Wow, I must have been more unclear than usual. I managed to confuse the
issue completely. Sorry. Blame it on Friday. I was trying to suggest
relaxing a number of associations, particularly CIM_PrinterComponent,
but only in the places where it is currently used, not new locations.
The changes I am suggesting are these:
1. Add the MIN(1) cardinality to the subunit side of the
CIM_PrinterComponent association.
2. Use this association only between CIM_Printer and the mandatory
subunits of a Printer. Our class diagram already states exactly which
of the subunits have minimum cardinality one: InputTray, OutputTray,
MediaPath, Marker, Channel, and Interpreter.
(Note that AlertLog by convention uses a different association to
Printer.) (Also note that, erroneously, I have PrinterSettingData
marked as cardinality one, too; I will fix that. It also uses a
different association by convention.)
3. For the other subunits, where the minimum cardinality is zero, relax
the model's association between CIM_Printer and the subunit to
CIM_Dependency. That is, use CIM_Dependency between CIM_Printer and the
optional subunits: Interlock, Supply, Finisher, and ConsoleLight.
4. If in the future someone were to make another subunit mandatory that
is currently optional, one could change the association between
CIM_Printer and the subunit to CIM_PrinterComponent. Since
CIM_Dependency is the parent class of CIM_PrinterComponent, this is a
legal change in CIM : no semantics are lost and additional semantics are
gained by using the subclass instead of the parent.
5. The two proposed associations between Finisher and MediaPath and
OutputTray are currently trivial subclasses of CIM_Depencency. Relax
the model's associations to CIM_Dependency, the parent class, since no
semantic value is gained by trivially subclassing this class.
6. If we decide to to #5, then we can withdraw the CRs to create these
two new associations before they get to TC for vote.
Boy, I hope that's clearer. Again, I apologize for the over-abbreviated
and confusing message.
rick
________________________________
From: McDonald, Ira [mailto:imcdonald at sharplabs.com]
Sent: Sunday, August 19, 2007 12:34
To: Landau, Richard; wims at pwg.org
Subject: RE: WIMS> CIM> Status of CRs in Core: passed. And new news.
Hi Rick,
About this PrinterComponent usage - NONE of our association MOFs use
the PrinterComponent association - they ALL use CIM_Dependency (which
is correct CIM usage). Only the CIM_Printer to subunit association
would be
affected. That is, CIM_AssociatedMediaPath has nothing to do with
PrinterComponent - it references PrinterElement (the data class) via
CIM_PrintMediaPath.
Choosing only the mandatory subunits is a slippery slope - it means that
a Printer MIB v3 can't make more subunits mandatory (which is indeed
plausible) and be easily reflected in the CIM model.
Cheers,
- Ira
Ira McDonald (Musician / Software Architect)
Chair - Linux Foundation Open Printing WG
Blue Roof Music / High North Inc
PO Box 221 Grand Marais, MI 49839
phone: +1-906-494-2434
email: imcdonald at sharplabs.com
-----Original Message-----
From: owner-wims at pwg.org [mailto:owner-wims at pwg.org]On Behalf Of
Richard_Landau at Dell.com
Sent: Friday, August 17, 2007 5:34 PM
To: wims at pwg.org
Subject: WIMS> CIM> Status of CRs in Core: passed. And new news.
The two CRs that were balloted this week, AssociatedPrintOutputTray and
AssociatedPrintMediaPath, will be forwarded to TC. Yay.
There was a question about cardinality. I conferred with Ira and
responded that the relationships between Finisher and MediaPath (and
OutputTray) are many-to-many, so cardinality need not be specified. I
updated the two CRs with the issue and response, resubmitted them to
DMTF, and reposted them to the WIMS-CIM FTP site.
Also, he (Crandall) did not understand that Finisher is optional while
MediaPath and OutputTray are mandatory. We perhaps did not make that
sufficiently explicit in the modeling, and he thinks we should. The
Visio diagram is clear about cardinalities, but we should make it
clearer in the association MOFs that some components are required and
others are optional.
For example, we should add MIN(1) to the CIM_PrinterComponent
association and then use that for only the mandatory subunits. For the
optional subunits (Interlock, Finisher, ConsoleLight, Supply), we can
use simply CIM_Component.
And he asked why we are doing these specific associations at all if they
do not contain any additional semantics. He feels that we could use
plain old CIM_Dependency (from which these two are subclassed) if there
is no additional information: no new properties, no new cardinality
restrictions. (Yes, eventually, we will learn all the rules.) If we
wish, we can withdraw these anytime before they are voted on in TC; we
have at least a week to decide. My thought: sure, why not. If we don't
need more specific classes, let's not define them. I will be happy to
redraw the class diagram.
rick
----------------------
Richard_Landau(at)dell(dot)com, Stds & System Mgt Architecture, CTO
Office
+1-512-728-9023, One Dell Way, RR5-3, MS RR5-09, Round Rock, TX 78682
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.5.484 / Virus Database: 269.12.0/961 - Release Date:
8/19/2007 7:27 AM
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.pwg.org/archives/wims/attachments/20070820/af6ffd06/attachment.html