24 ms·
I don't see causality as defined in this context as needing any extra "motivation". It's pretty clear that if an agent tries to do A and then B, your system pro
by arghbleargh 12y ago
I don't see causality as defined in this context as needing any extra "motivation". It's pretty clear that if an agent tries to do A and then B, your system probably should not first execute B and then A. Imagine if the instructions in one of your program threads were not guaranteed to be executed in order. You would never be able to write code that does anything meaningful that way.
- oct 12y agoThe causal ordering defined in the paper is stronger than that. It says that if process P performs event A and later sends a message to process Q, then any event that occurs in Q after Q receives the message causally follows A. I don't think the blog post was meant to imply that the order of events within a single process does not need to be respected, though it doesn't specifically distinguish the single process case.
- gioele 12y ago> Imagine if the instructions in one of your program threads were not guaranteed to be executed in order. Modern CPUs do not process instructions in the same order they are stored in memory. Out-of-order CPUs rely on the implicit order created by the dependency between instructions and operands. As in the Lamport paper, the order in which the instruction are performed is meant to preserve the order in which the effects are produced instead of the order in which the commands are given.