Lists Home |
Date Index |
I do understand that, Sean. I need a tool for many
things I have to build, so that doesn't distress me.
It distresses me if the thing to be built doesn't
have to be so complex that it requires a tool. It
distresses me if the timing for the "loud blast" is
a year or so after release and fielding. Spilt milk
but are we learning any lessons here?
The tool is not the original problem but now that
the recommendation has become viewed as a standard,
the tools are developed and being sold, and we have
people making use of XML Schema, we can't simply turn
it off. We also, and rightly, can't allow it to be
the only choice. Most of us have known since the
early drafts that building XSD without tools would
be hard slogging. I've done it and yeah, for a
db of reasonable size but even little structural
complexity, it is a bear. So is DHTML without a
So problem one, lesson one, is that we should consider
slowing down the spec or recommendation signoffs given
a work that potentially touches a lot of components
and users and markets. The W3C cannot be a bull
in the world information system china shop.
BTW: experts at the time told us the OSI network
would be a big improvement over TCP/IP and the
Internet. Did we throw baby out with the bathwater
then to climb on the WWW juggernaut? I don't know,
but I don't think source or historical success qualify
as good metrics of technical quality.
Let me say this again; if XML Schema is targeted to
and useful by object and relational programmers, it
has hit the largest market/mind share for such things.
Is market success enough? If so, then we have to
look at costs including effort and that includes
as you say, apparency, clarity, ease of use and
adoption, etc. If not, we don't have much to go on
when it comes time to choose among schema options.
I don't recommend it for document work and I'm not
here to say "use XML Schemas because they are the
W3C spec" or "because MS and Oracle like it" or
any of that. I'm saying: we have it, we are using
it, and we can't roll back the decisions at this
point. We just have to make sure that options
are available, understood, and noticed. I didn't
say "James did the wrong thing"; I said, the wrong
conclusions can be made of what he said simply
because he said it. That's groupie thinking.
I hope he gets free limos and bar service with it
plus has a new album coming out soon. :-)
So problem two: it does us not much good to plead
authority. Authorities signed off on XML Schema
and it still didn't make everyone happy and if
we aren't careful, will be costly to fix.
I can only suggest that the world standards groups
stare into these problems a bit more seriously instead
of working around cults of personality. From
the shadow war to the scandals in the boardrooms,
a call will be issued for better tools to get better
transparency over processes and products whose
use and fielding impact us globally. As the
ecosystems specialists are now publishing that
we are beginning to stress the planet with more
take out than the system can replace, we might
want to consider that there is also a finite amount
of money for tools and for people to inspect and
vette standards and specifications, so a similar
budgetary problem is emerging.
From: Sean McGrath [mailto:email@example.com]
>is out there, is apparently working for what it was designed
>or, and for all the warts, is now in commercial products.
>It is a viable option if a difficult one to comprehend at
>first. I did not find it hard to apply once I had a decent
>tool, but I don't accept that to mean it doesn't have warts.
Don't you see the trap here - you *need* a tool in order to wield
W3C XML Schema. The only way to grasp a W3C Schema is
via a visualisation.
You can stare at the text in Notepad until the cows come home
without getting a feel for the model. This is a baaad thing.
By contrast, with RelaxNG, tools are optional. Notepad is just
fine for wrapping your head around what the schema means.
Tools are good and useful but should be optional to understanding
a schema language. Otherwise, the open systems, vendor/platform
independence of the schema goes puff!