Re: Testing call this Monday

On Thu, 2011-03-31 at 23:21 +0200, Ms2ger wrote:
> On 03/31/2011 03:01 AM, Philippe Le Hegaret wrote:
> > Folks,
> >
> > It looks like everyone can be on the call this Monday at 9am ET. Let's
> > meet on #testing as usual.
> 
> Not sure who "everyone" is.

Ah, it was meant for Mike, Michael, Francois, myself, but you're welcome
to join us as well, including irc only.

> > On the testing interest group: this idea came out of Jeff to give a
> > "formal" status within W3C, raise the profile of the project, and allow
> > others to feel they're welcome to contribute to the project. Mike is
> > working on a draft charter for that. As usual, the trick is to restrict
> > the scope of the Group so that folks don't think we're going to test
> > everything everywhere. Maybe something along the lines of "HTML5 Testing
> > Interest Group"
> 
> What about "Web Browsers Testing"?

That's a possibility indeed.

> > but we'll still need to differentiate it from the HTML5
> > test suite task force. The goal of the interest group would not be to
> > produce tests in any case. By the way, we would need to find a chair (or
> > co-chairs?) for the IG and it would be nice if we avoid having me on the
> > critical path. On the other end, I have a vested interest in ensuring
> > that the testing project is successful so I'm ok to be the default
> > candidate to chair.
> >
> > On the starting point for the framework: I wonder how long we should
> > take to close the loop on that one. Either we take the current HTML
> > framework and continue it, or we use an other one for the starting
> > point. In any case, I would hate for us starting from scratch and
> > reinventing the wheel. James mentioned that he has one that can run the
> > HTML5 parsers already so maybe we could explore that path as well? Also, we'll
> > need to figure if the framework to be used on desktop browsers will be
> > the same one as the framework to be used on mobile browsers. Again, if
> > we can avoid duplicating efforts, that would be better. It may well be
> > that we start on a common framework and branch it later on.
> 
> Using the testharness.js the HTMLWG uses makes the most sense to me. (As 
> well as reftests [1] as used by the CSSWG for visual tests.)

Ah, by framework here, I meant the layer in charge of running several
test files and collecting the results. testharness.js is for a single
file. Regarding other techniques, we documented a few at:
 http://www.w3.org/wiki/TestInfra/goals#Test_methods

self-describing, testharness.js, and reftests are the immediate targets
for sure.

Philippe

Received on Thursday, 31 March 2011 22:38:12 UTC