Lists Home |
Date Index |
The odd thing about this thread is that people are defending
XML to me. I have a somewhat distinct memory of defending and
promoting markup for some rather long period of time, and so
far, most of these arguments are ones I've presented. This
feels like a college debating team exercise. My points:
1. The thread is not about killing XML. It is about
what might succeed it. I think that will happen. I
doubt it will be soon, but sooner than we think. The
dominant effect of the web has not been a radical
improvement in technology; it has been a radical
increase in the speed with which it is fielded and
replaced in some areas, and settled into suboptimum
minima in others. So as predicted in the NIST papers
of the late 80's, it is doing what was predicted,
following an annealing model.
2. I think XML for messaging is a good example of
pop computer culture. The masses get an idea based
on long tail politics and apply it widely. It works
ok but not as well as other solutions might. It is
standards for the sake of standards, or put otherwise,
to save us the argument, we accept the suboptimum condition.
Because of the pop culture (guys like Clay Shirky believe it
is the only solution, but Shirky is himself, a very
good example of pop culture at work since the majority
of his articles are a good ten years behind the original
thinkers), we get sub-optimum systems. For a mass market
automobile, that's ok. For a mission critical system,
it can be catastrophic. This is a buyer's market, so
I don't see conspiracies here; just bean counter
economics and the megaphone at work. (Carrie or Bo?
Do you really care because next year you get to do
I was curious about whether this list as a culture
was thinking ahead. Apparently not. That isn't a
slam as much as it is a confirmation that a) this
is a pop culture and b) the pattern of innovation
that has to be developed in isolation holds.
What is learned? Because of what one observes from
the Shirky Syndrome, it will be profitable to patent
the research, apply it to big ticket items, then
wait for the popular culture to need it and license it.
The MPEG/MHEG style of consortia has legs. The W3C
style of consortia will fade in importance. I'm not
promoting; just observing and taking notes.
I agree that one can challenge the 3-tier architecture
but it is itself, a phenomenon that emerges as a result
of percolation in the technologies, that is, the innovation
creates the impedance mismatch just as local views shorten
the lifecycle and reduce the scope of upper-level ontologies.
No size fits all. Some sizes fit most. 3-tier may go away
for awhile, but it will reemerge.
From: Didier PH Martin [mailto:email@example.com]
I agree XML is not the most efficient format. We got more efficient formats
in the past. But...
a) We now have a common alphabet, still not a common language. Comparing
this to actual tour of babel after several thousand of years, this is quite
b) The transformation tool (XSLT) that comes with it allows us to perform
translation form one language to another. Even perform transformations into
other syntax languages like for instance Java or ECMAScript. Moreover, we
can even perform this transformation in most modern browsers. This is a
tremendously useful tool for model to model transformations.
c) Easier to debug. For anything returning an XML based format, its quite
useful to use the browser to check the returned result set, easier to debug
than with binary formats: for example, to debug a SQL/xml statement or an
Xquery statement. I know, not a lot of tools out there are providing this,
but slowly and surely more and more do.
You know Len, for at least a couple of years and since new technologies take
a veeeeeery looooong time to percolate in developers mind, I'll wait that
more tool use XML as an interchange format.
However, in the near future, I think it is more useful to challenge the
actual three tiers architecture. The middle tier is used mainly to resolve
the impedance mismatch between relational database and clients. Most of the
time, tables and rows and transformed into objects and these objects used to
feed data into dumb terminals through HTML; this time instead of having
green screen we have colorful screen but still the same paradigm. New
progress made on the relational DB front allows us now to talk directly to
the RDB and get rid of the middle tier. For the moment, until we have more
tools and developers waking up to XML, I wouldn't change the format (at
least until it becomes really really pervasive) . But I think its time to
provide richer environments to users and go beyond the mainframe
architecture we are sticked with these days. So, if a don Quichote is in
search of a windmill its better to attack the three tier architecture than
the messaging format. Yes I know Len, big money, big players are too happy
to make money with this actual architecture...