A lot of
requirements are defined by idiots.
Or, to be a little kinder, by people who don't understand the constraints of the architecture, think they understand more than they actually do and follow up with an "All You Have To Do Is..."
HTTP is not a continuous-connection protocol like client/server is. It's a strict request/response protocol where each response corresponds to a request in a strict 1-to-1 ratio.
In order to get an update, therefore, the client
must submit a request. I think you understand that, but I wanted to lay out the fundamental strictures formally.
In order for a client to submit a request without human interaction, there must be some sort of active logic
in the client that submits the request. Commonly, this is going to be JavaScript, although for simple whole-page refreshes, another alternative is a timed-resubmit metadata tag on the page to be refreshed (providing that the client honors this tag).
For mobile devices, one of the other alternatives is to submit requests from a J2ME, Android, or even (
) dot-Net application installed on the phone itself. And guess what lucky person gets to create these apps?
A fundamental problem is that there simply
is no universal lowest-common-denominator undercover automated submit framework for mobile devices. Some phones are, bluntly, too stupid to support anything at all. Although that particular segment is probably nearly extinct by now.
A more realistic goal is to define your audience, find out what options you can choose for that particular audience, and develop for those options. Don't expect to be all things to all people. You'll either have to simply not support the fringe cases or provide an alternative such as a manual submit button.