Minor enhancements to EXPLAIN ANALYZE output formatting #18689
Description
Is your feature request related to a problem or challenge?
Here are several minor items to make EXPLAIN ANALYZE display nicer:
- Format large numeric values with human-readable units (e.g.,
10100 → 10.1 K,1532000 → 1.53 M) - Round time durations to two decimal places (e.g.,
elapsed_compute=11.295377 ms → 11.30 ms)
Demo(in datafusion-cli)
> explain analyze
select *
from generate_series(10000000) as t1(v1)
where v1 > 100;
+-------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| plan_type | plan |
+-------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Plan with Metrics | ProjectionExec: expr=[value@0 as v1], metrics=[output_rows=9999900, elapsed_compute=213.26µs, output_bytes=76.3 MB, output_batches=1221] |
| | CoalesceBatchesExec: target_batch_size=8192, metrics=[output_rows=9999900, elapsed_compute=69.983µs, output_bytes=76.3 MB, output_batches=1221] |
| | FilterExec: value@0 > 100, metrics=[output_rows=9999900, elapsed_compute=4.263662ms, output_bytes=76.3 MB, output_batches=1221, selectivity=100% (9999900/10000001)] |
| | RepartitionExec: partitioning=RoundRobinBatch(14), input_partitions=1, metrics=[output_rows=10000001, elapsed_compute=451.452µs, output_bytes=76.3 MB, output_batches=1221, spill_count=0, spilled_bytes=0.0 B, spilled_rows=0, fetch_time=11.416792ms, repartition_time=1ns, send_time=1.382266ms] |
| | LazyMemoryExec: partitions=1, batch_generators=[generate_series: start=0, end=10000000, batch_size=8192], metrics=[output_rows=10000001, elapsed_compute=11.295377ms, output_bytes=76.3 MB, output_batches=1221] |
| | |
+-------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1 row(s) fetched.
Elapsed 0.017 seconds.
From the demo output, I think output_rows, and the ratio numerator and denominator is applicable. e.g. output_rows = 1.31 M, and selectivity = 15% (15.11k / 100.32k), and all time measurements can be rounded.
Describe the solution you'd like
No response
Describe alternatives you've considered
No response
Additional context
No response