Github user yestinchen commented on a diff in the pull request: https://github.com/apache/flink/pull/4331#discussion_r129515022 --- Diff: flink-libraries/flink-cep/src/main/java/org/apache/flink/cep/nfa/NFA.java --- @@ -551,18 +579,47 @@ private boolean isSelfIgnore(final StateTransition<T> edge) { nextVersion, startTimestamp); } + + switch (skipStrategy.getStrategy()) { + case SKIP_PAST_LAST_ROW: + if (nextState.isFinal()) { + resultingComputationStates.add(createStartComputationState(computationState, outgoingEdges)); + } + break; + case SKIP_TO_FIRST: + if (nextState.getName().equals(skipStrategy.getRpv()) && + !nextState.getName().equals(currentState.getName())) { + ComputationState<T> startComputationState = createStartComputationState(computationState, outgoingEdges); + if (callLevel > 0) { + throw new RuntimeException("infinite loop! Will abort the match process, please rewrite your pattern query"); + } + // feed current matched event to the state. + Collection<ComputationState<T>> computationStates = computeNextStates(startComputationState, event, timestamp, callLevel++); --- End diff -- Because SKIP_TO_FIRST or SKIP_TO_LAST needs to start the next match process at the first or last matched event in specified pattern. For example, for a given event stream: `a1, b1, c1, a2` and a given match `(A B C)`. If we set the SkipStrategy to SKIP_TO_FIRST with a pattern name `B`, we should create a new `startComputationState` after `b1` is being processed. And the next match should start at event `b1`. So we need to manually feed `b1` to the newly created `startComputationState`.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---