Arjun,

3.1 was never a final release, it was a milestone release, 4.0 MR2 is just a continuation of this, the api and language won't fully stabalise and you can expect bugs unil we do a candidate release.

ShadowFacts were finished and shouldn't be visible to the user, however code generation in Java has known problems - which hibernate proxies also suffer from - where classes must have a default constructor and must not use the final modifier. Thanks to Objenesis, http://objenesis.googlecode.com/svn/docs/index.html, M2 was able to get over the default constructor limitation, but there is nothing we can do about final. The work towards allowing ShadowFacts to be optional is additional to this, i.e. its a new feature. As said before these are milestone releases, aimed to give you a "snapshot" at our R&D progress, we give no guarantees and you must expect rough edges and big changes - the rewards are you help us make a much better 4.0 final release.

Backwards compatability is always an issue, but that's why we did the version number change, we believe the changes and the advantages gained make this worth while.

Mark
Arjun Dhar wrote: 
Mark Proctor <mproctor <at> codehaus.org> writes:

  
Sorry thats the stateful working memory interface, StatefulSession - 
just incase my opening paragraph confuses anyone.
Mark Proctor wrote:
    


Hi Appreciate the quick response and accept all that is written; as for 1 and 
3.2.

TECHNICAL
================================
[1]
"Can you show me a use case where you neeed 
  
access to the WorkingMemory from a StatelessSession"
    
... Conceptually No, since you have it covered by allowing a person to Assert a 
list at a time (Faster when doing Batch mode; this is what I was doing) , 
except had built a wrapper method On the 'WorkingMemory' to achieve this. So 
will just have to do re-factoring to my code which again will not be backward 
compatible.

..So to summarize: Code written for JBoss Rules 4 may not work for 3.0 and 3.1. 
If that isnt and enginerring issue then kindly ignore.

[3.2]
Suggestion:: If Shadows are not fully implemented then they should be 
encapsulated and not be visible to users. From porting from 3.0 to 3.1 a 
NullPointerException due to a feature not to be delivered, can be considered a 
bug. From a blind QA perspective 3.2 is a bug!


ROOT CAUSE
================================
... I think if we look at both these points and from a project level, there is 
an issue of Backward Compatibility. I guess you guys have your hands full to 
care about that.

...But seriously, I understand the constraints you guys have to work with :o) 
but wanted to let you know that while I'm a critic (hope am not being picky) 
I'm also a big fan on what 
is being built.

Thanks again!
Arjun



_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev