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

XMLP Glossary: application

From: Henrik Frystyk Nielsen <frystyk@microsoft.com>
Date: Sat, 17 Mar 2001 15:16:09 -0800
To: "'Mark Jones'" <jones@research.att.com>, <skw@hplb.hpl.hp.com>
Cc: <xml-dist-app@w3.org>
Message-ID: <79107D208BA38C45A4E45F62673A434D0297CA5F@red-msg-07.redmond.corp.microsoft.com>


>This "application" issue confused me from the beginning with 
>SOAP.  I finally inferred from the SOAP spec that a SOAP 
>application was co-extensive with a SOAP processor.  You get 
>this hint when the spec talks about a "SOAP application 
>receiving a SOAP message" and performing various actions to process it.

What might be the confusing part is that the SOAP spec is careful not to
use application in any semantic manner - it is more or less used in
stead of "thingy". The semantic part is of course the processor.

>I tend to think of a SOAP or XMLP processor rather like a web 
>server. It effectively hosts capabilites for supporting MANY 
>applications in the traditional sense of the word.  It 
>provides a computational environment where some work gets 
>done, much like CGI, and has certain rules for communication, 
>invocation, and passing data.

In some sense this is true but the compositional model of SOAP is a bit
more fine grained than this - it allows several "modules" to work on the
same message. In many Web servers this maps to modules or event handlers
looking at a single HTTP request/response pair.

>The concept of an XMLP module is currently capturing some of 
>my sense of an application or an application component in our 
>current model.

I agree.

> It is a related set of block types and handlers 
>that encapsulate some piece of functionality.  Since there is 
>a composability model, the complete execution of an 
>"application" (either on a particular node or across nodes) 
>may involve a composition of modules -- that is why I would 
>not call an individual module an application. 

Although there is no reason why a module couldn't be an "application".
It is of course true (and will hopefully be the case) that modules may
play together to create more complex applications. If not then there
would be no need to have a compositional model in the first place.

>But it also 
>seems like an odd use of terminology to call an arbitrary 
>bunch of unrelated handlers that happen to exist on the same 
>node an "application".  To me, it is like calling a collection 
>of unrelated CGI scripts that are accessible from a given web 
>server an "application".  

I agree - maybe we should call it "hosted handlers" instead.

Henrik
Received on Saturday, 17 March 2001 18:46:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:59 GMT