break: remove roaring int/bool & run end bool arrays (#2020)

Roaring arrays require deserializing on each access (slow), and indices
are often well-compressed by other methods (e.g., FastLanes delta).

For validity, we're moving towards a more specialized approach in #2019
and away from the vanilla compressed array approach

Author: lwwmanning

Cargo.lock

@@ -153,9 +153,7 @@ vortex-io = { version = "0.21.1", path = "./vortex-io" }
 vortex-ipc = { version = "0.21.1", path = "./vortex-ipc" }
 vortex-layout = { version = "0.21.1", path = "./vortex-layout" }
 vortex-proto = { version = "0.21.1", path = "./vortex-proto" }
-vortex-roaring = { version = "0.21.1", path = "./encodings/roaring" }
 vortex-runend = { version = "0.21.1", path = "./encodings/runend" }
-vortex-runend-bool = { version = "0.21.1", path = "./encodings/runend-bool" }
 vortex-scalar = { version = "0.21.1", path = "./vortex-scalar", default-features = false }
 vortex-scan = { version = "0.21.1", path = "./vortex-scan", default-features = false }
 vortex-sampling-compressor = { version = "0.21.1", path = "./vortex-sampling-compressor" }

Cargo.toml

@@ -4,9 +4,9 @@ use rand::distributions::Alphanumeric;
 use rand::seq::SliceRandom as _;
 use rand::{thread_rng, Rng, SeedableRng as _};
 use vortex::aliases::hash_set::HashSet;
-use vortex::array::{ConstantArray, VarBinViewArray};
+use vortex::array::VarBinViewArray;
 use vortex::buffer::Buffer;
-use vortex::compute::{compare, try_cast, Operator};
+use vortex::compute::try_cast;
 use vortex::dtype::PType;
 use vortex::encodings::dict::dict_encode;
 use vortex::encodings::fsst::{fsst_compress, fsst_train_compressor};
@@ -18,14 +18,10 @@ use vortex::sampling_compressor::compressors::bitpacked::{
 use vortex::sampling_compressor::compressors::delta::DeltaCompressor;
 use vortex::sampling_compressor::compressors::dict::DictCompressor;
 use vortex::sampling_compressor::compressors::r#for::FoRCompressor;
-use vortex::sampling_compressor::compressors::roaring_bool::RoaringBoolCompressor;
-use vortex::sampling_compressor::compressors::roaring_int::RoaringIntCompressor;
 use vortex::sampling_compressor::compressors::runend::DEFAULT_RUN_END_COMPRESSOR;
-use vortex::sampling_compressor::compressors::runend_bool::RunEndBoolCompressor;
 use vortex::sampling_compressor::compressors::zigzag::ZigZagCompressor;
 use vortex::sampling_compressor::compressors::CompressorRef;
 use vortex::sampling_compressor::SamplingCompressor;
-use vortex::scalar::Scalar;
 use vortex::{IntoArrayData as _, IntoCanonical, ToArrayData};
 
 #[global_allocator]
@@ -41,16 +37,6 @@ fn primitive(c: &mut Criterion) {
     let uint_array =
         Buffer::from_iter((0..num_values).map(|_| rng.gen_range(0u32..256))).into_array();
     let int_array = try_cast(uint_array.clone(), PType::I32.into()).unwrap();
-
-    let bool_array = compare(
-        &uint_array,
-        ConstantArray::new(Scalar::from(0u32), uint_array.len()),
-        Operator::Eq,
-    )
-    .unwrap();
-
-    let index_array = Buffer::from_iter((0..num_values).map(|i| (i * 2) as u32 + 42)).into_array();
-
     let float_array = try_cast(uint_array.clone(), PType::F32.into()).unwrap();
 
     let compressors_names_and_arrays = [
@@ -63,9 +49,6 @@ fn primitive(c: &mut Criterion) {
         (&DEFAULT_RUN_END_COMPRESSOR, "runend", &uint_array),
         (&DeltaCompressor, "delta", &uint_array),
         (&DictCompressor, "dict", &uint_array),
-        (&RoaringBoolCompressor, "roaring_bool", &bool_array),
-        (&RoaringIntCompressor, "roaring_int", &index_array),
-        (&RunEndBoolCompressor, "runend_bool", &bool_array),
         (&FoRCompressor, "frame_of_reference", &int_array),
         (&ZigZagCompressor, "zigzag", &int_array),
         (&ALPCompressor, "alp", &float_array),

bench-vortex/benches/compressor_throughput.rs

@@ -20,9 +20,7 @@ use vortex::encodings::datetime_parts::DateTimePartsArray;
 use vortex::encodings::dict::DictArray;
 use vortex::encodings::fastlanes::{BitPackedArray, DeltaArray, FoRArray};
 use vortex::encodings::fsst::{fsst_compress, fsst_train_compressor};
-use vortex::encodings::roaring::{Bitmap, RoaringBoolArray, RoaringIntArray};
 use vortex::encodings::runend::RunEndArray;
-use vortex::encodings::runend_bool::RunEndBoolArray;
 use vortex::encodings::zigzag::ZigZagArray;
 use vortex::scalar::Scalar;
 use vortex::validity::Validity;
@@ -118,18 +116,9 @@ fn enc_impls() -> Vec<ArrayData> {
         .into_array(),
         NullArray::new(10).into_array(),
         buffer![0, 1].into_array(),
-        RoaringBoolArray::try_new(Bitmap::from([0u32, 10, 20]), 30)
-            .unwrap()
-            .into_array(),
-        RoaringIntArray::try_new(Bitmap::from([5u32, 6, 8]), PType::U32)
-            .unwrap()
-            .into_array(),
         RunEndArray::try_new(buffer![5u32, 8].into_array(), buffer![0, 1].into_array())
             .unwrap()
             .into_array(),
-        RunEndBoolArray::try_new(buffer![5u32, 8].into_array(), true, Validity::NonNullable)
-            .unwrap()
-            .into_array(),
         SparseArray::try_new(
             buffer![5u64, 8].into_array(),
             PrimitiveArray::new(buffer![3u32, 6], Validity::AllValid).into_array(),