[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [wsbpel] Issue 67 - Proposal for vote (restated)
Goran, Good questions. A bit of history, I think, is required. The original submission contained (and still contains) a statement about serializable scopes working in a fashion that is analogous to serializable database transaction isolation. Some of my product engineers wondered if this meant something like optimistic locking was permitted, in order to increase concurrent processing in certain cases. If this was so, shouldn't there be a standard BPEL-defined fault for optimistic lock failure when a scope completes? When I queried this on the list, it soon became clear that the authors' intent was not to allow such an "optimistic" approach, and that additional language was required to clarify this. Thus Satish's extra words, which will help avoid such intepretation problems in the future. Simply saying "serializable scopes shall not deadlock" is insufficient; the proposal clarifies this, without, I believe, unduly restricting implementations in this area. -Ron Goran Olsson wrote: >I'm a bit confused here. Why should the specification how a certain concept >is implemented? >Should it not just say " Note:that serialization means that serializable >scopes will not deadlock."? >I would have assumed that it is up to implementors to use whatever mechanism >they decide to ensure this. >Thanks, Goran > >----- Original Message ----- >From: "Satish Thatte" <satisht@microsoft.com> >To: <wsbpel@lists.oasis-open.org> >Sent: Friday, January 16, 2004 9:38 AM >Subject: [wsbpel] Issue 67 - Proposal for vote (restated) > > >As Ron wrote, the extra sentence he proposed adding led to some >potential confusion about implications for implementation. He has >withdrawn his suggestion. I would therefore like to simply restate my >proposal in its original form as the current proposal for a vote. > >I propose that we add the following sentences to section to Section 13.6 >following paragraph 2. > >"Note that serialization of variable access cannot lead to internal >deadlock in a BPEL process instance. The reason being that, >conceptually, a serializable scope is not started until it can gain >sufficiently exclusive access to all the non-local variables it needs." > >Satish > > > >To unsubscribe from this mailing list (and be removed from the roster of the >OASIS TC), go to >http://www.oasis-open.org/apps/org/workgroup/wsbpel/members/leave_workgroup.php. > > > >To unsubscribe from this mailing list (and be removed from the roster of the OASIS TC), go to http://www.oasis-open.org/apps/org/workgroup/wsbpel/members/leave_workgroup.php. > > >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]