[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Issue - 1: Non-permeable scopes
All this talk about links reminds me of our first issue, non-permeable scope semantics. To recap, adopting non-permeable scopes would mean that a link status would not be visible outside a scope boundry until that scope terminates. This amounts to an "isolation" property for links (where the scope boundry behaves like a traditional nested transaction). The primary benefit of this would be that in case of a fault, any link exiting the faulting scope would always be negative, which would be more in line with the behavior of deadpath elimination and transition condition evaluation (in case of failure or elimination, the link status is always negative). This would also encourage the proper use of fault handlers as an "undo" mechanism. With permeable scopes, it is very easy to write processes that "see into" a faulted scope, which works to defeat the whole fault handling and compensation mechanism. -Maciej
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]