Why is it difficult to control output via LIMIT in the `volcano model` and `Bottom-to-Top`? For example, in the `volcano model`, just like: ``` LIMIT N, M i = 0 for t in child. Next(): if i > M: return else if i < N: i++ continue emit(projection(t)) i++ ```
I think you got something mixed up. He mentioned that its easy to control output on the volcano model because you can just add a limit operator that calls next() x amount of times
24:00 doesn't ternary operator include branch as well? it should be `delta = (key > low) & (key < high)`?
Yes, you are technically correct. The compiler will simplify that down to just the comparison instructions as you point out.
1:04:00 why not do projection before select?
Postgres does use likely/unlikely.
Year, and some other OLAPs like Clickhouse, Doris do it too.
Why is it difficult to control output via LIMIT in the `volcano model` and `Bottom-to-Top`? For example, in the `volcano model`, just like:
```
LIMIT N, M
i = 0
for t in child. Next():
if i > M:
return
else if i < N:
i++
continue
emit(projection(t))
i++
```
I think you got something mixed up. He mentioned that its easy to control output on the volcano model because you can just add a limit operator that calls next() x amount of times