Child pages
  • Declarative Configuration

Documentation for a historical release of Infusion: 1.4
Please view the Infusion Documentation site for the latest documentation, or the Infusion 1.3. Documentation for the previous release.
If you're looking for Fluid Project coordination, design, communication, etc, try the Fluid Project Wiki.

Skip to end of metadata
Go to start of metadata

Less code is better. If you can represent aspects of your application as data rather than as imperative logic, there is a clear benefit to application design: data is transparent, and is easier to operate on and transform with algorithms. It's also easier to understand by both code in the application and by humans reading the design. In order to understand a data-oriented component, one merely needs to understand the layout of the data it works with, rather than looking at its implementation or understanding all the details of the contract behind its API documentation.

Standard conventions also help simplify your application architecture. Wherever possible, Infusion applies a standard layout to commonly appearing options for a component. For example, there is a common convention for declaring a component's events, selectors which bind to markup for both styling as well as function, and relationship to its model. Each component is also supplied with a set of defaults which also take the form of declaratively-specified model data.

The Infusion Renderer takes the concept of declarative configuration to a level not achieved with other frameworks. The renderer reduces the entire binding and controller function of an application to a declarative form, a component tree. In this form, the user interface is extremely amenable to inspection, interpretation and re-processing.

Historically, the desire to be able to treat logic as data has strong roots, for example in the LISP community. However, where all application code is on a common footing, designs become tangled and hard to interpret. By providing domain-specific forms for carefully selected parts of an application's functionality, typically at the Controller level, the complexity of code operating on this data can be reduced and transparency increased. It is a productive middle ground, between all application code becoming a candidate to be data (as in LISP), and none of it (as in Java).

  • No labels