Interpreting activties in timeline about eventual HOT 4 OPEN

Isn't the maximum size of a "group" 2? So it's not really that much extra processing - the finds are constant time

from eventual.

The code is matching activities right now, but don't see why it can't be broadened to match any schedule/end event, using the same heuristic style as your sampel code. While I do generally like more declarative code, I think in this case doing the grouping / find thing results in a lot of extra processing, doing a bunch of passes over nested lists. It can just be done in one pass by building the output map as you go:

let spanningActivities = Record<seq, TimelineActivity>{}
const otherActivities = TimelineActivity[]
 events.forEach((event) => {
    if (match(event, isWorkflowStarted)) {
      start = event;
    } else if (match(event, isScheduledEvent)) {
      //creates timelineactivity with start time and event type
      spanningActivities[event.seq] = {type: activityType(event.type), start: event.start}
    } else if (match(event, or(isCompletedEvent, isFailedEvent, isWorkflowTimedOut)) {
    //update timelineactivity with state and end time
     spanningActivities[event.seq].state = {status: activityStatus(event), end: event.end}
    } else if (event.match(or(isSignalReceived, isWorkflowStarted, isWorkflowFailed, isWorkflowCompleted, isWorkflowTimedOut)) {
    otherActivities.push(activityOf(event))
    }
}

return  [...Object.entries(spanningActivities), ...otherActivities]

As far as I can tell the outcome is the same, just a bit lower level, a bit more verbose, but saving a bunch of computation.

from eventual.

After grouping, the nested lists should be small, I doubt you'd be able to come up with a realistic history where you'd tell the performance difference. The groupby is doing most of the work here.

Clean/correct code > micro-optimizations. Then optimize if there is an actual issue.

Your implementation:

1. relies on the order
2. will always take the last event that matches the condition (mine takes the first)
3. depends on both events existing in order (not fault tolerant)
4. doesn't filter out the start or end event from "other"
5. ignores the non-state/end events that may relate to an "Eventual"/activity.

If your stateful accumulator can resolve this issues (with more code), then I think it would be fine, but... correct > micro-optimized.

from eventual.

Maybe not max (ex: we may get a "timeout" and then later get the "complete") or we may have intra-activity events in the future like a heartbeat checkpoint that gets stored in the workflow.

Though I'd constrain the size of a group to an average of < magnitude 10. (aka essentially constant).

from eventual.