[Mep-dev] Tuesday Challenge for 14 April 2009

[Paul Williamson]

On Fri, May 15, 2009 at 11:07 AM, Timothy J. Salo <salo at saloits.com> wrote:
> Furthermore, (again depending on what the MEP actually is)
> it's operating system (and hardware) should be able to run
> _without_ any display.

"Able to run headless" and "not able to run a GUI" are two very
different things. Yes, it needs to be able to run headless, for remote
applications.

If you have a reasonably high bandwidth connection to the remote
device, which we're already assuming, a very reasonable way to remote
control it is to use VNC or some similar screen-sharing technology.
There's no particular reason why the remote control has to squeeze
through a web browser, of all things. A web browser forces you to use
a bunch of fairly elaborate stuff in order to make a
halfway-reasonable simulation of a real GUI program, a la Web 2.0.
That's really an unnecessary constraint. Why not actually develop a
real GUI program, run it on the host where it has access to what it
needs (i.e., on the MEP processor), and (when necessary) present the
screen/keyboard/mouse remotely through a program that specializes in
doing exactly that?

When you don't need to be remote, you can use the local
video/keyboard/mouse facilities and get an even nicer experience. In
particular, I think this might be a boon during development.

It comes down, once again, to the general question about how much of
the user application is provided by MEP. If MEP is a networking black
box that just provides connectivity, then it doesn't need anything
very fancy and its user interface might as well be like your $40
Linksys router: a few boring web pages for setup, and no more. To
whatever extent MEP provides unique user capabilities beyond just
moving bytes around, it probably needs more user interface.

The hardware to provide local video, keyboard, and mouse are no longer
expensive. Nor is the CPU horsepower to run a GUI. That's one of the
things that's exciting about platforms like the Atom and the
BeagleBoard. There's no need to omit these functions if we can make
good use of them in some important use cases.

73  -Paul KB5MU