Re: Top-level HTTPS tests?

On 24/02/15 11:14, Mike West wrote:
> In https://lists.w3.org/Archives/Public/public-test-infra/2015JanMar/0000.html,
> James noted that it is now possible to load subresources over HTTPS,
> which is ever so excellent. That said, a number of features (Service
> Workers, Mixed Content, etc) require the test itself to be run under
> HTTPS. I'd like to start putting together test suites for some WebAppSec
> specs, but that's proving to be a difficult task.
> 
> In Blink, we've settled on a "*.https.html" naming convention for such
> tests which ensures that our test runner opens the test over HTTPS. That
> is, any test with `.https.` in the filename will point to an HTTPS URL,
> and not to an HTTP URL. Would such a system work for web-platform-tests?
> 
> Note that we don't have a ton of such tests in Blink, so I'm happy to
> change our behavior to align with whatever works for WPT and other user
> agents. The filename-based solution is pretty simple and
> straightforward, but we're flexible. :)

I think this makes sense; it matches the convention we use for other
kinds of metadata (e.g. .headers., .worker.) and is somewhat simpler to
write than <meta name=https content=true> (I frequently get requests for
tests to be faster to write). If anyone has a good argument for adopting
a different convention, now would be a great time to speak up!

Received on Tuesday, 24 February 2015 16:28:18 UTC