Steve Nelson, XML, and Why DSLs Should Be a Required Subject

An XML file can be seen as a Domain Specific Language for describing the solution to a problem within a constrained solution space. You can use XML to describe anything from controller functions to workflows to bean dependencies and configuration information. If you looked at the DTD, Schema or Relax NG description of the grammar, that would describe the underlying (abstract) grammar of any given XML document.

An important distinction here is between concrete and abstract grammars - and stay with me because this is actually quite useful. An abstract grammar describes the kinds of conceptual structures that a given language allows (for most languages you also have to add semantic constraints as often there will be syntactically valid statements that have no useful meaning). A concrete grammar is a particular way of writing an expression that conforms to that abstract grammar. So, for example if you have a DSL for describing workflows, I could write that using CF Script, a text file using indents and spaces for meaning, in XML or I could draw a diagram and these all might potentially be different valid concrete syntaxes to describe the same abstract grammar - to put it simply, they would all be valid, transformable ways of saying the same thing using different characters (or shapes) to denote the same underlying intent.

I have no idea what Steve is going to be saying at Frameworks, but one thing I will say is that a set of calls against an API can be thought of as a different concrete syntax which could be used in place of XML. For a certain subset of abstract grammars, an API can be a very appropriate concrete syntax to use, but you are still having to learn the same vocabulary (the abstract grammar and any associated semantic constraints of your DSL) to call your CFC or base class or API as you would be to write an XML file.

There are definitely strengths and weaknesses (as well as personal preferences) when choosing the most appropriate concrete grammar for a given DSL, but whether you use CFCs or XML, the language still needs to be defined and learnt. So the quote Jeff pulled:

In my opinion, using XML files to create a new framework programming language has more drawbacks than benefits. The biggest drawback is that the developer needs to learn yet another series of tags and syntax and keywords in order to use a framework

to me seems short of the mark (although I could be misunderstanding Steve completely). To me, there is a huge benefit in the expressiveness of a well designed DSL (whether you provide it as a CFC with a set of methods, an XML document, a custom textual syntax or a UML diagram). It reduces the number of characters required to express intent, cuts down on the likelihood of syntax errors, keeps your code DRYer and provides a host of other benefits. But whether you want to implement a DSL using an API or XML, the developer will need to learn your abstract grammar. I guess you could argue there is a minor benefit in not having to learn the concrete syntax of XML (although how hard are angle brackets to get?!), but just as you need to learn the tags and attributes for an XML concrete syntax, you need to learn the exact same concepts expressed using methods and properties if you implement the DSL using an API (probably a CFC) as your concrete syntax, so you can't beat up on XML for that.

I will be sure to attend Steve's talk next week to see what he is talking about and will post back here as I learn more!