Presentation: 3 Things I Wish I Knew When I Started Designing Languages

Track: 21st Century Languages

Location: Pacific DEKJ

Duration: 5:25pm - 6:15pm

Day of week:

Slides: Download Slides

Level: Intermediate - Advanced

Persona: Developer

This presentation is now available to view on InfoQ.com

Watch video with transcript

Abstract

When I first sat down to design a language I had a rough time.  I sat for months in front of a blank screen trying to invent, ex nihilo, a syntax that looked sufficiently different from anything I had ever seen. Isolation wasn’t working.  But every time I talked to someone about what I was working on they spun me far off course, either with scorn (e.g. “hahahaha, what do you need a new language for?”) or with opinions (e.g., “it needs to have promises and futures”).  Finally, after years of labor and dissemination, my first language never really gained more than one user. I had failed; I was discouraged; I wished I had never begun.

That was seven years ago.  Today, that language (called `Dedalus’) still permeates my life in every way.  Sure, it has an uncreative syntax and rigid rules that sometimes make it difficult to express things that are easy in more familiar languages. Sure, it is essentially impossible to sell to users -- even my own team members and students!  Nonetheless, I use Dedalus every day. It helps me to write useful programs, but more importantly, it frames how I am allowed to think about its target domain.  When I use Dedalus to model a system I thought I understood, I often learn something new.  It has created a scaffolding on which every piece of my subsequent research hangs. That failed language is probably the most important success of my life.

In this talk, I hope to convince you that there is more than one reason to engage in language design and more than one way to do it.  I hope to communicate why someone would (and indeed many of us should) do something so perverse as to design a language that no one will ever use.  Along the way, I will share some of the extreme and sometimes obnoxious opinions that guided my design process.

Speaker: Peter Alvaro

Asst Professor @UCSC, Researching Data-Centric Languages/Analysis Techniques & Worked on Failure Injection Testing @Netflix

Peter Alvaro is an Assistant Professor of Computer Science at the University of California Santa Cruz, where he leads the Disorderly Labs research group (disorderlylabs.github.io). His research focuses on using data-centric languages and analysis techniques to build and reason about data-intensive distributed systems, in order to make them scalable, predictable and robust to the failures and nondeterminism endemic to large-scale distribution. Peter earned his PhD at UC Berkeley, where he studied with Joseph M. Hellerstein. He is a recipient of the NSF CAREER Award and the Facebook Research Award.

Find Peter Alvaro at

Similar Talks

Scaling Patterns for Netflix's Edge

Qcon

Playback Edge Engineering @Netflix

Justin Ryan