ace project <[EMAIL PROTECTED]> writes:

> I've attached a (Word generated) HTML-file with the 
> proposed protocols/headers to be used. Plz bomb them
> :)

I have only a few comments right now.  More as I have time to delve into this 
further...

1. If you establish a TCP/IP connection, then provide a way to pass a
   "protocol version number" upon connection establishment.  If using UDP,
   then a protocol version number should be passed in each packet.  There are
   a few ways to do this:

   a. Add two extra bytes in the Level 1 or Level 2 header.  You likely don't
      want to add extra bytes, so the other options are probably better right
      now.

   b. Allocate 3 bits in FLAGS as a version number, with '111' reserved for
      versions greater than 6 (with some place to put them added later and
      known to exist because of the reserved '111' value.

   c. Allocate 1 bit in FLAGS to mean "initial protocol" (version 1.0).  If
      not enabled, then the non-initial protocol must define where/how the
      version number is encoded.  It should be well documented that any
      subsequent protocol version MUST define a way of specifying the protocol
      version number.

   If you don't provide for a protocol version number, you'll likely end up
   (in version 1.1 or 2.0) with things that you'd like to do but can't, while
   still maintaining backwards compatibility.

2. Providing only 12 bits of flags seems really skimpy to me.  This isn't
   really an issue, though, when you add the protocol version number feature
   described above.  If more bits are needed for a future protocol version,
   they can be added by that version.

3. FDM is declared as 50b in one place, and 20b in another.  Ditto for
   Nickname.

4. Flight Dynamics Model as text sounds risky.  What about having FDMs
   registered at FlightGear.org (with a numerical identifier assigned to them)
   and then just passing the appropriate FDM id?

5. Why is a termination character ('\r') required for plane and FDM when
   there's a length field provided?  It shouldn't be necessary.

6. Eight players seems like a severe limitation.  I can see the potential for
   hundreds of players.  It would be nice to design the protocol such that
   there's no limit on number of players.

You've got a really nice start here!

Cheers,

Derrell

_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Reply via email to