Re: Corrections to the Implemention Reports

On Oct 27, 2010, at 11:48 AM, John Jansen wrote:

> I suppose this should be directed to Peter:
>
> When we discover there is an error in an Implementation Report, what  
> is the best course of action to get that error fixed?
>
> Specifically, in the IE implementation report, the report is based  
> on our public Beta. In the public Beta we have a regression that  
> causes IE to fail the font-family-name-* tests. IE passed these  
> tests in builds leading up to the beta, and on internal builds since  
> the beta, but fail them explicitly on the Beta build.
>
> Should we re-publish our IR based solely on the Beta build pass/ 
> fail?  If so, is it OK to wait until after the next publish to the  
> Test suite, so we can include any new pass/fail data as well?

If the currently published IR doesn't match actual results from the  
public beta then I suggest you publish an updated IR asap (replace the  
one on w3.org and send a note to www-style and this list when it's  
updated).

You can publish a new IR against the most recent version of the test  
suite but it can only be based on publicly available builds.

I'll be adding the new version of the test suite to the harness today  
or tomorrow. As the harness stores results logged against a particular  
version of the tests, I'll be copying over all existing results for  
tests that didn't change. So a new IR really only needs to cover  
modified tests.

Peter