Erlang Factory Recap
Posted: May 5th, 2009 | Author: kevin | Filed under: Erlang | View CommentsSummary
The SF Erlang Factory was easily one of the best technical conferences I’ve attended in years. The amount of energy, passion, and technical discussions were far beyond anything I’ve experienced before. I literally felt like I was watching the Erlang community bootstrap itself into a larger presence.
What was most interesting to me was the interaction between the Erlang “old guard” — Ulf Wiger, Robert Virding, Francesco, Cesarini, John Hughes, etc — and the “new guard” — basically everyone using Erlang since Joe’s book came out in 2006. It was quite clear the new guard was taking Erlang in new directions. What was equally obvious was the old guard was just as excited. I definitely got the impression they were quite pleased to see Erlang gain some popularity and increased relevance.
Also striking was the complete lack of Erlang prima donnas. Many of the old guard have been using Erlang for well over a decade and could’ve been prima donnas. Instead, the conference vibe was friendly, open discourse with a high signal-to-noise ratio. I can only hope the community remains open and friendly as it inevitably grows.
What follows is a day-by-day talk-by-talk summary of the conference as I experienced it. If you want the gory details of the conference feel free to keep reading. If you’re looking for a high level summary, read no further than this: Erlang Factory rocked and you should’ve been there.
Thursday Morning
Woke up tired from the speakers’ dinner on Wednesday night but I was determined to wring as much as I could from the country’s only Erlang conference. Francesco gave an inspirational opening speech touching on the themes of community growth, Erlang’s growing popularity, and the future.
The next keynote was Robert Virding’s presentation of the “Erlang Rationale”. This was a great distillation of the points he made in his PDF of the same title.
“A lot of the system is a joke.” — Robert Virding, SF Erlang Factory 2009
Robert shared a lot about the history of Erlang, the background on some of the key design decisions, and the overall rationale which shaped Erlang’s early years. A lot of the information was historical but it filled in some holes I had in the context around certain language features.
The final keynote was by Engine Yard co-founder and my former colleague, Ezra Zygmuntowicz. Ezra discussed the challenges involved in managing cloud-y deployments and demoed the latest version of Nanite, the technical underpinning of Engine Yard’s Solo offering. The high point of Ezra speech was when he used Nanite to control most of the computers in the audience to speak those immortal Erlang words: “Hello, Joe. Hello, Robert.”
The sessions began after the keynotes and I was up in the first slot of the Enterprise Integration track. My talk on natter went over fairly well. I underestimated how much material I had and wound up ending about 20 minutes early. There were a several questions at the end and a few people stuck around to ask questions after the talk was over.
Next, I went to John Hughes’ presentation on QuickCheck, the industrial strength testing tool for Erlang. It’s an amazing bit of software which makes tracking down nasty bugs, like race conditions, much easier. If I was managing a team doing anything significant with Erlang I’d definitely investigate QuickCheck.
Thursday Afternoon
The first talk after lunch was Tom Preston-Werner’s talk on Erlectricity. Well, only the first half of the talk was about Erlecricity. The second half was the unveiling of a “new” binary format for interchange between Erlang and Java based on Erlang’s wire format. It looked pretty interesting but I think I’d need to do more Ruby <-> Erlang interop to truly appreciate it.
Next up was Justin Sheehy’s talk on webmachine. I was already familiar with webmachine since I use it in my Hands On Erlang classes. It was great to sit back and watch the audience’s reaction as Justin presented each feature.
“webmachine helps you create applications which are shaped like HTTP.” — Justin Sheehy, SF Erlang Factory 2009
The talk was a big hit and easily one of the best I attended at the conference.
The last talk of the day was Dave Fayram’s and Abhay Kumar’s talk on Katamari, aka fuzed. For anyone who hasn’t seen fuzed or heard about Katamari, it’s a kick-ass platform for managing distributed heterogenous services. Dave explained he had worked at a couple of start ups and was tired of doing the ah-crap-we-got-dugg shuffle and furiously trying to scale infrastructure. Katamari is serious tech and the Powerset guys continue to impress with their technical skills and thoughtful designs. The fuzed project will be updated with the latest Katamari goodness next month so the rest of the world can play, too. Yes, you should be excited.
I skipped the BOF sessions and decided to hoard what energy I had for the Erlounge at Gordon Biersch.
Erlounge, oh, Erlounge. The first rule of Erlounge is you do not talk about Erlounge. Let’s just say a good time was had by all and leave it at that :)
Friday Morning
Morning attendance was distinctly down compared to Thursday. I suspect Erlounge was to blame. Damien Katz led off with a very inspirational keynote on how he came to work on CouchDB. It was the same talk he gave at RubyFringe but more inspirational and touching in person.
I sat in on Martin Logan’s talk which covered introducing Erlang into the workplace. He used his own erlware project as an example of the ways he’s tried, and largely succeeded, to grease the skids for Erlang at Orbitz. Martin is a friendly guy with a wealth of Erlang experience. I’m sure I’ll use many of his tips in my consulting business.
Next up was Bryan O’Sullivan’s excellent talk on Haskell. Bryan presented Haskell in a very clear manner and it’s quite obvious he has an extremely good grasp of the technical bits.
“Our motto has been avoid success at all costs.” — Bryan O’Sullivan, SF Erlang Factory 2009
It was nice to experience a talk about Haskell which dispensed with the let’s-review-functional-programming and got right to the meat of Haskell. 2009 is definitely my year to learn Haskell.
Closing out the morning was Lennart Öhman’s talk on managing fail over and take over in Erlang systems. Lennart is the primary inventor of OTP behaviors so it was a treat to hear him speak on OTP. I was a little disappointed to find out his framework is not released yet but I still learned a bit from his talk.
Friday Afternoon
After lunch I attended Tony Arcieri’s talk on REiA. REiA is a Ruby-like language which runs on top of the BEAM VM. Tony’s started off with technical sleights of hand like defining a modules and named functions in the Erlang shell. The core of his talk was an overview of REiA and some of the design decisions he’s made during development. I don’t see myself using REiA soon but it’s interesting to watch languages evolve on the Erlang platform, nonetheless.
The last talk of the conference was Cliff Moon’s presentation on Dynomite. Cliff’s direct communication style coupled with Dynomite’s technical impressiveness combined to create a great talk. The high point of Cliff’s talk came when he downed a couple of production servers to illustrate how Dynomite’s replication and fault-tolerance mechanisms worked.
“So this is the dashboard for our production cluster.” — Cliff Moon, SF Erlang Factory 2009
If you need a key/value store with replication and fault-tolerance you really should give Dynomite a try.
Finally, it was time for the wrap-up session. Francesco built on his previous themes of growth and community and asked each of us to continue building cool things and grow the Erlang community.