My perspective, from a few decades in the SW industry. For an enterprise to _even start_ to consider using Nim, Nim must have the capability to efficiently work with a code base of significant size. This has a couple of implications.
First, IC is a mandatory first step. The mandatory second step, and close behind the first: a developer must be able to rapidlty analyze, develop features, diagnose problems, and refactor the design of code that was written by someone else in a collection of significant size. Typically without the benefit of much if any design documentation. Sound familiar to some? For Nim, the key enablers for the second part are nimsuggest, nim check and (until we get a native IDE) nimlsp. From what I have seen so far, they are currently _beta_ quality _at best_. So the other focus really needs to be on getting those critical components ready for prime time. REPL is down in my list. Way down - including for learning the language. I've used python's REPL and enjoyed it. But having a good IDE where I can code a quick experiment, have compile issues pointed out as I type, and click on a button to compile & execute said experiment, beats REPL hands down.
