Re: JMIMapper Error Handling

Yes, this is a good suggestion. I have had this in my mind for a long 
time - it would be nice to display all the errors during JMI ifcs 
generation to the Output window similarly to Java compilation. I just 
always had other things to do which I considered to be more important.
The same thing should probably apply also to XMI writer.
Could you please file a new enhancement request for this and for the 
primitive types package issue from your previous e-mail so that we won't 
forget it?
Thanks,
Martin

Brian Smith wrote:
> Hi,
> 
> (1) How extensively should the JMIMapper check the model for 
> well-formedness? That is, should it assume that the model is well-formed 
> before it starts? Should it give its "best effort" to find problems with 
> the metamodel? In particular, should a JMIMapper implementation verify 
> the model constraints before starting the actual mapping?
> 
> What should the JMIMapper do when it detects a violation of the rules 
> that are specific to the JMI specification. For example, a value of 
> javax.jmi.ignoreLifecycle that is neither "true" or "false", or a class 
> nested inside another class, etc.
> 
> If the JMIMapper is allowed to do anything besides silently ignore these 
> types of errors, then I suggest that we add a new (set of) checked 
> exception(s) that can be thrown by JMIMapper.generate(). In particular, 
> if the JMIMapper checks the constraints from the MOF metamodel, then it 
> should be possible to get the set of violated constraints. Similarly, if 
> a JMI-specific error was found, there should be some way of finding out 
> what that error was.
> 
> BTW, I know that neither CWM 1.0 nor CWM 1.1 fully conform to the MOF 
> 1.4 constraints (the obvious error is that some of the associations are 
> not public.)
> 
> Thanks,
> Brian
>