Erlang Factory SF 2011

Erlang Factory SF 2011 Report

Erlang Factory did not disappoint. I've been wanting to attend an Erlang Factory conference since I started working with Erlang in 2009 and this year the stars aligned so that I could attend (thanks Opscode!). I had a great time putting faces to names, meeting new people, and absorbing knowledge. Below are a few highlights from the whirlwind.

Better Erlang Programming through Toolistry

The conference kicked off with a keynote from Kostis Sogonas outlining a handful of tools produced by his lab that can help you to level-up your Erlang programming skills. Here's the list:

  1. Dialyzer - "I don't think you should be ignoring Dialyzer warnings; Dialyzer is never wrong" -- Kostis Sagonas. Dialyzer isn't a brand new tool. If you aren't using it, you should do something about that. Choice quote from Kostis: Getting started looks like this:

    # annotate your source with -type and -spec as described 
# in http://erlang.org/doc/reference_manual/typespec.html

# build a plt database for dialyzer
dialyzer --build_plt --apps erts kernel stdlib

# compile your erl code with debug_info
# If you're using rebar, just add the following to rebar.config
{erl_opts, [debug_info]}.

# run dialyzer and fix warnings
dialyzer ebin

    Adding type and function specs to your code serves as useful documentation (which is now parsed by Edoc in the latest Erlang release) and has the advantage over explicit guards of not imposing any runtime cost.

    Finally, visiting the Dialyzer website is worth a few minutes as there are some reports showing warnings produced by the in-depvelopment version of dialyzer. For example Heisenbugs and Intersection warnings.

  2. TypEr (pdf) - You can use this to bootstrap writing type specs for your code. Comes with Erlang. Get started with typer --help.

  3. Tidier - Next best thing to having a code review? Tidier attempts to refactor your code to improve its readability and performance. I'm Looking forward to taking this for a spin. The demos in the presentation showed an number of impressive automatic cleanups, but it remains to be seen if this is able to provide value for real-world cases. In any case, certainly on my list to spend some time with.

  4. PropEr - A property based test tool ala Quviq's QuickCheck. I think property based testing is what everyone is going to be using once they get clued in to how awesome it is. PropEr looks interesting because it understands type specs and can test those specs automatically. In theory, this means you can have an auto-generated regression test for a module which you can use to test a new underlying implementation.

Simon Thompson presented on Wrangler a refactoring tool that plugs into Emacs (and other editors) and looks to provide some useful automation for common tasks like function extraction, function generalization, and renaming along with heuristics for automatically identifying "clones" that are candidates for function extraction.

John Hughes provided an energetic introduction to property based testing via QuickCheck mini. With freely available tools like QuickCheck Mini and PropEr, the time has come to get over the hump with property based testing. I'm convinced there is tremendous value to be mined out of these tools with modest investment of effort.

Dave Smith and Joe Williams presented on Rebar and OTP release handling with Rebar. There seems to be growing momentum behind rebar as a standard build tool for Erlang. A few notes from the talk:

Coming Soon to Erlang

Wrap Up

There were many other interesting talks and even more great conversations. These notes are incomplete, but had to get something up before day-to-day concerns took over and the details leaked out of my head. To the folks at Erlang Factory and all the attendees, thank you! I hope to make it next year.

archived on 2011-03-28 in ,

blog comments powered by Disqus