[Draft] EARL Requirements
The ERT WG is chartered to produce a requirements document for EARL. A requirements document will help the group focus on a clear target in order to develop EARL as quickly as possible. The following is not a complete list of requirements but much more an intial attempt to capture the essential contents for a requirements document.
Target Audience
- Evaluation tool developers who want to output EARL
- Tool developers who want to interpret EARL
- Specification developers who want to incorporate EARL into their work (e.g. DCMI, IMS, QA, ...).
Deliverables
- EARL Reference Manual
- Detailed description of the EARL RDF schema, the classes, and the properties.
- EARL Primer
- Introductory document with many examples and guidance for people new to EARL and RDF.
Scope
- Persistence of test results
- When structural components that are not context sensitive change location (without actually changing), EARL results should remain valid.
- Example: informational image along with a longdesc move to a different location within a Web page, the longdesc tests should remain valid.
- Example: an illustrative image moved from @@@
- Example: security requirement does not allow the photograph and the name of an individual to be published beside each other, but the two are publishable otherwise.
- Example: katie will send example about multiple levels of security access. @@@
- Example: content-negotiation @@@
- Example: dynamic pages @@@
- Can we describe the context changes for which a given test is robust?
- Locating precisely the subject of the result
- EARL needs to provide sufficient information about the location in the content to which the results apply.
- Example: which of the 5 images on a page are not accessible?
- Example: header element of a table missing, should the error message point to the start of the table vs a specific cell?
- Example: querying a repository of test results for results to a specific area of the content (e.g. main content area in a portal).
- Test descriptions
- EARL only describes test results, not test descriptions or processes. However, in the spec we need to explain examples of how different tests can be related to each other.
Implementations
For each feature we need to have:
- code examples
- test suites (???)
Implementation Tracking
Maintain a list of tools which support EARL.
Issue Tracking
Bug tracking and issues list.
Last updated on $Date: 2005/03/07 15:29:49 $ by $Author: shadi $