[
Lists Home |
Date Index |
Thread Index
]
Actually:
"Four weeks of development at 5k USD per week plus one trip
to the customer site to install and test the import/export definition.
Eight weeks if these must be paired but still one trip.
The customer is obligated to provide all necessary
technical documentation and an authoritative technical
contact for the duration of the effort. Vendor assumes
no costs for any third party software required to implement
this functionality."
Now that's the bid. Still, no one has answered why I need
pointy brackets for that given comma-delimited ascii worked
just as well and is smaller.
No, imports and exports are not messages. They are reports.
A report never calls a function or invokes a message.
It might provoke a human. It doesn't invoke authentication.
It might have a header but that can be comma-delimited too.
It seldom calls for error handling but might be encrypted.
However those distinctions aside, the question is not, why messaging,
it is why *XML* messaging? The answer is usually, "XML is handy",
but that doesn't answer why a handier solution isn't available
because the one application everyone seems to use XML for is the
one application except for graphics that it is maximumly inappropriate.
Messaging solves the impedance mismatch; XML makes it a
thermodynamically expensive solution.
len
From: Liam Quin [mailto:liam@w3.org]
On Wed, Jun 01, 2005 at 10:36:26AM -0700, Ken North wrote:
> Rich Salz wrote:
> > appreciated if you've lived through other bridges:
> > com/corba
>
> That's the first answer that comes to mind.
>
> The COM versus CORBA debates produced a desire for a platform-neutral
solution,
> and we'd already seen the demonstrated interoperability of TCP/IP and
HTTP.
A story I sometimes tell...
Suppose you work at a large bank, and are tasked with implementing
Internet Banking. The main accounts database uses a database that
runs on an IBM mainframe, so IBM's message queueing software comes
to mind... the Web site uses a Sun Solaris system... and so you go
to the COBOL and IBM Assembly teams and ask them to implement your
CORBA IDL spec... they look at you blankly... so you take another
tack and ask them for a report generator that puts pointy brackets
around each field, and they smile happily and say "six weeks". :-)
This oversimplification represents just part of why xml-rpc caught
on so fast; I'll leave it to others to go further.
It's about impedance matching.
|
