Try using Unroll, source field set to `events`

IIRC, EB works on _raw. not on arbitrary fields in the event

if events was a thing under _raw, you might see the results you were expecting

Progress!If I unroll onto _raw, each new event has its own events array, with all 105 siblings, essentially squaring the memory footprint.If I unroll onto events, now I just need to eval all child fields to top-level fields.Long shot question - is it possible to return multiple events from a code function?I can try modifying raw to get EB working, but I don't think that's a long-term solution, in case we need the OG event later for debugging.

Code: No, code works inside a single event context only

Valid.If I overwrite _raw in code with a *stringified* array, then EB will work. Seems silly to json parse and then stringify to json parse again, so unroll is likely the best option here.

def experiment with that. Hit the CPU profiler in the stats popout

0

unroll takes 87ms; total time 260ms; though more than half appears to be writing the pipeline results to the preview pane

Looks like parse -> stringify -> parse is most performant. Thanks for the hint on the profiler!

EB takes 53ms; increases code from 2 to 3 for the stringify; total time 103ms - less than half. IDK what `run Microtasks` is, but it takes a lot longer after `unroll`

woof, that's a lot of time!If you can cleanse your data, I'd love to get a sample to see what sort of optimizations i can come up with. let me know.