Re: Additions to SVG 1.1 2nd Edition Conformance Tests

On 2010-03-21 16:28, Chris Lilley wrote:
> JR>  Although the test case then wouldn't show an error condition if a
> JR>  renderer decides to skip the # and draw the horizontal line up to 90;
> JR>  perhaps that error would have best been inserted in the red line instead
> JR>  of the black.
>
> Good point, I have made that change.
>    

Hm, maybe that wasn't such a good point, though. In fact we have two 
distinct failure conditions here:

- The path containing errors isn't rendered at all (which is what the 
test caught so far)
- The error recovery works *too* good and skips errors that should 
terminate parsing and renders more than it should (which is what would 
be caught when the error is introduced in the red path)

So maybe there should be two lines, each of which tests one of the above 
conditions, because a single test won't catch both.

Also, considering that many errors cause an element to be invisible the 
test as it was previously was probably more useful as most renderers not 
conforming to the spec would likely just skip rendering the path instead 
of trying overzealous error recovery.

Regards,
Johannes

Received on Sunday, 21 March 2010 19:20:48 UTC