Feature roadmap/Faster imaging

Workflow example for the network case above:

1. XOs leave the factory with an older image (e.g. 8.2.1) which includes this OFW code.
2. A custom image based on the same or a later release including activities, content, language and maybe other customizations is prepared and loaded on the XS.
3. An AP is setup and wired to the server.
4. The XOs are turned on, as many as they have power jacks for. On boot up we can require that they hold down some special keys (game keys).
5. Release the keys and the XOs find the AP and start listening on preconfigured multicast broadcast address.
6. The XS broadcasts on the preconfigured address and each XO starts receiving.
7. In case of errors which are more than what can be recovered by error correction the XOs, pick up again on the second broadcast or third or up to ... n (where is configurable or they just turn t he server off).
8. When the XO has a complete image it notifies by putting a special image on the screen.
9. On next reboot of the XO its running the new image. Subsequent boots do not do this multicast listening unless configured specifically by the user.

Some open issues to be addressed by the design proposal:

• Do we pre-configure an ESSID and multicast addresses and hard code those in the factory?
• Can we get it down to no intervention of hold down games keys only?
• What happens if you turn on "too many" XOs? Can we just let them turn on as many as they want and we'll fill them in some order or should we

limit it to set # of XOs at a time? Can we scale it by using more "channels" or more APs?

• Any other security points?