Figure 2 (the GIST layer model) can be misleading in that it seems to define all
sorts of possible protocol combinations, some of which are not even technically
possible, and others have not yet been formally specified for GIST. Comments
from Lars Eggert:

> >Section 3.1., paragraph 7:
> >>              Figure 2: Protocol Stacks for Signaling Transport
> >
> >  Nit: TLS doesn't currently operate over DCCP, and there are some  
> > issues with operating over some variants of SCTP.
> >

true; the diagram is supposed to be indicating the general structure rather than
precisely the detailed stacks that can be used. Maybe add a note that not all
protocol combinations are possible and the allowed ones are in 5.7?