W3C home > Mailing lists > Public > xml-dist-app@w3.org > January 2001

RE: Integrating XP Into Web Infrastructure

From: Mike Dierken <mike@DataChannel.com>
Date: Tue, 9 Jan 2001 10:23:35 -0800
Message-ID: <8E864C73E16B864BB594712EDB3C89A0409D88@belmail2.datachannel.com>
To: "'Dave Winer'" <dave@userland.com>, xml-dist-app@w3.org
Dave,
I think you've brought up a great point that might be the source of a lot of
miscommunication here & elsewhere.
It has to do with 'object-based programming' and 'function-based
programming' - classes .vs. procedures.
For example, people tend write code like one of the following:
a) server.get("resource-id")
b) get(server,"resource-id")

> If you did a GET on a logical structure on a server, you need an object
model.
Since the world is already doing GETs on logical structures on the server,
are you implying that there already is a Web Object Model? 
I think the Web can be viewed this way - which was the basic assumption of
an HTTP-to-Java dispatching framework I did & donated to Apache
(http://xml.apache.org/xang/overview.html)

One interesting thing is in my view of a Web Object Model, there are
resources, and resources have methods and methods have parameters and
parameters have names. In a Java object model, there are objects, objects
have classes, classes have methods, methods have parameters and parameters
have type and sequence. Trying to map the 'parameter list' from 'name' to
'type and sequence' is the hardest part.

> This idea competes with parameter lists.
This may be the center of confusion here. I don't understand how an 'object
model' competes with parameter lists.

Mike
PS
It's funny, because I *like* object models. They make sense to me. 
They give me a chance to simplify and refine and document my interfaces.

PPS
I think that one large use of XP for asynchronous messaging (and XML
messages in general) might bring to light a third style of programming,
namely 'property-based programming'. Sending stuff over a network seems to
work better when you toss around large-grained, semi-structured 'blobs',
rather than fine-grained data (like a number of functions that deal with
little parameters). Dealing with that structured blob seems gnarly if you
require a specific implementation to access that data - either requiring the
same code environment on both sides, or passing the actualy byte-codes along
with the data (like some object-based approaches).



> -----Original Message-----
> From: Dave Winer [mailto:dave@userland.com]
> Sent: Tuesday, January 09, 2001 8:00 AM
> To: xml-dist-app@w3.org
> Subject: Re: Integrating XP Into Web Infrastructure
> 
> 
> Aaron and Mark, you should explore this more fully.
> 
> If you did a GET on a logical structure on a server, you need 
> an object
> model.
> 
> This idea competes with parameter lists.
> 
> Maybe it's my Unix background, but I *like* parameter lists. 
> They make sense
> to me. They give me a chance to simplify and refine and document my
> interfaces.
> 
> But you could try to make the world align around GET and 
> POST, and see what
> happens.
> 
> I've been through this loop once before with Apple. They 
> insisted that an
> object model would be understandable to most scripters. I 
> disagreed. We did
> the procedural interface, they did the object model 
> interface. Now almost
> ten years later it might be a good idea to find out what 
> happened there. I
> honestly don't know.
> 
> Perhaps both approaches deserve a chance. Thanks to Aaron's 
> eloquent and
> respectful writing, I now understand what the issue is.
> 
> Dave
> 
Received on Tuesday, 9 January 2001 13:24:11 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:58 GMT