RE: [wsdm] Editorial comments

Title: Editorial comments


ok, I see that a lot of people like UML. Does then anybody 
have any idea how to represent structured event topics nicely in a class 
diagram?...

 

right now we have 

<wstop:Topic name="abc" 
messageType="ns:xyz"/>

 

represented as

<<event>> ns:xyz abc;

 

when I 
have

<wstop:Topic 
name="abc" messageType="ns:xyz">

    <wstop:Topic name="def" 
messageType="ns:pqr"/>

</wstop:Topic>

 

the best I can guess how to fit it into the class 
diagram would be

<<event>> ns:xyz 
abc;

<<event>> ns:pqr 
abc/def;

 

Unless I hear better, this is what I'll 
use.

 

I'm still not sure how would I expand topic aliases in 
this way.

 

Another point is that having a maze of classes with 
ralationships for every capability<->topic link is not very heplful. I 
went through that exercise.

-- Igor 
Sedukhin .. () 
-- (631) 342-4325 .. 1 CA Plaza, 
Islandia, NY 11788 

 

From: Mike Clements [mailto:] 

Sent: Monday, October 11, 2004 6:06 PM
To: Sedukhin, Igor 
S; 
Subject: RE: [wsdm] Editorial 
comments

It's easy for us to underestimate the 
value of the UML models since we already understand the spec. I find UML more 
intuitive and easier to read than XML.

 

It was only a few months ago when I 
was reading these specs for the first time. When I was doing that the 2 things 
that helped me the most in understanding it were the written text and the UML. 
The XML is an essential normative part of the spec, but it is virtually useless 
in gaining understanding. XML is intuitively readable only to robots and 
vulcans.

 

Pictures are useful. If UML is too 
detailed and expensive to keep up to date, let's replace it with some kind of 
non-normative block diagram. That way we get the intuitive advantages of 
pictures without the hassles of maintaining detailed and redundant 
UML.

 

Just my $0.02.

 

From: Sedukhin, Igor S 
[mailto:] 
Sent: Monday, October 11, 2004 1:13 
PM
To: 
Subject: [wsdm] Editorial 
comments

  

  
I suggest that we remove all UML 
  models which define the capabilities throughout MUWS and MOWS specs 
  because 
        1) it is hard to keep XML and UML syncronized 
        2) the information in UML is redundant to the normative definitions of 
  the XML 
        3) the value of these models is not incredibly useful for understanding 
  the specs 
        4) it makes sense to focus on XML information definitions and exchanges 
  which is what gives us interoperability 

  
-- Igor Sedukhin 
  .. () 
-- (631) 342-4325 .. 1 CA Plaza, 
  Islandia, NY 11788