On Thu, May 11, 2000 at 03:50:29PM -0500, Brandon wrote: > > > I'm not sure what he's talking about either. And handling unknown fields > > is also rigorously defined in the spec already. And as I've said before, > > Freenet protocol IS typed: field names are types. > > Right, handling unknown fields is defined. I'm saying that it has to be > changed if we want to be able to have a single node speak both typed and > untyped protocols. > > What do you mean when you say field names are types? > > When I say types I mean number, boolean, string, etc.. > The current Freenet protocol has every value as being a string. So the > protocol is untyped. > > The internal representation of the message, on the other hand, is typed. > > Oskar's binary protocol is typed, meaning that each field value is either > a number, boolean, string, or whatever and the different types are > represented differently in the protocol. > > We need some way to make the two protocols interoperable. > > Oskar wants to make the text protocol typed, in the sense of my previous > use of the word. > > I'd prefer that the text protocol be untyped but we still allow for typed > protocols. >
Why is there a need for a different, binary protocol in a test bed? Are you referring to a simulation? If so, you don't need to worry about the protocol spec, just pass the objects around. And if you are talking about protocol translation, yes, that's a gateway. David Schutt _______________________________________________ Freenet-dev mailing list Freenet-dev at lists.sourceforge.net http://lists.sourceforge.net/mailman/listinfo/freenet-dev
