- From: Curt Arnold <carnold@houston.rr.com>
- Date: Fri, 19 Mar 2004 14:29:13 -0600
- To: www-dom@w3.org
On Mar 19, 2004, at 10:48 AM, Kasimier Buchcik wrote: > > Hi, > > on 3/19/2004 5:17 PM Philippe Le Hegaret wrote: >> On Fri, 2004-03-19 at 10:08, Kasimier Buchcik wrote: >> >>> Hi, >>> >>> there are two tiny questions burning under my fingernails: >>> >>> 1. Does anyone has a definition at hand about what to return if >>> DOMImplementationSource.getDOMImplementation was called with an >>> empty >>> (or just whitespaces containing) "features" argument? >> >> >> I would propose null as a return value. > > Ok, although if no feature was specified one could tend to misinterpret > this in the form of: "no features required, so it will give me any > implementation". A tiny hint in the specs would be helpful here. > Currently, Xerces-J will interpret an empty features list as "no features required" and will return the first available implementation for getDOMImplementation and every available implementation for getDOMImplementationList. I believe that behavior is consistent with the spec. If you start out with a long features list on a call to getDOMImplementationList and start removing features, you will always get the same or larger set of implementations. It would be inconsistent that removing the last feature would clear the list. I've got two tests on tap (one each for getDOMImplementation and getDOMImplementationList) that I'll commit to the test suite unless there are any objections.
Received on Friday, 19 March 2004 15:29:14 UTC