W3C home > Mailing lists > Public > ietf-discuss@w3.org > January 1999

The marketplace, RPC and IPP (was RE: Application "core protocol" , BOF/WG idea)

From: Chris Newman <Chris.Newman@innosoft.com>
Date: Fri, 29 Jan 1999 11:22:20 -0800 (PST)
To: "Manros, Carl-Uno B" <cmanros@cp10.es.xerox.com>
Cc: discuss@apps.ietf.org
Message-id: <Pine.SOL.3.95.990129104734.25085B-100000@elwood.innosoft.com>
On Thu, 28 Jan 1999, Manros, Carl-Uno B wrote:
> I think we should also look into why earlier attempts have failed.

It's often hard and sometimes impossible to identify why a protocol didn't
get deployed or used.  It may have nothing to do with technical merit. 
I'd rather focus on operational problems or successes with deployed
protocols, which can be documented in a mostly objective fashion.

> RFC 1831 defines a working RPC mechanism, which could potentially have 
> been used by multiple application protocols but wasn't.

My speculation, which I can't back up with research, is that RPC
mechanisms are a poor choice in general for standards-based protocols.
It's much harder to design an extensible and simple API than it is to
design an extensible and simple wire protocol.  In addition, APIs by their
nature tend to have significant biases in the direction of programming
language or operating system.  Finally, RPCs are designed to "hide" the
network -- I think the network and network latency in particular needs to
be explicitly factored into the design at several levels.  Non-RPC
protocols tend to force that to happen in practice.

> The IPP WG looked at this as an alternative to using HTTP, but came to
> the conclusion that there were few if any implementations of RFC 1831,
> and it was certainly not "on everybody's desktop" like HTTP.

It seems really strange to think of a general purpose RPC mechanism and a
hypertext/MIME transfer application protocol as similar beasts... Which
brings me to a question I've been wondering about: TCP is on more desktops
than HTTP and has a significantly smaller code footprint, so why didn't
you just build IPP on TCP?

> Do we know why RFC 1831 was never picked up by the market place, and
> how we can avoid making yet another generic application protocol,
> that does not get implemented?

The marketplace is fickle so there's no way to be assured of success.  The
best chance is to design for technical excellence, as illuminated by our
past successes and operational problems.  I'll note that IMAP was around
for 10 years before the marketplace really picked it up.  Was IMAP ahead
of it's time, too complex, not properly publicised or was the marketplace
just being stupid?  I couldn't guess the relative influence of those
factors, and I don't think it's feasible to study.

		- Chris
Received on Friday, 29 January 1999 14:24:32 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 23 March 2006 20:11:26 GMT