I like the approach that Cory has taken with his protocol code, and
think this could be extended further. Most RPC systems support a notion
of a stub generator that emits specialized code to marshal/unmarshal
message types on the wire. Generating appropriate NesC components to
handle this as well as the buffer management would be fantastic. It
could even be extended to support specific retransmission policies, etc.
moving the complexity from runtime to compile time. Good course project
idea...
That's what MIG does, except that the packing/unpacking is merely from Java object representations to TinyOS structures. RPC is quite a different beast; the purpose of marshalling/unmarshalling there has much to do with pointers and platform independence; the appearance of a function call is very different than packing a message structure.
The whole issue of platform dependence on issues such as byte packing (between motes and TOSSIM, for example, were it to run on a Motorola instead of an Intel) has raised the question of whether nesC should have a message type. Field accesses would be transformed into the right packing/unpacking operations. There would be some limitations over standard structs (e.g., no taking pointers). The idea is to do it at the language level, instead of through components. Once you do this, all of the buffer management issues go away.
Phil
-------
"We shall not cease from exploration And the end of all our exploring Will be to arrive where we started And know the place for the first time."
- T. S. Eliot, 'Little Gidding'
_______________________________________________ Tinyos-users mailing list [EMAIL PROTECTED] http://mail.Millennium.Berkeley.EDU/mailman/listinfo/tinyos-users
