Friday, October 5, 2012

Strange Loop 2012: Day 1

So, I'm going to change the pace of these Strange Loop blog posts a bit. I'm going to make them more of a narrative rather than technical reviews.

After a rather unremarkable Continental breakfast, we were off to the main venue, the beautiful Peabody Opera House for the first "official" day of sessions at Strange Loop.

To kick things off a keynote by Michael Stonebraker (of Ingres and Postgres fame) titled, "In-Memory Databases: the future is now". The talk was very illuminating in some respects with some interesting points, findings and solutions to the problem of large data transaction processing. However, things derailed a bit when the speaker made some rather contentious and controversial comments about NoSQL databases, which in the eyes of many attendees were either ill-informed or misguided. In essence, Mr. Stonebraker claimed that databases without ACID support aren't worth your while and data replication schemes of NoSQL databases all they do is "produce garbage".

After they keynote, I decided to attend a session titled, "Monad Examples for Normal People [...]" which was rather disappointing. For one, the speaker flew through his presentation: he was done in about twenty-two minutes even though sessions are supposed to be fifty minutes. Secondly, the talk was highly technical and I would venture to say that less than five percent of all attendees actually understood the concepts being presented. It was not at all taught or designed for "normal people", it was designed around "normal people" who have a deep understanding of Monads.

The second session I attended was named, "Renormalize -- the sequel to SQL". It was a fairly interesting presentation with the speaker proposing what amounts to cherry picking features from both SQL and NoSQL databases. The speaker was also promoting the Akiban database server that implements the concepts he spoke about during the presentation.

The next session I went to was called, "As we may do -- Augmented Reality and Computer Vision" which was in the lukewarm side of things. The presenter, Neil Milsted, kept murmuring and speaking so softly that attendees had to ask him to speak louder a few times. He had a neat demo of a Rubik's cube solver using computer vision, but regretfully there wasn't enough light in the room for the computer to see the Rubik's cube, but after a few minutes the speaker managed to lit up the face of the cube and he was able to show off the solver to the audience.

The next session, in my opinion, was the highlight of the day. "Relational Programming with miniKanren" simply blew my mind. miniKanren is not just a language, it's a meta-language that among other features generates programs that solve a given problem assuming such program exists! So, for instance the classic 5 + 1 = X, it will generate a program that will return 6. But what if you have X + 2 = 4? It will generate a program that will produce the value 2 and so forth. Mind you, they were not coding to solve the equation (that would be entirely too pedestrian), they were coding a program to write other programs that satisfy the relation or equation.

Next up, was David Nolen's talk about ClojureScript which as you might surmise, it's an altJS language based on Clojure. David's presentation skills together with his deep understanding of Clojure and JavaScript made this session one of the best of the day, at least in my opinion.

Following that, I decided to attend a session titled, "Type vs. Tests: an Epic Battle?" and left about ten minutes after it started.

After waiting around and hawking some t-shirts from the few vendors present, I went to a presentation by Intel's Stephan Herhut about River Trail, a JavaScript library and firefox extension for enabling parallel processing in JS. It was actually very interesting to see the concepts and theory behind this library as well as seeing it in action through some impressive demos. For instance he showed a browser-based scene rendering with normal JS and it was crawling at a whooping three or four FPS. Then he showed the same scen rendering written on River Trail and the frame rate went up to nearly twenty FPS. He then showed other instances where the frame rate was many-fold increased  from plain JS to River Trail. Good presentation overall and very interesting product.

To close the first "official" day of sessions, a keynote by Google's Lars Bak on the history, present and future of virtual machines and more specifically virtual machines that live in the browser. A good mix of technical details and anecdotes, was a good way to close the day in a positive note.

As if a long day full of sessions wasn't enough, the organizers put together what they call "unsessions" which are sessions led by either speakers, volunteers, locals or attendees about a wide variety of topics and in a more informal format. All the unsessions were packed to brim! (that should tell a lot about the highly motivated crowd that attends Strange Loop).

Of all the options at hand, I attended a basic-level unsession about MongoDB. It was actually pretty fun with attendees and presenters sharing "war stories", lessons learned and tips dealing with Mongo.

The last unsession I went to was "Mentoring Software Apprentices", which was fantastic. It was led by 8th Light's Doug Bradbury and was about how the best way to attract and keep talent in your organization is to "grow them". Techniques and suggestions as to how to create a solid relationship with the candidates and arm them with the tools to flourish in the organization. Lots of good information was shared and exchanged among the speaker and attendees.

And as the cliché says, that's a wrap!

No comments:

Post a Comment