Layered User Interfaces Incubator Group Charter

Scope

There is plenty of market potential for tools and services that assist people with developing and maintaining user interfaces for business processes, especially in the context of the roll out of Service-Oriented Architecture (SOA) within an enterprise. Declarative approaches based upon XML offer the following benefits compared to traditional application development models:

• reduced costs for development and maintenance throughout the application lifecycle
• improved security, accessibility and usability
• easier delivery to a wide range of devices and platforms including HTML, SVG, Flash, Silverlight, .NET and Java
• simplifying the provision of mobile access to business processes
• enabling people outside of the IT department to develop applications without the need for extensive programming skills
• facilitating people with different roles to work on different aspects of the development process, e.g. task models, domain models, user interaction, and graphical look and feel
• easier integration with SOA and enterprise knowledge management systems (business ontologies)
• easier transfer of application maintenance responsibilities from one organization to another
• reduced effort for compliance with requirements for adoption of new regulations and auditing procedures (e.g. Sarbanes-Oxley and Basel II)

A prerequisite for this is a means to separate out different concerns. This can be achieved through the adoption of a layered architecture:

1. Application task models, data and meta-data
2. Abstract UI (device independent)
3. Concrete UI (device dependent)
4. Physical realisation for specific devices

Each layer can be considered as the result of a transformation of the layer immediately above it. Each layer embodies a model of behavior at a progressively finer level of detail. Transformations have access to the delivery context, which models user preferences, device capabilities and environmental conditions. The model view controller pattern is used to cleanly separate the user interface, the dialog behavior and the data it operates on.

All of the layers (with the possible exception of the physical realisation) can be represented in XML, with the mappings between the layers described using XSLT or other related techniques (e.g. QVT and ATL). The mapping to the physical realisation can be implemented via an optional mapping to a low level markup language followed by a compilation step. This enables developers to concentrate on what the application needs to do, rather than the details of how that is to be achieved on a particular platform. The approach is intrinsically extensible, and skilled programmers can add support for new markup and associated implementation classes as needed for new kinds of controls.

High level development suites can be provided to shield developers from the underlying XML representations. For example, a data model could be manipulated as a diagram, while the user interface could be defined via drag and drop operations together with editing values in property sheets. The development suite is responsible for maintaining the mappings between layers and verifying their consistency. Developers can choose to provide alternative transformations as needed to address different groupings of delivery contexts. This will facilitate the mobelisation of business processes. For business people on the move, mobile devices can be a lot more convenient than having to pull out and power up a notebook computer.

In Scope

The Layered UI XG will focus on use cases, requirements and proposals for the Concrete UI layer and a declarative treatment of event handlers and task models. This work will seek to combine XForms with XML markup for concrete user interfaces with a range of layout policies and UI controls as appropriate to the use cases and target platforms. This will be aligned to the WAI-ARIA ontology of roles, states, and properties, as a basis for supporting assistive technology.

The work on task models will consider end-to-end behavior, including both client and server-side aspects, for example:

• mapping URIs to user interfaces as appropriate to the delivery context
• event driven models of business processes
• the role of human agents in workflows
• the role of subcomponents in user interfaces

Form submission could be treated as equivalent to generating an event and handled via server-side state transition models that in turn generate new events. Such events could be targeted at named queues identified by URIs. Web pages would be generated by event handlers in response to client requests. The XG will compile a taxonomy of use cases as a basis for collecting requirements. On the client side, the XG will consider the potential for using task models to describe the role of different parts of the user interface, including dynamic bindings to server-side data, and the means to respond to external events.

A study will be made on the potential of State Chart XML (SCXML) for executable models of business processes, along with XQuery and XSLT for database access and report generation. A gap analysis will be conducted to identify opportunities for further work, including the potential for using business rule languages for event handlers in state models, and treating input and output queues as objects in their own right for use in work flows.

This work could include a study of the role of diagrams as a means to support high level authoring tools. XML would allow for round tripping diagrams through authoring tools and subsequent extraction of machine manipulable semantics. The diagrams could be used as a basis for organizing access to different categories of information throughout the application lifecycle from initial requirements capture to withdrawal from service. Relevant standards include UML and BPMN.

The XG may also want to review approaches based upon orchestration or choreography of messages. Choreography focuses on the views of individual participants involved in an exchange of messages, whilst orchestration models encompass all parties, giving a global view of the system. Relevant standards include BPEL and WS-CDL.

Out of Scope

Themes define the detailed presentation and behavior for physical realisations of the Concrete UI. The mechanisms for "theming" a Concrete UI are out of scope for this XG, although it should be possible to identify which theme is to be applied. Note that the use of graphics (including SVG) as part of the Concrete UI remains in scope.

Voice and multimodal interfaces are out of scope, but could be the subject for a follow on XG.

Success Criteria

The XG will be judged successful if its work is picked up in part or in whole by W3C Working Groups, such as the Ubiquitous Web Applications Working Group, whose charter is due for renewal in early 2009. In addition, it is hoped that the XG will be successful in highlighting the role of XML for authoring as distinct from markup intended for direct interpretation by client platforms.