mustafasrepo commented on PR #9221: URL: https://github.com/apache/arrow-datafusion/pull/9221#issuecomment-1956008485
After thinking about the possible solutions, another approach might be for the table below a | -- | 1 2 null 3 null 4 constructing a vector with same length where each entry contains the index of the previous non-null entry. For the table above, this would be non-null_pointers | -- | -1 0 0 1 1 3 To find `LAG(2)` for the row=5 (where `a=4`). We would follow pointers twice (idx 5 -> 3, 3 ->1). Hence result would the value at index 1, which have `a=2`. For the `LEAD` we might need to reverse data, apply `LAG` as you suggest. If we encounter -1, that would indicate result is `None`. Because there is no more previous data. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
