Introduction: Any programming language is intended to take you so far
to solve real world problems. Object oriented language like C# or JAVA is
directed to create applications that, solves problems using the generic syntax,
allowing one to build components / classes that solves the business problem. DSL
on the other hand empowers to develop a layer through quasi syntax (and its
corresponding semantics) that solves the business domain problems.
Definition: In
essence, DSL is a meta-programming approach (business centered specific
programming on top of conventional programming language) that develops a layer
of domain specific syntax/components which can be, invoked typically over
command line or structured textually through syntax or structured graphically,
and hence interpreted through a custom parser. This facilitates a conventional
programming language to extend and solve a specific business problem at hand.
Essential: There is an urgent
need to build a repository of DSL for any business domain which inherently
abstracts the repeating miniscule problems and allows one to flexibly weave the
DSL components making it powerful, reusable, well tested over time and extensible.
Popular Examples:
>UNIX’s
command line (grep/awk/sed) entities that allow one to solve recurring everyday
programming problems of data manipulation and can be invoked through scripts
and command line.
>HTML is by itself a configuration
(textual/xml) based DSL that allows to solve the rendering hypertext over
browsers.
>CSS is another
meta-programming language that allows specifying styles for its kin HTML.
Challenges: There is clearly an
overhead in building a foundation of DSL components with economical syntax that
greedily solves the business problem at hand.
Note: There are scores of books
devoted for DSL running to 100s of pages. But I have made a sincere attempt to
capture my understanding both from the theory I have read and practical coding
experience.
No comments:
Post a Comment