Lists Home |
Date Index |
Paul Prescod wrote:
> Today I step through some Expedia forms to buy an airline ticket, or
> to buy a book from Amazon. Tomorrow I write a Python program to
> step through for me. It demonstrably works, even with plain old
> HTML. I know it works because I do it!
And how many hours does it take to do it? And how fragile is it, because
you're reading something which is meant to be read by humans, not programs?
Or if you are talking XML, how much time will you spend agreeing your
own little XML protocol with Expedia or Amazon? Will you send your query
as GET parameters and force the server to institute lots of state
management so you can build the query up call by call rather than
sending an entire XML query in one block? Or if you use POST, what CGI
variable will you use to hold the query block? Or will you just send one
naked XML block? How will they document what elements they expect to
receive and send?
Why not get productive, do it the XML way? Why not write an XML grammar
which answers all these questions, and an XML protocol to implement the
transport? Why not call them WSDL and SOAP?
I can "Add Web Reference" in Visual Studio in about a minute. Leave out
the obvious assigment statements to and from my query and result
structures, and it takes me two lines of code to do a synchronous WS
call, three to do it async. And I'm sure the Java guys aren't far behind.
Some of it's a bit new and rough, but no-one has yet persuaded me that
there are any show-stopping flaws in it.
I really don't understand what you're offering as an alternative to WS.
If it's Python and HTTP GET then I think I can see cavalry charging the