Expert to Expert: Martin Fowler and Chris Sells - Perspectives on Domain Specific Languages

Play Expert to Expert: Martin Fowler and Chris Sells - Perspectives on Domain Specific Languages

The Discussion

  • User profile image


    Around 15:58 Martin says something about being at Intentional and seeing things that made Oslo look kind of 19th century.  When pressed for details, he couldn't pluck a named feature out of the air. I wonder if the interaction between the man and the machine seemed magical?  In the words of ACC, "Any sufficiently advanced technology is indistinguishable from magic." And magic is something that has to be seen to be recognized, not unlike indecency. 


  • User profile image
    Ryan Riley

    Interesting interview, especially the piece about Excel as a (sort-of) language workbench. I'm gathering that many in the DSL forefront see the future of programming as writing DSLs for specific platforms (i.e. Oslo), frameworks (i.e. Rails), or the actual language used for writing the code in a specific business domain (i.e. DDD). I find that an interesting proposition, though I can only imagine the number of DSLs we'll see as most programmers (or at least I do) like digging into the nuts and bolts and making things fit their own mindset.

  • User profile image

    Martin Fowler, on C9?

    Bloody awesome!

    We want more Smiley

  • User profile image

    My criticism of Oslo would be that it focuses too much on data and not enough on behavior.  I think that's what makes it "19th century". 

    On the other hand, a focus on data sells, especially in the Microsoft space.  You just have to be prepared to be snubbed by the "intellectuals", who admittedly don't represent a huge market segment.

    Another problem I have with Oslo is its use of a database as an artifact repository; something like TFS with version control would seem to be more of a fit.  A question: why don't you tell us to put our code in databases?  Is the DSL any different?

  • User profile image
    Eric Giles

    It is my understanding that the DSLs, textual information, etc. are just text and therefore are more than suited to being housed within something like TFS. I think that the purpose of the repository being within SQL Server is because the creation of the structure is only half of the story. The other half, is the runtime consumption of the information stored witin that repository and even updating, inserting information into the database. It is probably more of a mental leap to think in terms of moving from the model being a design-time artifact to a runtime one in the form of model-supported or model-driven applications.


  • User profile image

    Hmm, I want more!  Expressionless

Add Your 2 Cents