- From: Robert Clary <bclary@netscape.com>
- Date: Sat, 30 Nov 2002 16:07:30 -0500
- To: Philippe Le Hegaret <plh@w3.org>
- CC: Dimitris Dimitriadis <dimitris@ontologicon.com>, www-dom-ts@w3.org
Philippe Le Hegaret wrote: > > I'm wondering if we should have more than one test suite for ECMAScript: > One for HTML, one for XHTML, and one for XML. It seems like by trying to > mix all of them, we run into unnecessary troubles. I believe that the issue with the current design has to do with the inability to select the content type of the test document which is served as well as the inability to choose the implementation that is being tested. In the current approach we have lost that ability and hide the differences between implementations in handling content types (for example Internet Explorer's native handling of XML versus MSXML's handling of XML). This is not necessarily related to the version of JsUnit that is being used but more to do with the implementation of the builder objects and how test documents are chosen and loaded. In the straw man proposal I made in late May, I proposed an approach where the content type and implementation could be chosen. I believe that the ability to specify the content type and the implementation is crucial to a complete and fair test. In addition, several tests make assumptions with regard the ability of the parser being used (validating vs non-validating etc.). The current tests do not make it clear whether a particular test may or may not be supported by the particular implementation being tested. Finally, the current test framework skips certain tests which are not appropriate for the current test environment but do not provide a report of what tests were skipped or why. The test framework should report on any such skipped tests. /bc
Received on Saturday, 30 November 2002 16:08:17 UTC