- From: Rebecca Hauck <rhauck@adobe.com>
- Date: Thu, 19 Apr 2012 09:27:02 -0700
- To: Chris Lilley <chris@w3.org>
- CC: Dirk Schulze <dschulze@adobe.com>, "public-css-testsuite@w3.org" <public-css-testsuite@w3.org>
Great. Thanks, Chris. And I did encounter the problem you called out in one of the other tests where the translation was incorrect but enough to cover up the 50,50 rect so I went ahead an made the red failure rects the same size and the test rects. Adding 1px margin for anti-aliasing is a good idea, too - I'll incorporate that as well. Cheers, -Rebecca On 4/19/12 7:03 AM, "Chris Lilley" <chris@w3.org> wrote: >On Wednesday, April 18, 2012, 11:27:02 PM, Rebecca wrote: > >RH> Hi Chris, > >RH> Yes, thanks for the feedback and I'm also in agreement. However, I'm >RH> proposing the following change, where I eliminate the coordinates >from the >RH> description altogether and add a second visual element that would a >cause >RH> a diff if the test fails. This is similar to your suggested >improvement >RH> to the scale-XXX tests, where the pass/fail result is affected or >defined >RH> by a fixed visual element on the page. > >Great. My main concern was that careful testers would need to screenshot >and count pixels, while slapdash ones might give false positives. > >RH> Here's a sample code snippet: > >RH> <p>The test passes if there is a green square and no red.</p> >RH> <svg> >RH> <rect x="100" y="100" width="50" height="50" fill="red"/> >RH> <rect width="100" height="100" fill="green" >RH> transform="translate(50 50)"/> >RH> </svg> > > >RH> I've attached pngs demonstrating PASS & FAIL for this test. Would an >RH> approach like this be ok? > >The general approach would satisfy my comment. > >For specific tests, though, its good to ensure that inaccuracies are >visible. To take your code snippet above, the red rect is only 50x50 so >if the implementation ends up with any transform from translate(25,25) to >translate(75,75) the red rect is still covered. > >(Unlikely with that specific transform syntax but could happen if other >units are used or for more complex, combined transforms). > >Allowing a 1px margin for anti-aliasing and thus specifying a 98x98 red >rect would be more sensitive to errors in implementations. > > >RH> Thanks, >RH> -Rebecca > > > >RH> On 4/18/12 8:46 AM, "Dirk Schulze" <dschulze@adobe.com> wrote: > >>>I agree with your comments and we are already improving the tests, but >>>push when we are ready. So far translate and scale tests are affected by >>>the refactoring. > >>>Greetings >>>Dirk > >>>On Apr 18, 2012, at 8:40 AM, "Chris Lilley" <chris@w3.org> wrote: > >>>> Hello public-css-testsuite, > >>>> "The test passes if there is an orange square at 50,50" > >>>>http://test.csswg.org/harness/test/CSS3-TRANSFORMS_DEV/svg-translate-00 >>>>1/ > >>>> Easier to see if there was a marker at 50,50 to which the top left >>>>corner of the square could align. > >>>> Also "at 50,50" could be made more precise "whose top left corner is >>>>at >>>>50,50". > >>>> -- >>>> Chris Lilley Technical Director, Interaction Domain >>>> W3C Graphics Activity Lead, Fonts Activity Lead >>>> Co-Chair, W3C Hypertext CG >>>> Member, CSS, WebFonts, SVG Working Groups > > > > > > > > >-- > Chris Lilley Technical Director, Interaction Domain > W3C Graphics Activity Lead, Fonts Activity Lead > Co-Chair, W3C Hypertext CG > Member, CSS, WebFonts, SVG Working Groups >
Received on Thursday, 19 April 2012 20:41:25 UTC