SL33: Using Well-Formed XAML to Define a Silverlight User Interface

XAML and XML

XAML is based on XML, and shares many of its language features. Some of the language features that are directly relevant to the stated intent of SC4.1.1 and to 4.1.1 related techniques include:

• Well-formedness: The definition of well-formed XAML is the same as the XML definition. XAML processors, including the Silverlight runtime XAML parser, will block loading XAML that is not well formed.

• Duplicate attributes: Unless specially configured for scenarios such as design-time support, XAML processors will block loading XAML where elements contain duplicate attributes.

• Quote matching: mismatched quote matching for attribute values in XAML constitutes XAML that is not well formed.

Some XAML language features that are analogous to XML but have some technology-specific differences include:

• Identifiers: XAML defines a Name directive, which is analogous to xml:id in that Name serves as the unique identifier of an element. However, XAML defines an additional concept of a XAML namescope, which permits a XAML document to contain multiple XAML namescopes as a factoring technique. Thus, identical Name values are permitted in a XAML document so long as each is defined in a separate XAML namescope. XAML namescopes are associated with elements, such that the extent of each XAML namescope is understood by XAML processors.

• Schemas and vocabularies: A notable difference between XAML and XML is that a XAML vocabulary is not typically represented in existing XML schema definition formats such as XSD or DTD. XAML includes inheritance and reference features that cannot adequately be expressed in XSD or other existing XML schema representation formats. This affects the "elements are nested according to their specifications" consideration of SC4.1.1. XAML definitely has the ability to enforce nesting restrictions as represented by a XAML vocabulary. However, XAML validity for a vocabulary is deliberately fluid, in order to support extension by user code. XAML validity is determined by a combination of a XAML processor, a XAML concept known as a XAML schema context, and the code that backs the XAML and defines any objects being instantiated as a parsing result. Typically, design time tools such as Microsoft Visual Studio can adequately duplicate the runtime validity characteristics of a XAML vocabulary. Using these tools, application authors can both verify XAML validity as well as receive design-time information for how to correct any XAML validity errors.

XAML parsing and HTML parsing

In the Silverlight implementation, XAML is like HTML in that it is loaded and parsed just-in-time. Silverlight XAML is not precompiled to binary or MSIL (the language-neutral CLR runtime format). Instead,Silverlight XAML is transmitted or stored as plain text, either loose or packaged as resources in a library. Thus Silverlight XAML is human readable as well as machine readable.

However, unlike HTML, Silverlight XAML is only intended to be loaded and interpreted by the Silverlight runtime, rather than multiple possible user agents that each implement an HTML engine. HTML is a language where the behavior is also specified. In contrast, XAML is a language for referencing constructs that are defined in runtime libraries, and the functional specification of the XAML language itself is minimal (intrinsics; language rules; primitive types). Layout, appearance, type-member sets, roles, etc. are all left up to specific frameworks and vocabularies that use XAML. Behavior associated with a given XAML construct is based on type definitions made in a runtime library. For Silverlight XAML, the types are from Silverlight core libraries, but often the definitions come from libraries that are available to the Silverlight runtime as part of an application's packaging for distribution.

XAML is generally speaking strict, and will raise parsing errors if XAML contains elements that are not recognized. Such parsing errors generally present the information in the XAML from resulting in any objects being created, which in turn prevents a Silverlight application from running. This is different from typical (non-xHTML) HTML, where implementations are permitted to contain nonrecognized elements or attributes and ignore them.