- From: François Daoust via GitHub <sysbot+gh@w3.org>
- Date: Fri, 03 Jun 2016 09:41:14 +0000
- To: public-secondscreen@w3.org
> Should we define "feature"? Or is that defined implicitly by the test suite for each conformance class? Most specs leave that implicit. I could not find a generic definition of "feature" in the specs I had a look at. In some cases, groups have gone one step further and actually list the features, c.f. the WebMention spec for a recent example: http://www.w3.org/TR/2016/CR-webmention-20160524/#feature However, it may not be easy to apply that to our case, because our algorithms have more paths and conditions to check. We could start with the list of algorithms: ``` For the purposes of evaluating exit criteria, each of the following is considered a feature: - For controlling user agents: - Constructing a PresentationRequest - Starting a presentation - Reconnecting to a presentation - Getting the presentation displays availability information - Monitoring the list of available presentation displays - Establishing a presentation connection - Sending a message through PresentationConnection - Receiving a message through PresentationConnection - Closing a PresentationConnection - Terminating a presentation in a controlling browsing context - Handling a termination confirmation in a controlling user agent - For receiving user agents: - Creating a receiving browsing context - Monitoring incoming presentation connections - Sending a message through PresentationConnection - Receiving a message through PresentationConnection - Terminating a presentation in a receiving browsing context ``` However, the granularity does not look good enough. For instance, for "Starting a presentation", we would probably want to be much more precise, e.g.: - Starting a presentation in the absence of user interaction - Starting a presentation from a non secure context - Starting a presentation from within an iframe - Starting a presentation when a presentation is already starting. - etc. That's a bit tedious though, we would end up with the list of test cases in the test suite (and we have not built that list yet). Feel free to add a generic some wording or build on the above text. Personally, I would leave the definition of feature implicit :) > If (for example) Chrome implemented the 1-UA receiving user agent and Mozilla implemented the 2-UA receiving user agent, that would fulfill the criteria of having two interoperable and independent implementations of the receiving user agent conformance class, correct? Correct. > Editing comment: Should the first sentence of each definition be capitalized? :) They should :) I updated the PR accordingly. -- GitHub Notification of comment by tidoust Please view or discuss this issue at https://github.com/w3c/presentation-api/pull/314#issuecomment-223536159 using your GitHub account
Received on Friday, 3 June 2016 09:41:15 UTC