People are doing lots of useful things with the web. That doesn't
make them appropriate for standardization.
> >Yes, it's easy to prototype new protocols using http, but when you try to
> >productize those prototypes, you may find there are more features in
> >http 1.1 than you wanted to support in your product.
>
> HTTP 1.0 and 1.1 support will be on standard "stock" web servers and web
> clients, regardless of what IPP is implemented on. We don't have to use
> all features of HTTP 1.1 if we don't need to.
No, but the features you don't want or need may get in the way
and cause operational difficulties or subtle incompatibilities.
Telnet's NVT protocol was reused for FTP, but the telnet option
negotiation (which was built-in to the telnet/NVT code on some
platforms) got in the way. For this reason, some FTP clients
had to roll their own NVT code rather than using the built-in
telnet code.
> >(or are there some other kinds of "stock Internet software" that I'm
> >not thinking about?)
>
> Yes. For example, Interent Explorer just uses the underlying Win32
> Internet APIs that all others applications could use as well.
This is the same thing I was talking about.
If nothing else, printing protocols and the web each need to evolve
separately; we need to be very careful about how we tie them together.
Keith