[misterbowfinger]

The example of why Elixir is less productive makes 0 sense:

Doh Phoenix and Elixir are trying to do all the best to help developers be productive, when you are a technology that promises such a huge scale you need to introduce practices that need to be decoupled => bit slower for developers.

Good example of this how a “model” writes to database. Module (with schema) need to call a changeset, changeset call repository, repository writes to database (example). That’s 3 manual steps where in Rails you have in in one.

Phoenix & Ecto work that way because 1) Elixir is functional, and 2) ActiveRecord mixes a lot of concerns that suck in the longterm.

IMO the real thing holding back Phoenix/Elixir is deployment & operational issues. It's being worked on a lot with Distillery, but from what I understand there's still a lot of kinks to be worked out.

[andy_ppp]

I personally think the deployment story heavily conflicts with container management. I’m going to give edeliver on real hardware a go at some point and things should work very well and be very simple.

[gamache]

Only if you want Erlang clustering or hot upgrades. In the usual army-of-one deployment scenario, Elixir and Erlang work inside a container just fine.

[andy_ppp]

Sorry I wasn't clear; what I'm saying is using the default deployment story with Erlang/Elixir is actually simpler than Kubernetes in most cases.