Re: Fallback to w3c-test.org

On 04/24/2014 02:43 PM, James Graham wrote:
> So at the moment the serve.py script tries to do some cleverness and
> fall back to w3c-test.org if the local subdomains don't resolve e.g. it
> will try to use www.w3c-test.org in place of www.web-platform.test if
> the latter doesn't resolve.
>
> Unfortunately this system is totally broken because the ports on the
> local machine and on w3c-test.org don't match at all. To deal with this,
> it would be necessary to assign each domain / subdomain a port either
> from loclhost or from the well-kown ports on the fallback server. This
> would impact the way that substitutions are used e.g.
>
> {{domains[www]}}:{{ports[http][0]}}
>
> wouldn't make any sense because the port would depend on the domain.
> Since tests are using various combinations of host/domain/port, fixing
> this would require a great deal of effort not just to design a
> fallback-aware substitution mechanism, but also to convert all tests to
> use it.
>
> I don't think I have time to do this work (that is; I think there are
> other things I could be doing that are higher value). Does anyone else
> think that the the fallback to w3c-test.org is valuable enough that they
> are prepared to put in the effort to unbreak it? If not I suggest we
> remove this feature and simply require people to set up their
> environment correctly.

This is the first time I hear about this feature, and I'd rather it 
didn't exist at all. It seems rather surprising that fetching something 
on a local domain would silently forward you to w3c-test.org.

HTH
Ms2ger

Received on Thursday, 24 April 2014 16:36:29 UTC