On Jan 10, 2006, at 5:16 PM, Jon Ferraiolo wrote: > Eric, > > I agree with your analysis, particularly with respect to the > transform attribute. (And if other implementers agree with Maciej, > for other attributes as well.) The spec already appears to call for completely ignoring invalid attribute values, so I'm satisfied with the specified behavior. I'll let other implementors speak for themselves. > The SVG WG should have a test for the ‘tranform’ attribute for > incorrect entries such as the example you provide. It is too much > to ask for tests for every possible incorrect formulation of every > attribute, but the example you give below represents a case which > is more likely due to accidental error, such as misspelling > ‘translate’. Clearly it is impractical to test every possible invalid value. But you can start with things that seem especially likely to be handled wrong, and build up the test suite over time based on actual bugs found in impelementations. Other working groups apparently take a similar approach. Regards, Maciej
Received on Wednesday, 11 January 2006 01:39:41 UTC