Re: Review of filter tests.

On Tue, 12 Oct 2010 14:07:10 +0200, Tavmjong Bah <tavmjong@free.fr> wrote:

>
> Hi,
>
>  I have reviewed five of the filter test. Four look good, one has a
> problem (filters-offset-02.b.svg). As I don't have access to CVS yet (or
> know where the relevant file is), I'll post the results here.

I've updated the tests based on your comments, see below.

>  I also have one question: Why doesn't the test template pass the W3C
> validator at http://validator.w3.org?

Probably because of the test template, which is in the namespace  
"http://www.w3.org/2000/02/svg/testsuite/description/".

I note that the validator is wrong about <font-face> in <defs>. That's  
perfectly valid markup.

> filters-felem-02-f.svg
>
>  Passes: Firefox 4, Opera,
>  Fails: Batik, Chrome, Inkscape
>
>  Review: OK (But does the polygon (star) really
>  need 15 decimal places of precision?)

Not really no. I cut it down to 3 places, and marked the test as reviewed.

> filters-gauss-02-f.svg:
>
>  Passes: Opera, Chrome, Batik
>  Fails: Inkscape, Firefox 4.
>
>  Review: OK

Marked the test as reviewed.

> filters-gauss-03-f.svg
>
>  Passes: Chrome, Batik
>  Fails: Inkscape, Firefox 4, Opera
>
>  Review: OK

Marked the test as reviewed.

> filters-image-03-f.svg
>
>  Passes: Firefox 4, Opera, Batik
>  Fails: Inkscape, Chrome,
>
>  Review OK except typo(?) 20.4% -> 20.8% in "width specified"
>  (also, unused filter def: "x_specified").
>
>  Here are the details as they are a bit of a pain to figure out:
...

Added the details to the testcase, removed the unused filter definition,  
fixed the typo and marked the test as reviewed.

> filters-offset-02.b.svg
>
>  Passes: Opera
>  Fails: Inkscape, Firefox 4, Chrome, Batik
>
>  Review: Not OK:
>
>    Left most square:
>
>         No primitive filter region defined for feImage.

You meant feOffset, not feImage, correct?

>         According to the spec, the primitive filter region
>         defaults to the smallest subregion for all referenced
>         nodes which in this case is the primitive filter region
>         of feFlood. This region is not large enough to account
>         for the displacement. If the primitive filter region is
>         enlarged, Firefox 4 and Chrome will pass.

It seems slightly clumsy to me that you'd have to specify either of x,y or  
width/height on feOffset to account for the dx,dy offset since offsetting  
filter results is what feOffset is meant for in the first place.

This sounds a bit related to ISSUE-2334, which has a pending edit based on  
http://lists.w3.org/Archives/Public/www-svg/2010Jul/0009.html that I  
haven't got to yet.

Cheers
/Erik

-- 
Erik Dahlstrom, Core Technology Developer, Opera Software
Co-Chair, W3C SVG Working Group
Personal blog: http://my.opera.com/macdev_ed

Received on Tuesday, 12 October 2010 13:14:37 UTC