Lists Home |
Date Index |
Bullard, Claude L (Len) wrote:
> I don't think so. If this is a composable set of specifications,
> it is a matter of the programmer not paving their own roads to
> hell by putting together components that don't work. So the
> WSIO profiles bear watching.
The community has to discover ways to use things together through trial and
error and then codify that experience for the benefit of others -- in my
non-statistically-verified experience, the probability that a profile
written in advance will be even slightly useful is in the single digits, so
I see little point wasting my time reading one. A profile written in
advance of serious, real-world experience is about as useful as a
stockmarket prediction, a system for winning at slots, or CIA intelligence
reports about WMD.
Look, for example, at how the model 2 architecture evolved in J2EE after
people realised that println statements in servlets were a lousy way to
generate HTML and embedded code in JSPs was a lousy way to control program
flow--in fact, look at how J2EE itself evolved. Java was originally intended
for embedded devices, and then for interactive Web pages: I don't think that
anyone saw it crapping out in those areas and becoming the dominant
technology for server-side Web app development instead. However, since the
Java language, JVM, and libraries are self-contained, well-designed building
blocks, they adapted easily to the new niche.
Ditto for XML. Remember the promise of XML in the browser, with stylesheet
linking, client-side XSLT, XLink, XPointer, etc? Like Java, XML ended up
moving to the other side of the firewall. The XML that did go client side,
RSS, is not what any of us expected.
I don't disagree that the profiles are useful; however, I want them to be
like modern science, derived from experimentation and observation; what
we're getting in this profiles reminds me more of alchemy.
All the best,