[sdo] [SDO-62]: DataObject containment cycles

[Radu Preotiuc-Pietro <radup@bea.com>]

Radu Preotiuc [07/Dec/06 07:31 PM] 
We have discussed this and related issues quite a bit on the 12/07 call,
but I still am unclear as to what was the agreement regarding the
following point:

if we add an option to turn this check off, does this mean an SDO
implementation is "allowed" to throw StackOverflowErrors? Because if
that's not acceptable, it feels like removing this check could
potentially mean that additional checks are needed in other places.


Frank Budinsky [18/Dec/07 08:03 PM] 
I believe this was intended to be resolved in 2.1.1. Did we not agree to
make this behavior implementation dependent?


Frank Budinsky [18/Dec/07 08:13 PM] 
To answer Radu's question, I believe StackOverflowException is allowed,
if that's what an implementation wants to do. I think another reasonable
behavior is to ignore the cycle at serialization time, and just
serialize up to the loop and discard the back reference. I believe
that's what Tuscany does today. Tuscany also throws
IllegalStateException in some places where it detects a cycle. I don't
think that we should try to define a single correct behavior for this,
at least not in 2.1.1.


Frank Budinsky [20/Dec/07 07:34 PM] 
Current thinking is that this might be handled by providing
configuration options to control whether or not exceptions are thrown.
This is, however, a 3.0 discussion.



Notice:  This email message, together with any attachments, may contain information  of  BEA Systems,  Inc.,  its subsidiaries  and  affiliated entities,  that may be confidential,  proprietary,  copyrighted  and/or legally privileged, and is intended solely for the use of the individual or entity named in this message. If you are not the intended recipient, and have received this message in error, please immediately return this by email and then delete it.