Re: Initial resultsfor WOFF user agent testing

On Tuesday, September 13, 2011, 8:44:27 AM, Jonathan wrote:

JK> Thanks for assembling this, Chris.

JK> WRT the metadata-related tests, it's understood that a UA can
JK> "pass" from a spec conformance POV if it simply does nothing with the metadata.

That is correct, and is how the metadata* tests are worded currently. For tests with valid metadata, they say:

The Extended Metadata Block is valid and may be displayed to the user upon request.


For tests with invalid metadata, they say:

If the UA does not display WOFF metadata, the test passes if the word PASS appears below.
  The Extended Metadata Block is not valid and must not be displayed. If the UA does display it, the test fails.

This means that, for all the metadata* tests, implementations that don't display the metadata all pass,as long as they implement the "metadata must not affect font display" condition (which is separately tested.

JK> However, I wonder if it would be more helpful if the results made
JK> a distinction between "pass by ignoring" and "pass by appropriate
JK> display", so that we have a better picture of what is actually being implemented and tested.

Yes. I think a better criterion for metadata tests would be:

If the UA does not display WOFF metadata, skip the test

and then either 

The Extended Metadata Block is valid. If the UA displays WOFF metadata, the test passes if it is displayed.

or

The Extended Metadata Block is invalid. If the UA displays WOFF metadata, the test fails if it is displayed.

The PASS FAIL font display would be removed from those tests (the font still needs to be used, it could say something else though.




-- 
 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, 13 September 2011 08:42:26 UTC