tustvold commented on code in PR #2680:
URL: https://github.com/apache/arrow-rs/pull/2680#discussion_r965630634
##########
arrow/src/compute/kernels/sort.rs:
##########
@@ -311,41 +312,26 @@ pub fn sort_to_indices(
)));
}
},
- DataType::Dictionary(key_type, value_type)
- if *value_type.as_ref() == DataType::Utf8 =>
- {
- match key_type.as_ref() {
- DataType::Int8 => {
- sort_string_dictionary::<Int8Type>(values, v, n, &options,
limit)
- }
- DataType::Int16 => {
- sort_string_dictionary::<Int16Type>(values, v, n,
&options, limit)
- }
- DataType::Int32 => {
- sort_string_dictionary::<Int32Type>(values, v, n,
&options, limit)
- }
- DataType::Int64 => {
- sort_string_dictionary::<Int64Type>(values, v, n,
&options, limit)
- }
- DataType::UInt8 => {
- sort_string_dictionary::<UInt8Type>(values, v, n,
&options, limit)
- }
- DataType::UInt16 => {
- sort_string_dictionary::<UInt16Type>(values, v, n,
&options, limit)
- }
- DataType::UInt32 => {
- sort_string_dictionary::<UInt32Type>(values, v, n,
&options, limit)
- }
- DataType::UInt64 => {
- sort_string_dictionary::<UInt64Type>(values, v, n,
&options, limit)
- }
- t => {
- return Err(ArrowError::ComputeError(format!(
- "Sort not supported for dictionary key type {:?}",
- t
- )));
- }
- }
+ DataType::Dictionary(_, _) => {
Review Comment:
NGL I worry a bit that this has the combinatorial fanout that absolutely
tanks compile times... Perhaps we could compute the sort order of the
dictionary values and then use this to compare the keys? This might even be
faster
--
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]
