RE: spec development process was: vendor prefixing

[Brian Manthos:]
> 
> > > Why is it not until CR that "test suite" comes up?
> > > Shouldn't the test suite be underway during WD or ED stages?
> > >
> > > For example, why isn't every WD accompanied by a test suite that
> > > consists of (at least) every Example from the WD draft?
> >
> > Writing testcases earlier is better - having a test editor/owner aims
> > to enable that, in part - but since our public testsuites can't use
> > prefixes and pre-CR implementations must use prefixes the practical
> > utility of going through the exercise of publishing testcases that
> > can't yet run anywhere is somewhat limited.
> 
> Isn't the issue of "test cases unprefixed" vs. "tested implementations
> prefixed" already addressed as part of the implementation report process
> that happens in CR phase?
> 
> Why can't that same process be applied during WD?
No one is saying it can't be. But since WDs don't have to be backward-compatible
with previous WDs maintaining testcases is a challenge, never mind cross-browser
testcases. So while I agree that waiting for CR in order to write testcases is 
not ideal and getting an earlier start is good - hence, in part, the incentive 
of requiring an IR from unprefixed implementations of a CR - the practical cost 
of handling multiple colliding moving parts could outweigh the expected benefits.

Received on Monday, 12 December 2011 17:19:13 UTC