Issue 7972: OCL Constraints in many levels (ocl2-rtf)
Source:  (, )
Nature: Uncategorized Issue
Severity: 
Summary: I been using rational rose and bold for delphi in some projects. This have worked very well for me.
When adding constraints to my models i have some times whished that there whas a way to do this
on different levels. Eg. error-constraints (if persisted the object would be a dirty data) , 
 
warning-constraints these can be broken but there is high probability that the object is ill formed from the
system user perspective (example a new customer whith no billing adress) and finally a hint-contraint that
when broken indicates that the object containes strange data (example a new customer object with the 
same phone number as a existing customer)
 
My own solution to this have been to add contraints of the first type to the model. This have enabeld me
to create generic code dealing with if the user should be allowed to save a object or not. 
 
The other types of constraints have been added as coments as a way to make the model as complete as
possibel. The implementation of checking and dealing with these constraints later in the project have ben 
solved in a mutch less generic and cumbersom way.
 
I thinīk that if the standard included a way to specify different levels of ocl statements in the model this would 
benefit the model driven way to make software
Resolution: 
Revised Text: 
Actions taken:
December 10, 2004: received issue
Discussion: 

End of Annotations:=====
ender: wask@amethyst.omg.org 
X-Mailer: QUALCOMM Windows Eudora Version 6.1.2.0 
Date: Fri, 10 Dec 2004 13:26:34 -0500 
To: juergen Boldt <juergen@omg.org> 
From: Fred Waskiewicz <wask@omg.org> 
Subject: Fwd: OCL Constraints in many levels 




From: "Hans Andersson, BlueFront AB" <hans.andersson@bluefront.se>
To: <wask@omg.org>
Subject: OCL Constraints in many levels
Date: Fri, 10 Dec 2004 16:25:45 +0100
X-Mailer: Microsoft Outlook, Build 10.0.2627

Hi
I been using rational rose and bold for delphi in some projects. This have worked very well for me.
When adding constraints to my models i have some times whished that there whas a way to do this
on different levels. Eg. error-constraints (if persisted the object would be a dirty data) , 
 
warning-constraints these can be broken but there is high probability that the object is ill formed from the
system user perspective (example a new customer whith no billing adress) and finally a hint-contraint that
when broken indicates that the object containes strange data (example a new customer object with the 
same phone number as a existing customer)
 
My own solution to this have been to add contraints of the first type to the model. This have enabeld me
to create generic code dealing with if the user should be allowed to save a object or not. 
 
The other types of constraints have been added as coments as a way to make the model as complete as
possibel. The implementation of checking and dealing with these constraints later in the project have ben 
solved in a mutch less generic and cumbersom way.
 
I thinīk that if the standard included a way to specify different levels of ocl statements in the model this would 
benefit the model driven way to make software. 
 
Pardon my bad english im from sweden.
Best regards Hans Andersson

==================================================================
Fred Waskiewicz                                    OMG
Director of Standards                               250 First Ave., Suite 100
wask@omg.org / +1 781 444-0404 x127    Needham, MA, USA 02494
==================================================================