[
Lists Home |
Date Index |
Thread Index
]
Mike Champion wrote:
>
>...
> But what if your website or service has dozens or hundreds of producers
> of inputs, and they all deal with dozens of consumers of inputs.
At this point it should be in everyone's interest to standardize. In
fact, it is often the case that people don't even attempt to do X until
they have a standard for X because they are so afraid of dealing with
the situation that you describe.
> ...
> You have no authority to get the various consumers of messages
> to get them to band together and insist on some common schema, and
> you have an economic disincentive of some sort to reject
> inputs that don't conform to your idea of a valid message.
It's all costs and benefits. It costs one thing to figure out how to
deal with a message and another thing to reject the order. You do
whichever makes you the most money. You also push people to at least use
extensible technologies like XML and RDF to try to minimize the extent
to which your process breaks when one or two new things are added.
> ... What
> is to be done? Well, joining the plumbers union and learning an
> honest trade comes to mind <grin> but assuming that you want to
> live in the environment where no one has the authority to impose
> an authoritative schema, and there are disincentives to employ
> "draconian" measures when imperfect inputs are received, what
> do you do? I suggest that one does the kinds of things that Walter
> is talking about.
I don't dispute that. But Walter claims that the more typical solution
of an industry consortium is both misguided and dangerous. I don't
understand that part. If it can't be done, don't do it. But if it can be
done, do it!
> The RSS situation seems to be a beautiful illustration of what
> the dilemmas that arise in the Real World. First, there's little
> agreement what "valid" RSS is.
I think that the particular personalities involved may have something to
do with that particular issue. Other vocabularies have stabilized. XSLT
was quite confusing for a while because of Microsoft's two early phases
of "experimentation". But they fell in line (as they promised to) when
the real standard came out and now it is quite clear what "valid" XSLT
is, or XML Schema, or SVG. Even XHTML has a pretty clear place you go to
figure out what is right or wrong (if you care, which admittedly most
don't). If there was sufficient money riding on getting RSS right, there
would be a consortium founded and a bunch of big vendors would put their
stamp on a canonical schema.
> ...
> Is there any reason to think that this won't be repeated many times
> over as XML worms its way deeper and deeper into the business
> infrastructure? Do people REALLY think that competitors will band together
> to define and enfore standard schemata, when they all have an incentive
> to use whatever hacks they can to take business from those who reject
> invalid messages that contain sufficient information?
If you don't even *define* the schema then there is a danger that the
market won't even come into existence. Look at all of the companies
taking a wait-and-see attitude towards web services while they "wait for
standards to solidify." Once the schema exists then all of the usual
market pressures will work out the way they did before there was XML.
Sometimes the market leader defines the defacto profile. Sometimes they
follow the spec closely. Sometimes conformance testing shames them into
being strict. Sometimes they have a "strict" or "loose" flag. etc. etc.
--
Paul Prescod
|
