RE: "Binary XML" proposals

[Miles Sabin <MSabin@interx.com>]

David Brownell wrote,
> > If you mean particular identifiable implementations, then no,
> > not unless I'm allowed to count BINDs 4, 8, and 9 separately.
>
> Nope, and not client-only implementations either! :)

No fair ... rationale please.

> DNS is not an example of a "widely implemented" protocol; 
> "widely deployed" is rather different.  (Arguably, you just 
> picked a bad example ... where there's really only one 
> significant implementation.)

But that's simply not true. DNS is maybe slightly less widely
implemented than HTTP clients/servers. But far less less widely 
than you're making out. How many HTTP servers can you name? How 
many proxies?

In any case, there's a difference between "open, implementable
by anyone" and "open, implemented by lots of people". Even if
there were only one example, DNS would still be in the former
category, and I think that's good enough.

> > Interoperability isn't simply due to a lack of diversity.
>
> Actually, for DNS it's been a major factor.  Original specs did
> not match the implementation, and for all I know that's still
> an issue ... because that implementation was so widely deployed
> that it became the real protocol spec.

That's not exactly unusual in the protocol space. If anything it
goes to show that there's a counter-tendency to the one you're
worried about. Yes, there's a trend to closed proprietary systems,
but there's also a trend in the opposite direction. And all
vendors, at least some of the time, have an interest in supporting
the counter trend ... eg. when they don't have a clearly dominant
market position.

> > I understand your concern, and I share it. But I think you're
> > overestimating the extent to which text is a defence. 
>
> I just said "orders of magnitude", I didn't say how big the 
> original pool of "interoperability defenders" would be!

I still disagree. My experience, having (legitimately) reverse 
engineered both text and binary proprietary file formats, is that 
there's not a huge difference between the two in terms of effort.

The real killer isn't text vs. binary: it's who's in control of
change. If a vendor can make egregious changes from release to 
release then you're stuffed, text or binary. This, I think, counts
in favour of a binary standard even by your own criteria: it
would make it harder (nb. harder, not impossible) for anyone to 
make unilateral changes.

Cheers,


Miles

-- 
Miles Sabin                               InterX
Internet Systems Architect                5/6 Glenthorne Mews
+44 (0)20 8817 4030                       London, W6 0LJ, England
msabin@interx.com                         http://www.interx.com/