Hacker News Reader

Comment by losvedir | Hacker News Reader

Maybe a better link: https://elixir-lang.org/blog/2019/01/14/elixir-v1-8-0-releas... . One thing about that blog post that really caught my eye:

> There is only one last major feature planned for upcoming Elixir versions, which is the addition of mix release to Elixir itself, streamlining the experience provided by packages like distillery.

Aside from having native support for releases, which would be killer, I was really struck by the idea that the language will be basically "finished". You rarely see that in software these days, but I think there's a lot to be said for trying to get to that point.

View on HN

A great point. Since it's built on top of the erlang VM, and it has all the cool tools built into the eralng VM and OTP, Elixir has been able to rapidly go from a prototype language to a production grade environment to work in extremely quickly.

But also to basically decide that the language ergonomics and syntax is "finished" this quickly is truly impressive. The addition of macros, combined with the huge focus on developer experience, I think allows the core team to say "its feature complete" while not worrying about any particular missing features in the language (you can probably just implement it yourself anyway)

Has Elixir input driven any new features in Erlang's BEAM VM?

At least some features, yes, and also some optimizations.

José Valim was the one who added "Add {continue, Term} and handle_continue/2 to gen_server" - https://github.com/erlang/otp/commit/69e009e3e1ad899a4609ff3...

José's contributions are at https://github.com/erlang/otp/commits?author=josevalim and Michał Muskała's at https://github.com/erlang/otp/commits?author=michalmuskala - there may be other Elixir devs in https://github.com/erlang/otp/graphs/contributors. (I even got in a tiny optimization myself once: https://github.com/erlang/otp/commit/07b0f4315b079cce7aeef7b...)

Interesting. Thanks for the links!

I think Jose and team have done a great job with this, especially judiciously limiting what features get added to the core language.

This is also an area where Elixir is benefiting greatly from the maturity and ongoing development of the Erlang VM. Elixir is, in a sense, simply a text based interface on top of the BEAM.

This is also pretty huge from an ecosystem point of view. It means you can have books, video series, tutorials etc. targeting this final feature set which is very nice. Not that it's a huge issue but it's cool to buy a programming book knowing that it'll contain a language that will not change for a long time. It's not a huge deal, I own the PragProg book (I don't recall if it's >=1.6 or if there was a version before) but now I know I'll very likely get the 1.8-version when it is released.

Being that the language is user-extensible, where do you see it going in the future? Will "finished" mean feature-complete, or will we see anything worth incorporating into the language?

This isn't directed specifically at you, but to anyone who might have ideas.

The Elixir team took a deliberate step to limit the features built into the language itself, and prefer non-core solutions developed as external packages instead.

For example, they recently scrapped plans on building property testing into the language core.

Which is something I find very refreshing, TBH, having worked with the JS ecosystem, I somehow find this approach to be rather relieving for some reason.

It's pretty much sum up in this keynote (The Next Five Years of Elixir) https://www.youtube.com/watch?v=suOzNeMJXl0

Thanks, we changed to that from https://github.com/elixir-lang/elixir/releases/tag/v1.8.0.

Reply on news.ycombinator.com