Re: [css3-transforms] Testing as stand-alone SVGs.

On Thursday, May 10, 2012, 8:27:13 PM, Dirk wrote:

DS> Hi Tav,

DS> On May 10, 2012, at 5:10 PM, "Tavmjong Bah" <tavmjong@free.fr> wrote:


>> Hi,

>>    I see that a lot of work has gone into creating CSS3 Transforms tests
>> which include SVG embedded in either HTML or XHTML.[1] What is missing
>> is the ability to test standalone SVG. I have written a short script
>> which converts the HTML/XHTML based test into standalone SVG tests using
>> the template that the SVG group has approved.[2] I've uploaded an
>> example of a converted file to the SVG WG Mercurial site.[3]


DS> That is really awesome! We planned to port the test once we have enough tests as well.

Yes, it is excellent.

DS> I would not upload the tests at this point of time. We still make
DS> some changes to the spec, that influence the tests as well. Please
DS> hold on on uploading the converted tests till we are in LCWD.

I disagree. Given the conversion is automated, it is no problem to run it multiple times. The advantage of having tests available earlier is that implementors have earlier access to them.

>>    How can these files be included in the test framework that is accessed
>> from the Editor's draft? 

I believe it is simply a matter of linking to the sections of spec being tested, in the tests.

> Two things would be desirable:

>> 1. To be able to tag a browser as passing/failing for a stand-alone SVG
>> in the same way that is done with the HTML/XHTML versions.

Right.

>> 2. The ability to include results from non-browsers. This is important
>> for SVG. Four out of the eight SVG renderers used for the SVG 1.1 second
>> edition Implementation Report are not web browsers.[4]

That can be done; there are two ways to do it. One is to have a text file of passes/fails which is imported by the harness. The other way is to run the tests in the browser, using the options on the startup page to set an appropriate user agent string. The non-browser implementation is run in parallel, and the pass/fail/skip/dunno options are selected based on what the non-browser does (ignoring whether the browser passes or fails).

>>    It appears that the SVG specific tests don't actually test the use of
>> CSS but include the transforms as presentation attributes. Is there a
>> plan to include testing as CSS styling properties?

We need both types of tests (presentation attributes and style rules - both are using (the same) properties). It seems possible to take the presentation attribute tests and autogenerate a set of style rule tests - either with a style attribute, or with generated ids on the elements that had presentation attributes and #id selectors in the stylesheet in the style element.

We also need some tests that have both (this need not be all the tests, just enough to test the feature) with different values, to ensure that the correct specificity is implemented. Presentation attributes have specificity zero.

DS> That are really good questions. I talked with Peter about that
DS> this morning and he needs more help from the maintainer of the SVG
DS> test suite. So the SVG wg should reach out to Peter directly.

DS> Greetings
DS> Dirk

>>                        Tav

>> [1]http://test.csswg.org/suites/css3-transforms/nightly-unstable/html/reftest/
>> [2]http://www.w3.org/Graphics/SVG/WG/wiki/SVG2/Testing_Requirements
>> [3]https://svgwg.org/hg/svg2-tests/file/c5a34c41f233/contributors/tavmjong/incoming/svg-matrix-001.svg
>> [4]http://www.w3.org/Graphics/SVG/Test/20110816/status/implementation_matrix.html







-- 
 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 Tuesday, 22 May 2012 12:33:53 UTC