Upgrade arrow2/convert and use native buffers for the tensor u8 types (…

…#1375)

* Use buffers for the tensor u8 types

Cargo.toml

@@ -54,8 +54,8 @@ rerun = { path = "crates/rerun", version = "=0.6.0-alpha.0", default-features =
 
 ahash = "0.8"
 anyhow = "1.0"
-arrow2 = "0.16"
-arrow2_convert = "0.4.2"
+arrow2 = "0.17"
+arrow2_convert = "0.5.0"
 bytemuck = { version = "1.11", features = ["extern_crate_alloc"] }
 cfg-if = "1.0"
 clap = "4.0"
@@ -82,9 +82,9 @@ mimalloc = "0.1.29"
 ndarray = "0.15"
 nohash-hasher = "0.2"
 parking_lot = "0.12"
-polars-core = "0.27.1"
-polars-lazy = "0.27.1"
-polars-ops = "0.27.1"
+polars-core = "0.29"
+polars-lazy = "0.29"
+polars-ops = "0.29"
 puffin = "0.14"
 rfd = { version = "0.11.3", default_features = false, features = [
   "xdg-portal",

crates/re_data_ui/src/image.rs

@@ -285,7 +285,7 @@ pub fn tensor_summary_ui_grid_contents(
             re_ui.grid_left_hand_label(ui, "Encoding");
             ui.label(format!(
                 "{} JPEG",
-                re_format::format_bytes(jpeg_bytes.num_bytes() as _),
+                re_format::format_bytes(jpeg_bytes.len() as _),
             ));
             ui.end_row();
         }

crates/re_log_types/src/component_types/mesh3d.rs

@@ -8,7 +8,6 @@ use arrow2_convert::{serialize::ArrowSerialize, ArrowDeserialize, ArrowField, Ar
 
 use crate::Component;
 
-use super::arrow_convert_shims::BinaryBuffer;
 use super::{FieldError, Vec4D};
 
 // ----------------------------------------------------------------------------
@@ -262,7 +261,7 @@ pub struct EncodedMesh3D {
 
     pub format: MeshFormat,
 
-    pub bytes: BinaryBuffer,
+    pub bytes: Buffer<u8>,
 
     /// four columns of an affine transformation matrix
     pub transform: [[f32; 3]; 4],
@@ -275,7 +274,7 @@ pub struct EncodedMesh3DArrow {
 
     pub format: MeshFormat,
 
-    pub bytes: BinaryBuffer,
+    pub bytes: Buffer<u8>,
 
     #[arrow_field(type = "arrow2_convert::field::FixedSizeVec<f32, 12>")]
     pub transform: Vec<f32>,

crates/re_log_types/src/component_types/mod.rs

@@ -17,7 +17,6 @@ use lazy_static::lazy_static;
 use crate::Component;
 
 mod arrow;
-mod arrow_convert_shims;
 mod bbox;
 mod class_id;
 mod color;

crates/re_log_types/src/component_types/tensor.rs

@@ -7,8 +7,6 @@ use arrow2_convert::{serialize::ArrowSerialize, ArrowDeserialize, ArrowField, Ar
 use crate::Component;
 use crate::{TensorDataType, TensorElement};
 
-use super::arrow_convert_shims::BinaryBuffer;
-
 // ----------------------------------------------------------------------------
 
 /// A unique id per [`Tensor`].
@@ -145,7 +143,7 @@ impl ArrowDeserialize for TensorId {
 #[derive(Clone, PartialEq, ArrowField, ArrowSerialize, ArrowDeserialize)]
 #[arrow_field(type = "dense")]
 pub enum TensorData {
-    U8(BinaryBuffer),
+    U8(Buffer<u8>),
     U16(Buffer<u16>),
     U32(Buffer<u32>),
     U64(Buffer<u64>),
@@ -159,7 +157,7 @@ pub enum TensorData {
     //F16(Vec<arrow2::types::f16>),
     F32(Buffer<f32>),
     F64(Buffer<f64>),
-    JPEG(BinaryBuffer),
+    JPEG(Buffer<u8>),
 }
 
 impl TensorData {
@@ -180,7 +178,7 @@ impl TensorData {
 
     pub fn size_in_bytes(&self) -> usize {
         match self {
-            Self::U8(buf) | Self::JPEG(buf) => buf.0.len(),
+            Self::U8(buf) | Self::JPEG(buf) => buf.len(),
             Self::U16(buf) => buf.len(),
             Self::U32(buf) => buf.len(),
             Self::U64(buf) => buf.len(),
@@ -979,7 +977,7 @@ impl DecodedTensor {
 
             TensorData::JPEG(buf) => {
                 use image::io::Reader as ImageReader;
-                let mut reader = ImageReader::new(std::io::Cursor::new(buf.0.as_slice()));
+                let mut reader = ImageReader::new(std::io::Cursor::new(buf.as_slice()));
                 reader.set_format(image::ImageFormat::Jpeg);
                 let img = {
                     crate::profile_scope!("decode_jpeg");

crates/re_log_types/src/data_table.rs

@@ -1165,7 +1165,7 @@ fn data_table_sizes_basics() {
         2_690_064,    // expected_num_bytes
     );
     expect(
-        DataCell::from_arrow("some_bools".into(), cell.to_arrow().slice(1, 1)),
+        DataCell::from_arrow("some_bools".into(), cell.to_arrow().sliced(1, 1)),
         10_000,    // num_rows
         2_690_064, // expected_num_bytes
     );
@@ -1182,7 +1182,7 @@ fn data_table_sizes_basics() {
         2_840_064,    // expected_num_bytes
     );
     expect(
-        DataCell::from_arrow("some_u64s".into(), cell.to_arrow().slice(1, 1)),
+        DataCell::from_arrow("some_u64s".into(), cell.to_arrow().sliced(1, 1)),
         10_000,    // num_rows
         2_680_064, // expected_num_bytes
     );
@@ -1205,7 +1205,7 @@ fn data_table_sizes_basics() {
     expect(
         DataCell::from_arrow(
             crate::component_types::Label::name(),
-            cell.to_arrow().slice(1, 1),
+            cell.to_arrow().sliced(1, 1),
         ),
         10_000,    // num_rows
         2_950_064, // expected_num_bytes
@@ -1229,7 +1229,7 @@ fn data_table_sizes_basics() {
     expect(
         DataCell::from_arrow(
             crate::component_types::Point2D::name(),
-            cell.to_arrow().slice(1, 1),
+            cell.to_arrow().sliced(1, 1),
         ),
         10_000,    // num_rows
         5_100_064, // expected_num_bytes
@@ -1253,7 +1253,7 @@ fn data_table_sizes_basics() {
     expect(
         DataCell::from_arrow(
             crate::component_types::Point2D::name(),
-            cell.to_arrow().slice(1, 1),
+            cell.to_arrow().sliced(1, 1),
         ),
         10_000,    // num_rows
         3_920_064, // expected_num_bytes
@@ -1289,7 +1289,7 @@ fn data_table_sizes_basics() {
     expect(
         DataCell::from_arrow(
             crate::component_types::Point2D::name(),
-            cell.to_arrow().slice(1, 1),
+            cell.to_arrow().sliced(1, 1),
         ),
         10_000,    // num_rows
         5_560_064, // expected_num_bytes

crates/re_log_types/src/size_bytes.rs

@@ -303,7 +303,7 @@ fn estimated_bytes_size(array: &dyn Array) -> usize {
             estimated_bytes_size(
                 array
                     .values()
-                    .slice(values_start, values_end - values_start)
+                    .sliced(values_start, values_end - values_start)
                     .as_ref(),
             ) + std::mem::size_of_val(array.offsets().as_slice())
                 + validity_size(array.validity())
@@ -372,7 +372,7 @@ fn estimated_bytes_size(array: &dyn Array) -> usize {
                         let values_end = offsets[idx_end] as usize;
                         fields.get(cur_ty as usize).map_or(0, |x| {
                             estimated_bytes_size(
-                                x.slice(values_start, values_end - values_start).as_ref(),
+                                x.sliced(values_start, values_end - values_start).as_ref(),
                             )
                         })
                     })
@@ -394,7 +394,7 @@ fn estimated_bytes_size(array: &dyn Array) -> usize {
                 array
                     .fields()
                     .iter()
-                    .map(|x| estimated_bytes_size(x.slice(0, array.len()).as_ref()))
+                    .map(|x| estimated_bytes_size(x.sliced(0, array.len()).as_ref()))
                     .sum::<usize>()
             };
 

crates/re_query/src/entity_view.rs

@@ -112,7 +112,7 @@ impl ComponentWithInstances {
 
         let arrow_ref = self.values.as_arrow_ref();
         (arrow_ref.len() > offset)
-            .then(|| arrow_ref.slice(offset, 1))
+            .then(|| arrow_ref.sliced(offset, 1))
             .or_else(|| {
                 re_log::error_once!("found corrupt cell -- mismatched number of instances");
                 None

crates/re_viewer/src/ui/view_bar_chart/ui.rs

@@ -57,7 +57,7 @@ pub(crate) fn view_bar_chart(
             for ((ent_path, instance_key), tensor) in &scene.charts {
                 let chart = match &tensor.data {
                     component_types::TensorData::U8(data) => {
-                        create_bar_chart(ent_path, instance_key, data.0.iter().copied())
+                        create_bar_chart(ent_path, instance_key, data.iter().copied())
                     }
                     component_types::TensorData::U16(data) => {
                         create_bar_chart(ent_path, instance_key, data.iter().copied())