W3C home > Mailing lists > Public > www-tag@w3.org > August 2012

Re: Web Architecture

From: <BillClare3@aol.com>
Date: Mon, 27 Aug 2012 10:17:15 -0400 (EDT)
Message-ID: <2090e.251a3172.3d6cdb6b@aol.com>
To: nrm@arcanedomain.com
CC: www-tag@w3.org, timbl@w3.org, ashok.malhotra@oracle.com, ossi@w3.org, ossi.nykanen@tut.fi, ht@inf.ed.ac.uk, masinter@adobe.com, ylafon@w3.org, jeni@jenitennison.com, robin@berjon.com, jeff@w3.org
Your comments raise interesting questions at the heart of  my proposal.  An 
architecture for  the Web needs to be based on a clear notion of just what 
the Web is – and that  may not be as easy to state as it sounds.  For 
instance, Wikipedia defines the Web as “a _system_ 
(http://en.wikipedia.org/wiki/Information_system)  of interlinked _hypertext_ 
(http://en.wikipedia.org/wiki/Hypertext)   documents accessed via the _Internet_ 
(http://en.wikipedia.org/wiki/Internet) ”.  More on  this at the end of the note.

In a message dated 8/26/2012 8:21:56 P.M. Eastern Daylight Time,  
nrm@arcanedomain.com writes:


Thank you so much for your contribution. Let's see  whether some of the 
TAG's members have comments on this. I have only had  time for a quick 
based on that, there are at least a few high level  concerns that occur to 

* The Web is, obviously, a system that is  already deployed on a massive 
scale. The AWWW document written in 2004 is  not an attempt to craft an 
ideal architecture for a global information  system in the abstract, 
although it does make some effort to identify  general principles. Rather 
attempts to document the architecture of the  Web as it is now, and as it 
may evolve incrementally from the base we  have. I don't see in your draft 
lot of connection to the existing core  mechanisms of the Web, such as 
HTTP,  etc.

Although HTTP and URI are indeed core mechanisms, the  question here is, at 
the risk of great heresy, are they essential to a Web  architecture.  
Integration of data  and services from many sources is a useful goal and I’m not 
sure that a Web  architecture wants to be exclusive.  As examples, XML is 
not just a markup language anymore and JSON is often  a useful alternative. 

*  Two key concerns in the design of the Web is scalability and  
discoverability, I.e. the ability for users and software to dynamically  
explore the Web by following links. As described in the Self-Describing  
[1], the system is architected to ensure that, with knowledge of the  URI 
specification and the specifications to which it transitively refers,  
clients can correctly interact with resources and correctly interpret the  
responses they receive (or else discover reliably that they are not  
prepared for such correct interpretation.) It's not immediately obvious to  
me how your proposal addresses such concerns.
Scalability is a concern that can be addressed in many  ways.  I suppose an 
architecture  might preclude it but I’m not sure how a general architecture 
would explicitly  provide for it.  A particular  architectural approach 
such as “map/reduce” or “nothing shared” can foster it,  where applicable. 
Discovery is also fundamental and the approach is to  integrate existing 
capabilities such as RDF and WSDL- and even to generalize  them where 
feasible.  For instance,  WSDL is largely a particular instance of the concept of a 
communications  protocol, which is an instance of an interface.  The concept 
of interface  might also define preconditions,  exception handling, and 
other properties. 

Thank you again for offering this  proposal.
So to the original question - is the Web to be defined in terms of a set  
of protocols from which it has evolved, or are their useful abstractions of  
these protocols that can be exploited.  That of course leaves an onus for an 
alternative definition, which the  proposed framework suggests; i.e. a 
structured set of interfaces for data,  services and applications that 
communicate with each other and with a firm  foundation in capabilities from XML and 
related standards.  And this a large part of the alternative  value of such 
an approach – and, yes, of its challenge.

Noah Mendelsohn
Chair: W3C Technical Architecture  Group

On 8/26/2012 6:59 PM, BillClare3@aol.com wrote:
> W3C TAG  members,
>    It seems typical that over time  architecture groups start with broad
> visions and then tend to become  focused on more and more narrow issues.So
> with all the accelerating  innovations spawned by the Web is it time for a
> revisiting of a broad  vision ?
>     I suspect many might wish to  answer yes to that question, but would
> question how to  proceed.Attached is a short paper that was written for a
> slightly  different purpose, but which tries to address the question from
>  perspectives of XML language and standards, from that of models and
>  frameworks, from that of integration of resources, data, services and  
> and ultimately from the perspective of applications.In  particular, the
> paper focuses directly on basic architecture  principles, identified by 
> W3C, as orthogonality, extensibility,  error handling and
> interoperability.In addition it attempts at a basis  for completeness as a
> framework for applications and their  development. It is much simpler in
> content and at a higher level than  the W3C recommendation on Web
> architecture from 2004 – and perhaps  that is a good thing.
>     So is it time for new  foundations?And is this feasible ?
>     Perhaps  these notes can useful for stimulating discussion within the
> group and  might be useful for soliciting formal sponsorship.
> Thanks for  your consideration.
> *          Bill Clare**
>  *
Received on Monday, 27 August 2012 14:18:14 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:56:47 UTC