Re: NG an evolution of or a departure from HTTP?

You raise two points: one about the value of "loose coupling" between clients
and servers, and the other about the value of human-readable wire encodings.

I agree about the value of what you call "loose coupling".  I think it is what
I've been calling "anarchic evolution".  That is, independently developed
evolutionary changes to some existing application's network interfaces can be
concurrently and incrementally rolled out into a widely deployed existing
system.  It's not quite correct to say that existing distributed system
technology cannot facilitate this.  While some people will argue that subtyping
of object types suffices, I claim it doesn't do the job very well.  Putting a
property list in every type you want to be extensible does the job, but not
very elegantly.  I'm working on an idea I call signature extension that should
provide a more elegant solution.  I don't have a decent introduction or
exposition ready yet, but you can find an out-take of a technical discussion of
the idea on the PDG home page (accessible to W3C members only).

On the binary-vs-text encoding issue, there is a lot of debate.  There are
worthy pro's and con's on both sides.  I myself favor modular software
architectures, enabling case-by-case selection of the best thing for the job at
hand.

Received on Thursday, 20 August 1998 15:49:06 UTC