- From: Dean Jackson <dean@w3.org>
- Date: Fri, 6 Aug 2004 23:36:43 +1000
- To: Jonathan Watt <jonathan.watt@strath.ac.uk>
- Cc: www-svg@w3.org
On Thu 05 Aug 2004, Jonathan Watt wrote: > I'm looking for WG clarification on why SVG_WRONG_TYPE_ERR isn't > required to be thrown by all DOM methods that require arguments > to provide an SVG interface when there is no sensible action to take if > the argument doesn't support the interface. I specifically want to know > what should happen when createSVGTransformFromMatrix is passed an object > that is not an SVGMatrix. Should SVG_WRONG_TYPE_ERR be thrown? Should an > implementation defined exception be thrown - in which case why not an > SVG_WRONG_TYPE_ERR? Or should the identity matrix be used and no error > reported? Or is there some other course of action that makes sense? I'm not the expert on this but it seems to me as if the SVG_WRONG_TYPE_ERR should be thrown in these cases. I don't like the idea of default values if you pass in dodgy objects. What do you think? If you agree, I'll propose this as an errata. > In the case of createSVGTransformFromMatrix there is the possibility of > using the identity matrix, although it would seem bad not to notify the > user of the error to help them debug, but for some methods there is no > real course of action. For example, in SVGSVGElement, what should happen > for getIntersectionList or getEnclosureList if rect is not an SVGRect or > if referenceElement isn't an SVGElement or null? Hmmm... in the DOM subset implement for JSR 226, we throw NOT_SUPPORTED_ERROR when you pass null into a getter, and INVALID_ACCESS_ERR when used in a setter (this is for trait access, and in an API where we want to keep the total number of classes to a minimum). I'll ask you, what do you think "null" should do? I'd suggest SVG_WRONG_TYPE again. Dean
Received on Friday, 6 August 2004 09:37:16 UTC