Skip to main content

Benjamin Oakes

Photo of Ben Oakes

Hi, I'm Ben Oakes and this is my geek blog. Currently, I'm a Ruby/JavaScript Developer at Liaison. Previously, I was a Developer at Continuity and Hedgeye, a Research Assistant in the Early Social Cognition Lab at Yale University and a student at the University of Iowa. I also organize, ICRuby, OpenHack Iowa City, and previously organized NewHaven.rb. I have an amazing wife named Danielle Oakes.

JWT: JSON Web Tokens

by Ben

JSON Web Tokens are an open, industry standard RFC 7519 method for representing claims securely between two parties.

Source: JWT.IO

Neither Microsoft, Nokia, nor anyone else should fork Android. It’s unforkable.

by Ben

Neither Microsoft, Nokia, nor anyone else should fork Android. It’s unforkable. | Ars Technica.

An interesting take on Google’s Android strategy. In summary, Android is Open Source, but Google is really pushing that apps depending on proprietary Google Mobile Services (GMS). This explains why some apps aren’t in the Amazon Appstore. I’d be interested in how Cyanogen Mod is solving the GMS problem.

At any rate, the only service that is really discussed is in-app purchases… I’ve never made an in-app purchase, so that wouldn’t be a deal breaker for me (although I have purchased apps before). I could see how that would be a big blow to the current mobile gaming industry. I would imagine Google’s mapping service is another… but it’s hard to say what other GMS services I would really care about.

Fear of Shipping and Service-Oriented Architecture

by Ben

This passage from The Travis CI Blog: How to Get Over the Fear of Shipping stood out to me:

Complex dependencies

Over the last two years, we split up Travis CI into lots of smaller apps. While this was great help to scale out the platform, there was an effect we initially didn’t foresee.

Most of the apps depend on one large library containing all code, it’s called travis-core.

Whenever something in that repository changes, we ideally need to deploy all our apps to production to make sure the change doesn’t impact any of them.

This big, single dependency is an impediment to us shipping code, and it’s a common problem.

Do your projects have a common dependency, a library of shared code?

We’re slowly breaking this part up, pulling the relevant code into all of our single apps rather than share a big bundle amongst all of them. While this may cause some duplication, it allows us to ship faster in return and with more confidence.

The duplication is outweighed by every app only knowing about our data model what it needs to. We can reduce database models to the information relevant to each app, slowly replacing direct database access with API access, an architecture where every app knows as little as possible about the others and about the overall data model.

I’ve seen this anti-pattern happen first hand. Building a set of SOA apps really makes SOLID design principles (especially “single purpose”) very salient. (Think of each app as a class in normal object oriented programming.) If one app needs to know too much about another app (or worse, it needs to share access to a database!), then that app probably isn’t focused enough.

This can be easily observed in the shared library code. If your shared library (e.g. travis-core above) has too much churn and too many deploys just so the business logic is consistent across multiple clients of the shared library, that’s a sign that some reorganization needs to happen. Like the Travis developers mention, this can easily lead to concerns when deploying. Know how it feels to have to update Rails in multiple apps when there’s a security fix? Imagine if you had to update an internally-maintained gem every time you deployed.

When I’ve seen SOA done well, there’s a well-defined API that any clients must use, and nothing else. If you try to skimp on a well-defined API, SOA will likely be more trouble than it’s worth.