WIMS> CIM> FW: Question to CIM Core about evolution vs replacement of classes.

WIMS> CIM> FW: Question to CIM Core about evolution vs replacement of classes.

Richard_Landau at Dell.com Richard_Landau at Dell.com
Thu Jun 15 14:17:33 EDT 2006


This has not climbed onto the CIM Core agenda yet.  Much urgent business
in front of it.  (The original message went to entire Core group.
Second msg only to chair.  I will reword more strongly and resend to
group.)  
 
rick

________________________________

From: Landau, Richard 
Sent: Tuesday, June 06, 2006 17:40
To: 'crandall at brocade.com'
Subject: FW: [wg-cimcore] Question about evolution vs replacement of
classes. 


John, can we put this question that I posed on the Core agenda, please,
for a short discussion in the near future.  Here's a short version just
to get the idea across.

 

If I add this many properties to a class, is it still the same class?  

          1        Y     N

          2        Y     N

          5        Y     N

        10        Y     N

        20        Y     N

        50        Y     N

      100        Y     N

 

Is there an obvious breaking point?  Is there a breaking point at all?
Or is it reasonable to add lots of properties that were mistakenly left
out in the first draft?  

 

Thanks.

 

rick

 

________________________________

From: Richard_Landau at Dell.com [mailto:Richard_Landau at Dell.com] 
Sent: Friday, June 02, 2006 10:27
To: wg-cimcore at dmtf.org
Subject: [wg-cimcore] Question about evolution vs replacement of
classes. 



(For discussion in mail threads or on the concall) 

Procedural question for CIM Core: How much can one change a class before
it isn't the same class anymore?  

In the PWG (Printer Working Group) work on updating the CIM printing
classes, we have concluded that a couple of the classes will require
major facelifts.  CIM_Printer, for example, currently expresses about
thirty properties.  A few of these will be deprecated and moved to
CIM_PrintService, but the real question concerns properties that it does
not currently express at all.  

The industry model of a printer device includes a *lot* of properties
that are not currently in CIM_Printer.  We will certainly attempt to
reduce the number by collecting many of the properties into Capabilites
and Settings classes, and perhaps by splitting some features into other
separate classes (but not to the point of hyper-normalization).  Even
so, the size of the class is likely to double.  

So the question is, Is it plausible to add twenty, or thirty, or more
properties to a class and still call it the same class?  

- Shall we proceed on the assumption that we are simply going to improve
CIM_Printer (and the others)? 
- Do classes ever revert from final status to experimental status
because of such changes? 
- Will Core insist that we define an entirely new class, e.g.,
CIM_PrintDevice, with all the new properties and then deprecate the old
class in favor of the new one?  

I'm not asking for a decision of any sort at this point, but for a sense
of the working group's process in such cases.  

Thanks. 

rick 

---------------------- 
Richard_Landau(at)dell(dot)com, Stds & System Mgt Arch, CTO Office 
+1-512-728-9023, One Dell Way, RR5-3 MS 8509, Round Rock, TX 78682 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.pwg.org/archives/wims/attachments/20060615/75f3c397/attachment.html


More information about the Wims mailing list