#14 Inconsistency with OMG Specification

[Nima Sharifi Mehr]

Dear OpenORB developers,

I found that some inconsistencies with OMG specification
in OpenORB Concurrency Control Service is available
which are very crucial in practice and have been
overlooked in it's implementation:

1 - In the implementation of LockSet's "lock" method you
should notice that it can be called on behalf of a implicit
transaction and if this call blocked and the transaction
be aborted then the call should return with the
Transactions::TransactionRolledBack exception.

2 - A lock can be held multiple times in the same mode
then you should use some solutions such as a counter
for managing this situation.

3 - the "get_coordinator" method in LockSet interface
has been ruined,why?

[Lars Kühne]

Logged In: YES
user_id=401384

Unfortunately, I don't have any experience with the
ConcurrencyControlService.

You seem to already have had a look at the sources - is
there any hopy you could develop a patch that fixes these
problems?

[Viacheslav Tararin]

Logged In: YES
user_id=422452

Yes, all notes is suitable. But for fix it, we need fully
rewrite ConcurrencyControlService. According for this: "Is
anydoby how use the service?"