http://www.sauria.com/blog/2009/04/20/dsldevcon-2009/ Open Source, Modern Programming Languages, OS X, Photography, and ... Thu, 19 Jan 2012 07:08:30 +0000 hourly 1 http://wordpress.org/?v=3.2.1 http://www.sauria.com/blog/2009/04/20/dsldevcon-2009/comment-page-1/#comment-16145 Herding Code 45: Larry O’Brien on Domain Specific Languages | Herding Code Wed, 29 Apr 2009 21:36:15 +0000 http://www.sauria.com/blog/2009/04/20/dsldevcon-2009/#comment-16145 [...] Ted Leung’s overview of DSL DevCon  [...] [...] Ted Leung’s overview of DSL DevCon  [...]

]]> http://www.sauria.com/blog/2009/04/20/dsldevcon-2009/comment-page-1/#comment-16125 Zak Jacobson Wed, 22 Apr 2009 22:30:46 +0000 http://www.sauria.com/blog/2009/04/20/dsldevcon-2009/#comment-16125 I was also at this excellent conference. I'd put myself on the business side of the fence although I am a developer. I work for Intuit which uses DSLs for implementing US Federal and State income tax law. Intuit has been using DSLs for 15+ years. The process for developing the DSLs has historically been a lot of up front design, implement for a long time, hand it over to the business and add new features over time. We recently, with Brad Cross (also presented at dsldevcon), tried a new approach of developing the language while using the language. We built an internal DSL in Groovy, paired with a business analyst and produced working software in a couple of weeks. We developed just enough of the language to meet the goals of the pilot project. The process worked extremely well. Of course, we haven't scaled it up yet. The language we developed was very expressive and intuitive for the tax analysts. We were able to iteratively develop language constructs and abstractions to find the sweet spot of readability and terseness. Using a Groovy based internal DSL let us take advantage of existing tools, especially refactoring in IntelliJ, so that we could refactor the language as we went. I was also at this excellent conference. I’d put myself on the business side of the fence although I am a developer. I work for Intuit which uses DSLs for implementing US Federal and State income tax law. Intuit has been using DSLs for 15+ years. The process for developing the DSLs has historically been a lot of up front design, implement for a long time, hand it over to the business and add new features over time. We recently, with Brad Cross (also presented at dsldevcon), tried a new approach of developing the language while using the language. We built an internal DSL in Groovy, paired with a business analyst and produced working software in a couple of weeks. We developed just enough of the language to meet the goals of the pilot project. The process worked extremely well. Of course, we haven’t scaled it up yet. The language we developed was very expressive and intuitive for the tax analysts. We were able to iteratively develop language constructs and abstractions to find the sweet spot of readability and terseness. Using a Groovy based internal DSL let us take advantage of existing tools, especially refactoring in IntelliJ, so that we could refactor the language as we went.

]]>