APIcon 2014: Making the Case for RAML

Michael Vizard
May. 28 2014, 08:34PM EDT

No matter how great an application may theoretically be, it’s increasing becoming apparent that the difference between success and failure is the quality of the end user experience that winds up driving actual usage of its API.

Looking to make it easier to develop APIs that drive application usage, MuleSoft (parent company of ProgrammableWeb) is backing a RESTful API Modeling Language (RAML) that is intended to take make it easier for developers of an API to identify patterns and more easily discover what function an API provides.

Speaking at the APIcon 2014 conference today, Dillon Compton, a product manager for MuleSoft, showed how developers can not only use RAML to design APIs, but also describe them more clearly and simply. The goal, says Compton, should be to use RAML as an organizing principle around which much of the application can actually be built.

After invoking RAML to apply more discipline to the API development process, Compton says there should be significantly less friction in an API economy that now only consists of about 13,000 public APIs, but also millions of private ones.

The challenge facing the developer community today is that unless APIs are easy to understand and consume, the rate at which the overall API economy will grow could slacken. Unless that API is simple and intuitive to understand Compton says the assumption now is that it’s an unstable API that might result in higher maintenance costs if invoked by a third-party.

Compton says that despite the existence of other API modeling frameworks, MuleSoft felt there was a need for a framework that more easily generated APIs that could be read by humans. That capability not only increases the accessibility of the API, it also makes it much less likely that anyone will use an API in a way that was unintended.

At the moment, Compton says there are 3,000 users of the hosted tools that MuleSoft has made available. While MuleSoft also created an API Designer for RAML, Compton says that MuleSoft is pleased to see that the larger RAML community is now creating its own tools. The goal, says Compton, was to create a modeling language specification that people were excited enough about to actually use. Accomplishing that goal required MuleSoft to initially build the first set of RAML tools, says Compton.

Essentially a subset of YAML, Compton says that MuleSoft would love to engage with providers of other API modeling frameworks, "and I hope to get those conversations started while API Blueprint and Swagger are in town for APIcon."
In the meantime, Compton says that developers should look forward to a formal 1.0 release of RAML sometime this summer as part of an overall effort to improve the application programming experience.

Michael Vizard