Pipeline::dispose() currently takes OperationContext* parameter. According to the method's comment this is necessary, because a pipeline can be disposed when the pipeline's own operating context is for some reason not valid anymore, but it is hard to spot / understand such cases.

In cases when it is actually so, we could probably call Pipeline::reattachToOperationContext(OperationContext*) before calling Pipeline::dispose(OperationContext*). 're-attaching' does some additional steps:

• propagate to the stages
• validate

The advantage of such change would be to mark explicitly the cases when the operation context needs to be re-attached.