Re: [xml-dev] Bits on the wire: more drum-banging

[John Cowan]

> todd scripsit:
>
> > I would never advocate  an API only approach because it requires a port
> > of the API to the
> > machine you are using and the language you are using. Not exactly a
> > recipe for mass
> > adoption. Using the protocol approach we can have the best of both
worlds.
>
> As someone who generally favors bits over API, I think the posterchild
> for the API approach is the sockets API.  I can scarcely imagine anyone
> who would avoid this in favor of manipulating TCP, UDP, and IP bits
> themselves.  If there's no implementation of sockets (even a semi-broken
> one like Winsock), network programming is a plain non-starter on that
> platform until it's ported.
>

You've just illustrated very nicely that "bits on the wire" is itself an
abstraction.  To try to get to the base abstraction, you could use assember
and tweak bits, but you still have to talk to the network adapter, and that
most likely has some API.  You could even think of assembler as an
abstraction.

The good abstractions let people be creative and effective, and work with
other abstractions that may not be be so tractable.  So we like to ignore
the IP packets and (in our thinking) pick up the bits after they have been
pulled out, error-corrected, network-order-corrected, and so on.  With XML
we only get them after they have also been assembled into characters.

It works for me.

Cheers,

Tom P