The purpose of the DOM Binder is to relax the coupling between a component and its markup. Whilst the Fluid framework is built on jQuery, and uses that framework's selector engine throughout to perform queries on the DOM, the extra level of indirection provided by the DOM Binder allows a complete separation of concerns. Fluid components are parameterised by a set of named selectors, managed by the DOM Binder instance attached to each top-level component. In this way, explicit selectors never appear in component code - leaving the components free of any baked dependence on markup structure.
There are other benefits to having DOM searches (via jQuery) managed in a central location - component authors can get access to fine-grained control over caching and lifetime of search results, which might otherwise become expensive if performed repeatedly - along with its basic
locate() method, each DOM binder has a
fastLocate() method which will not perform a DOM search if there is a cached result.
Components access elements in the DOM through unique selector names. The component defines default values for the selectors, but implementors are free to override the defaults if they need to change the structure of the markup.
Each standard Fluid component has a DOM binder created automatically as a result of its call to the standard framework function
initView. This call takes a set of options from the member
selectors from the component's top-level options, and uses them to initialise the DOM binder.
Component developers declare the component's selectors in the defaults for the component:
When the View is initialised with
fluid.initView(), the DOM binder is created and attached to the top-level that(the component itself) as the member named
dom. In addition, or convenience, the DOM Binder's
locate() function is added as a top-level instance memember on the View. For example, to get a named element, you can simply call
that.locate(name)) (see below for more information).
The other crucial configuration passed to the DOM binder is the overall container for the component - by convention, this is passed as the first argument to the component's constructor function (recall that the standard signature for a fluid component is
fluid.componentName(container, options). Unless they are otherwise qualified, all searches performed by the DOM binder attached to a particular component will be automatically scoped to its own container.
The component must use the DOM Binder for any access to the DOM elements that make up the component, using the
(For information about parameters, for this and other DOM Binder functions, see DOM Binder API.)
The other methods on the DOM Binder are less frequently used, and are not attached to the top-level that in the way that
locate() is. They need to be accessed through the DOM Binder's own object, which by
initView is available as
The signature and function of
fastLocate are exactly the same as that for
locate. The difference is that, if the results of the search are already present in the DOM binder's cache, they will be returned directly without searching the DOM again. This can be very much more rapid, but runs the risk of returning stale results.
The DOM binder's cache is populated for a query, whenever a query is submitted via
clear() method completely clears the cache for the DOM binder for all queries. It should be used whenever, for example, the container's markup is replaced completely, or otherwise is known to change in a wholesale way.
refresh() method refreshes the cache for one or more selector names, ready for subsequent calls to
fastLocate(). It functions exactly as for a call to
locate() except that
refreshrather than just a single one.
The Inline Edit component requires three parts in its user interface:
The component declares selector names for these elements, and provides defaults, in its call to
Here, the default selectors use class names. To use these defaults, an implementer can simply attach these class names to the relevant elements in mark-up. Alternatively, the implementer may choose to override some or all of these selectors with other selectors. For example:
In this example, the implementer is using element IDs to identify the text and edit fields. Because a custom
editContainer is not included, it will default to the class declared by the component.
To access the DOM elements, the Inline Edit component uses its DOM Binder and the selector names:
In this way, the Inline Edit component is completely ignorant of the mark-up it is working with, or even the selectors used. When the markup changes, code doesn't break.