OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

 


 

   Re: Hostility to "binary XML" (was Re: [xml-dev] XML 2004 weblogitems?)

[ Lists Home | Date Index | Thread Index ]

Michael Champion wrote:
> FWIW, the show-stopper argument in a number of discussions at XML 2004
> that I heard against the idea that "XML text is ubiquitous, don't mess
> with it" was from the wireless people:  XML is NOT ubiquitous in our
> world, because of the excessive bandwidth requirements.  Our technical
> constraints are fundamental and not going away anytime soon, so don't
> expect Moore's Law to make everything alright the way it has made
> convenient but inefficient approaches work on the desktop and on the
> server.  On othe other hand, we desperately want the tools and
> standardization that XML offers, just not that verbose serialization.
> We could write a standard for our industry, but we want *one*
> internet, not a wired one and a wireless one."

Yes, that is also the argument they (we, I guess I should say) made at 
the binary workshop, and for quite a few of them had been making already 
for some years before that.

I fully understand the way in which XML as a ubiquitous format is more 
than extremely valuable, and consequently highly valued. But on the 
other hand it only flies for a constrained, localized, and qualified 
definition of ubiquity. To make a deliberately exagerated (and unfair) 
comparison, some people see XML as ubiquitous and won't hear of anything 
else in the same way that some people see Win32 as ubiquitous and don't 
see the point in targeting other platforms. The point is, who's ubiquity 
are you talking about? Let me illustrate with a fresh quote from Russell 
Beattie's blog:

"The thing I have to constantly impress on people who are just getting 
up to speed in the mobile space is the numbers involved. The hundreds of 
millions of devices shipped just this year alone by just Nokia, the 
billions of subscribers out there, the massive growth that's going to 
happen over the next few years. Many companies just can't imagine this 
sort of scale. Apple shipped 4 million iPods in the past quarter, Palm 
shipped 1.5 million Treos and Dell shipped 8 million PCs and 185k Axims. 
Very nice, but Nokia shipped over 50 million handsets in the same 
timeframe. See what I'm talking about?"

Mobiles do use XML -- just look at how SVG has stormed over that world 
-- but nowhere near as much as they could if it didn't chop off half the 
battery's life each time you parsed a little too long :)

In light of the above numbers, the question is "would binary XML hurt 
XML's ubiquity, or would it take it to another level entirely?" We could 
after all well be, sooner rather than later, talking about an order of 
magnitude difference.

> I doubt it too.  I think people have accepted that there is a problem,
> but there is intense skepticism that a single standard format will
> cover all (or even 80%) of the requirements.

Skepticism, or at least some form of aporia, is also present in the XBC 
WG. We are almost done with the "destructive" phase of collecting use 
cases and properties, which naturally easily leaves one with the 
impression that it's just too big. But then that was the intention to 
start with, and we're now starting to look at removing the dog's 
breakfast from the mix and seeing if there isn't a sufficiently large 
subset that would justify a standard. More on that in the coming months.

> One data point that some
> people from Microsoft brought up in the Binary XML Town Hall:  They
> have *tried* to come up with one binary serialization that will
> satisfy even their internal customers, and haven't found one.

Yes, they said that at the workshop too. But that needs to be qualified: 
it's quite possible that there's no good option that can cover for 
instance both efficient transmission between MSSQL client and server, 
and compilation of XAML documents to an intermediate format, but those 
are fairly specific. Also, while I know for a fact that Microsoft has 
top-notch engineers at its disposal it's happened before that they've 
shipped less-than-average products or failed to find the right solution 
to something.

> There is
> also EXTREME skepticism that a typical W3C design by committee job
> will come up with anything useful.

I don't think that's fair. The W3C's had failures and successes -- and 
anyone who thinks any organisation will get things right every single 
time is obviously on crack. I think that a critical mass of people is 
now aware of pitfalls they might not have thought of before. To put it 
differently, most people on the XBC WG have used XML Schema, and a fair 
number have actually implemented it ;)

> The sense I got from Michael Leventhal's presentation on the XBC was
> that you are doing the Right Thing.  I didn't hear anyone in the
> audience disagree that this is a very valuable exercise, even if many
> are convinced that it will ultimately conclude that the
> (non-wireless?) world is not ready for a single binary XML standard. 
> Most of us would  be extremely happy to be proven wrong on that point,
> if the data are there to back it up.

Well if someone went ahead and produced a mobile-specific standard, it's 
what the desktop minority will end up using a couple years down the 
line, simply because that's where the center of gravity is. I'd line 
that up as a pretty good reason that *if* a new standard is to be 
created for binary XML, it certainly can't be mobile-specific.

> I ended my talk with a *personal* [don't hold any past, present, or
> future employer responsible!] recommendation "Leave evolution to
> Darwin, not Berners-Lee".  In other words, it's time to experiment, to
> develop specs that meet the needs of some specific industry, to see if
> parsing and compression technology for XML text can be dramatically
> improved .... and THEN to come back with data and best practices in
> hand to see if W3C Recommendations can be agreed upon.

I agree with a lot of what you say here, except that your timeline is 
off by about 4-5 years :) We've got the industry-specific standards, 
we've got the more generic standards, we've got deployment experience -- 
in some cases we're talking millions of terminals. The XBC WG is just us 
guys coming back with the data and best practices in hand to see if W3C 
Recommendations can be agreed upon.

For sure there's always room for innovation in the space, but we're way 
past the point where it's experimental. In fact, it's been quite a while 
since I last saw something that caught my attention as truly innovative 
in the field. The question now is really about if we want one W3C 
standard for all -- at the cost of having possibly two universal formats 
instead of one -- or do we want XML and the three or four binary XML 
standards (from other groups) that'll survive market competition?

-- 
Robin Berjon





 

News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS