RE: UI Guidelines?

>-----Original Message-----
>From: SULLIVAN, BRYAN L [mailto:bs3131@att.com]
>
>Anant,
>
>I note that DAP has been down this road though before, unsuccessfully. The
>permissions model, if more than single-shot allow/deny, gets into
>complicated areas of defining a session or managing persistent permissions.
>And there is no guarantee that the user who gave permission is the user
>whose streams are being captured later. In BONDI/WAC, we developed what I
>think was a simple and flexible model for procedural API permissions (even
>given the limitations noted above), but DAP failed to reach consensus on
>that approach, and as a result DAP shifted to the declarative/markup API
>approach you see in HTML Media Capture. Having started down that path,
>isn't the declarative approach an option here, and being user-intentional,
>doesn't it provide adequate consent verification? Determining how
>getUserMedia would be accessible only through markup (e.g. through
>"onclick" or as an input type) would need to be decided. But I think it
>might avoid the UI/permissions-definition issues that a prompt approach
>carries.
>
>I do think that getting specific about how the chrome should show what
>devices are in use is a good discussion to have, but this also gets close
>to differentiation needs for browsers.
>
>So overall I would be less worried that the UX would vary significantly
>across browsers (real experience might result in convergence on the one
>that works best, or we might find that the multiple approaches do not
>confuse users after all), as compared to spending significant time going
>down the prompt system rabbit hole.
>
>But  if we want to explore this again, wire frame illustrations at least
>would help clarify the depth of detail and choice that you are recommending
>we define. It might turn out simpler that what we had considered before.


I think that mandating a particular UI in an API spec is not appropriate. However,
the spec may certainly contain non-normative sections on UI suggestions that support
the security concepts described in the spec. (This is similar to how HTML5 doesn't mandate a particular UI for certain form elements, but does have suggestions on what they could look like.)


>Thanks,
>Bryan Sullivan

Received on Tuesday, 13 March 2012 18:27:04 UTC