[
Lists Home |
Date Index |
Thread Index
]
On Wednesday 19 December 2001 01:28 pm, Champion, Mike wrote:
> People need flexible/open networking to exchange objects or data between
> applications. They look around and see that HTTP is everywhere and it
> works good enough for the networking layer, and they see that XML is
> flexible and (almost) good enough for the flexible object
> serialization/data exchange layer. Sure, they don't "need" HTTP and XML,
> but both are more or less free, and more or less universal. Why even
> bother complaining that they aren't the Right Thing since Worse is Better
> wins every time? <grin>
My experience tells me that "Worse Is Better" is the way evolution works, but
that eventually, evolution works towards an optimal solution (sometimes to
the detriment of the species). As such, to me, HTTP and XML represent the
australopithicus of open systems: it can stand upright, but run, sing and
dance... we need another few generations.
> Seriously, and getting back to Simon's original point when he originated
> this thread: XML (1.0 or 1.1 as presently conceived) is far from perfect as
> a serialization format for objects or binary data. BUT once the data is in
> XML, it is (in principle) liberated from the application or class
> definitions that produced it. One might think of some SOAP message as a
> kludgy serialization of some business object, but for others it's an XML
> "document" that they can whack on with
Right. That's my point though: the interoperability here is valuable enough
that you really don't want to open the kimono too much. We've already seen
enough in this thread to convince me that adding control characters would be
detrimental to interoperability... and to suggest that there is a pent-up
demand for some pervasive alternative.
FWIW. I should also not that HTTP and XML *per se* don't work: it's the
subsets that people use that do.
|
