Domain-Specific Languages made simpler

draegtun · on Oct 2, 2012

This was also cross posted to blogs.perl.org (http://blogs.perl.org/users/jeffrey_kegler/2012/08/domain-sp...) where comments can be made.

skrebbel · on Oct 2, 2012

I might be missing something but how does "a new parsing algorithm" imply "easier DSLs"? I mean, with stuff like xtext/xpand around and things? You can describe a DSL grammar in a pretty straight forward way and immediately get excellent (Eclipse) IDE support for the grammar and your newly generated spec. And with JetBrains MPS around, and so on?

wonderzombie · on Oct 2, 2012

Yeah, this proposition had me scratching my head, as well. I guess the OP's method is, in some sense, easier, but it doesn't strike me as easy. When I imagine easy, I imagine ad-hoc DSLs in Ruby or (dare I say it) Lisp.

draegtun · on Oct 3, 2012

> When I imagine easy, I imagine ad-hoc DSLs in Ruby or (dare I say it) Lisp.

These are what Martin Fowler calls an internal domain-specific language. These Ruby/Lisp/Perl/Scala/etc embedded DSL's blur the lines between an API and a DSL. What the OP is referring to is a full blown (external) DSL.

p4bl0 · on Oct 2, 2012

For those interested by the topic, you may also want to give a look at Racket[1]. See for instance this ACM Queue article[2] by Matthew Flatt.

[1] http://racket-lang.org/

[2] http://queue.acm.org/detail.cfm?id=2068896