bar]>