mbutrovich commented on code in PR #1511:
URL: https://github.com/apache/datafusion-comet/pull/1511#discussion_r2018762152
##########
native/core/src/execution/shuffle/shuffle_writer.rs:
##########
@@ -422,27 +432,29 @@ impl ShuffleRepartitioner {
.collect::<Result<Vec<_>>>()?;
// use identical seed as spark hash partition
- let hashes_buf = &mut self.hashes_buf[..arrays[0].len()];
+ let hashes_buf = &mut
scratch.hashes_buf[..arrays[0].len()];
hashes_buf.fill(42_u32);
// Hash arrays and compute buckets based on number of
partitions
- let partition_ids = &mut
self.partition_ids[..arrays[0].len()];
+ let partition_ids = &mut
scratch.partition_ids[..arrays[0].len()];
create_murmur3_hashes(&arrays, hashes_buf)?
.iter()
.enumerate()
.for_each(|(idx, hash)| {
- partition_ids[idx] = pmod(*hash,
*num_output_partitions) as u64
+ partition_ids[idx] = pmod(*hash,
*num_output_partitions) as u32;
});
// count each partition size
- let mut partition_counters = vec![0usize;
*num_output_partitions];
+ let partition_counters = &mut scratch.partition_starts;
+ partition_counters.resize(num_output_partitions + 1, 0);
+ partition_counters.fill(0);
partition_ids
.iter()
.for_each(|partition_id|
partition_counters[*partition_id as usize] += 1);
// accumulate partition counters into partition ends
// e.g. partition counter: [1, 3, 2, 1] => [1, 4, 6, 7]
Review Comment:
Agreed. I don't think that is any more readable. Thank you for the
explanation!
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
