On Mon, 19 Feb 2001, Brandon wrote: > > > Ooohhh! Client protocols! Adam and I were just talking about that. > > > > I'd like a slightly modified version of Whiterose's binary protocol > > I am very much against both binary client protocols and making up our own > client protocols. Binary doesn't really give you anything since the > parsing time is trivial on a client protocol no matter what since it's so > simple. Also, a faster client protocol doesn't matter since you have to > wait on the node to node text protocol anyway. > > The advantage of using a standard protocol is that you can turn Freenet > nodes into components which other applications can immediately talk to > without any additional coding. For instance if the node spoke CORBA then > you could use embed Freenet as a GNOME component.
Embed Freenet? What? How? Why? > > This can be implemented much more quickly than a text-based, > > base64-encoded protocol, which is why I like it. > > That's short term thinking. I'd rather have a protocol which gets the most > usability over time. Hrmm. I still think that someone who's writing a MP3 player would be more likely to add support for a simple protocol than link some weird Freenet library into their program to communicate using a complex protocol, considering code bloat and all. A simple binary protocol should be decided upon and implemented for this reason, IMHO. Although the whole situation disturbs me somewhat. This shit could quickly get completely out of control, when a Freenet node needs to support CORBA, XML, SOAP, HTTP, FNP, FTP, IRC, and whatever other buzzword is in style. We need to think twice before implementing difficult and complicated protocols. HTTP support is easy to implement and widely usable, so that was a good choice. I support a simple binary protocol that client writers can use. But beyond that, I'm not sure. -- Mark Roberts mjr at statesmean.com _______________________________________________ Devl mailing list Devl at freenetproject.org http://www.uprizer.com/mailman/listinfo/devl
