From 2ec8ce8057b58745d147e3124da8501dd927a3e7 Mon Sep 17 00:00:00 2001
From: Jeremy Leibs <jeremy@rerun.io>
Date: Mon, 9 Sep 2024 19:45:30 -0400
Subject: [PATCH] Switch ArrowDict to signed and fix static datatype

---
 crates/store/re_chunk/src/util.rs            | 8 ++++----
 crates/store/re_chunk_store/src/dataframe.rs | 2 +-
 crates/store/re_dataframe/src/range.rs       | 8 ++++----
 3 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/crates/store/re_chunk/src/util.rs b/crates/store/re_chunk/src/util.rs
index 1cde20e9d6a2..091f06962f6d 100644
--- a/crates/store/re_chunk/src/util.rs
+++ b/crates/store/re_chunk/src/util.rs
@@ -84,7 +84,7 @@ pub fn arrays_to_list_array(
 pub fn arrays_to_dictionary<Idx: Copy + Eq>(
     array_datatype: ArrowDatatype,
     arrays: &[Option<(Idx, &dyn ArrowArray)>],
-) -> Option<ArrowDictionaryArray<u32>> {
+) -> Option<ArrowDictionaryArray<i32>> {
     // Dedupe the input arrays based on the given primary key.
     let arrays_dense_deduped = arrays
         .iter()
@@ -96,7 +96,7 @@ pub fn arrays_to_dictionary<Idx: Copy + Eq>(
 
     // Compute the keys for the final dictionary, using that same primary key.
     let keys = {
-        let mut cur_key = 0u32;
+        let mut cur_key = 0i32;
         arrays
             .iter()
             .dedup_by_with_count(|lhs, rhs| {
@@ -140,7 +140,7 @@ pub fn arrays_to_dictionary<Idx: Copy + Eq>(
     };
 
     let datatype = ArrowDatatype::Dictionary(
-        arrow2::datatypes::IntegerType::UInt32,
+        arrow2::datatypes::IntegerType::Int32,
         std::sync::Arc::new(data.data_type().clone()),
         true, // is_sorted
     );
@@ -149,7 +149,7 @@ pub fn arrays_to_dictionary<Idx: Copy + Eq>(
     // unique values.
     ArrowDictionaryArray::try_new(
         datatype,
-        ArrowPrimitiveArray::<u32>::from(keys),
+        ArrowPrimitiveArray::<i32>::from(keys),
         data.to_boxed(),
     )
     .ok()
diff --git a/crates/store/re_chunk_store/src/dataframe.rs b/crates/store/re_chunk_store/src/dataframe.rs
index f348667b677f..2b2bc6705703 100644
--- a/crates/store/re_chunk_store/src/dataframe.rs
+++ b/crates/store/re_chunk_store/src/dataframe.rs
@@ -482,7 +482,7 @@ impl ChunkStore {
                                 archetype_name: None,
                                 archetype_field_name: None,
                                 component_name: *component_name,
-                                datatype: datatype.clone(),
+                                datatype: ArrowListArray::<i32>::default_datatype(datatype.clone()),
                                 is_static: true,
                             })
                         })
diff --git a/crates/store/re_dataframe/src/range.rs b/crates/store/re_dataframe/src/range.rs
index 3da271784eff..e56258bc01bc 100644
--- a/crates/store/re_dataframe/src/range.rs
+++ b/crates/store/re_dataframe/src/range.rs
@@ -91,7 +91,7 @@ impl RangeQueryHandle<'_> {
                         .map(|col| match col {
                             ColumnDescriptor::Component(mut descr) => {
                                 descr.datatype = ArrowDatatype::Dictionary(
-                                    arrow2::datatypes::IntegerType::UInt32,
+                                    arrow2::datatypes::IntegerType::Int32,
                                     descr.datatype.into(),
                                     true,
                                 );
@@ -311,7 +311,7 @@ impl RangeQueryHandle<'_> {
         // see if this ever becomes an issue before going down this road.
         //
         // TODO(cmc): Opportunities for parallelization, if it proves to be a net positive in practice.
-        let dict_arrays: HashMap<&ComponentColumnDescriptor, ArrowDictionaryArray<u32>> = {
+        let dict_arrays: HashMap<&ComponentColumnDescriptor, ArrowDictionaryArray<i32>> = {
             re_tracing::profile_scope!("queries");
 
             columns
@@ -612,7 +612,7 @@ mod tests {
                     })
                     .unwrap()
                     .as_any()
-                    .downcast_ref::<ArrowDictionaryArray<u32>>()
+                    .downcast_ref::<ArrowDictionaryArray<i32>>()
                     .unwrap()
                     .values()
                     .clone()
@@ -639,7 +639,7 @@ mod tests {
                     })
                     .unwrap()
                     .as_any()
-                    .downcast_ref::<ArrowDictionaryArray<u32>>()
+                    .downcast_ref::<ArrowDictionaryArray<i32>>()
                     .unwrap()
                     .values()
                     .clone()