chore: remove used_coding as it is not used currently

rust/lance-encoding/src/data.rs

@@ -15,7 +15,6 @@
 //! and decoders easier (since they can rely on a normalized representation)
 
 use std::{
-    collections::HashSet,
     ops::Range,
     sync::{Arc, RwLock},
 };
@@ -31,17 +30,6 @@ use lance_core::{Error, Result};
 
 use crate::{buffer::LanceBuffer, statistics::Stat};
 
-/// `Encoding` enum serves as a encoding registration center.
-///
-/// All the encodings added to Lance should register here, and
-/// these encodings can be dynamically selected during encoding,
-/// users can also specify the particular encoding they want to use in the field metadata.
-#[derive(Eq, Hash, PartialEq, Debug)]
-pub enum Encoding {
-    Bitpack,
-    Fsst,
-    FixedSizeBinary,
-}
 /// A data block with no buffers where everything is null
 ///
 /// Note: this data block should not be used for future work.  It will be deprecated
@@ -101,29 +89,6 @@ impl PartialEq for BlockInfo {
         *self_info == *other_info
     }
 }
-// `UsedEncoding` is used to record the encodings that has applied to a `DataBlock`
-#[derive(Debug, Clone)]
-pub struct UsedEncoding(Arc<RwLock<HashSet<Encoding>>>);
-
-impl Default for UsedEncoding {
-    fn default() -> Self {
-        Self::new()
-    }
-}
-
-impl UsedEncoding {
-    pub fn new() -> Self {
-        Self(Arc::new(RwLock::new(HashSet::new())))
-    }
-}
-
-impl PartialEq for UsedEncoding {
-    fn eq(&self, other: &Self) -> bool {
-        let self_used = self.0.read().unwrap();
-        let other_used = other.0.read().unwrap();
-        *self_used == *other_used
-    }
-}
 
 /// Wraps a data block and adds nullability information to it
 ///
@@ -138,8 +103,6 @@ pub struct NullableDataBlock {
     pub nulls: LanceBuffer,
 
     pub block_info: BlockInfo,
-
-    pub used_encoding: UsedEncoding,
 }
 
 impl NullableDataBlock {
@@ -165,7 +128,6 @@ impl NullableDataBlock {
             data: Box::new(self.data.borrow_and_clone()),
             nulls: self.nulls.borrow_and_clone(),
             block_info: self.block_info.clone(),
-            used_encoding: self.used_encoding.clone(),
         }
     }
 
@@ -174,7 +136,6 @@ impl NullableDataBlock {
             data: Box::new(self.data.try_clone()?),
             nulls: self.nulls.try_clone()?,
             block_info: self.block_info.clone(),
-            used_encoding: self.used_encoding.clone(),
         })
     }
 
@@ -233,8 +194,6 @@ pub struct FixedWidthDataBlock {
     pub num_values: u64,
 
     pub block_info: BlockInfo,
-
-    pub used_encoding: UsedEncoding,
 }
 
 impl FixedWidthDataBlock {
@@ -271,7 +230,6 @@ impl FixedWidthDataBlock {
             bits_per_value: self.bits_per_value,
             num_values: self.num_values,
             block_info: self.block_info.clone(),
-            used_encoding: self.used_encoding.clone(),
         }
     }
 
@@ -281,7 +239,6 @@ impl FixedWidthDataBlock {
             bits_per_value: self.bits_per_value,
             num_values: self.num_values,
             block_info: self.block_info.clone(),
-            used_encoding: self.used_encoding.clone(),
         })
     }
 
@@ -323,7 +280,6 @@ impl DataBlockBuilderImpl for FixedWidthDataBlockBuilder {
             bits_per_value: self.bits_per_value,
             num_values,
             block_info: BlockInfo::new(),
-            used_encoding: UsedEncoding::new(),
         })
     }
 }
@@ -444,7 +400,6 @@ pub struct OpaqueBlock {
     pub buffers: Vec<LanceBuffer>,
     pub num_values: u64,
     pub block_info: BlockInfo,
-    pub used_encoding: UsedEncoding,
 }
 
 impl OpaqueBlock {
@@ -457,7 +412,6 @@ impl OpaqueBlock {
                 .collect(),
             num_values: self.num_values,
             block_info: self.block_info.clone(),
-            used_encoding: self.used_encoding.clone(),
         }
     }
 
@@ -470,7 +424,6 @@ impl OpaqueBlock {
                 .collect::<Result<_>>()?,
             num_values: self.num_values,
             block_info: self.block_info.clone(),
-            used_encoding: self.used_encoding.clone(),
         })
     }
 
@@ -494,8 +447,6 @@ pub struct VariableWidthBlock {
     pub num_values: u64,
 
     pub block_info: BlockInfo,
-
-    pub used_encodings: UsedEncoding,
 }
 
 impl VariableWidthBlock {
@@ -525,7 +476,6 @@ impl VariableWidthBlock {
             bits_per_offset: self.bits_per_offset,
             num_values: self.num_values,
             block_info: self.block_info.clone(),
-            used_encodings: self.used_encodings.clone(),
         }
     }
 
@@ -536,7 +486,6 @@ impl VariableWidthBlock {
             bits_per_offset: self.bits_per_offset,
             num_values: self.num_values,
             block_info: self.block_info.clone(),
-            used_encodings: self.used_encodings.clone(),
         })
     }
 
@@ -989,7 +938,6 @@ fn arrow_binary_to_data_block(
         bits_per_offset,
         num_values,
         block_info: BlockInfo::new(),
-        used_encodings: UsedEncoding::new(),
     })
 }
 
@@ -1155,15 +1103,13 @@ fn arrow_dictionary_to_data_block(arrays: &[ArrayRef], validity: Option<NullBuff
             bits_per_value: bits_per_index,
             num_values,
             block_info: BlockInfo::new(),
-            used_encoding: UsedEncoding::new(),
         }
     } else {
         FixedWidthDataBlock {
             data: LanceBuffer::Borrowed(indices.to_data().buffers()[0].clone()),
             bits_per_value: indices.data_type().byte_width() as u64 * 8,
             num_values,
             block_info: BlockInfo::new(),
-            used_encoding: UsedEncoding::new(),
         }
     };
 
@@ -1247,7 +1193,6 @@ impl DataBlock {
                     bits_per_value: 1,
                     num_values,
                     block_info: BlockInfo::new(),
-                    used_encoding: UsedEncoding::new(),
                 })
             }
             DataType::Date32
@@ -1277,7 +1222,6 @@ impl DataBlock {
                     bits_per_value: data_type.byte_width() as u64 * 8,
                     num_values,
                     block_info: BlockInfo::new(),
-                    used_encoding: UsedEncoding::new(),
                 })
             }
             DataType::Null => Self::AllNull(AllNullDataBlock { num_values }),
@@ -1325,7 +1269,6 @@ impl DataBlock {
                     data: Box::new(encoded),
                     nulls: LanceBuffer::Borrowed(nulls.into_inner().into_inner()),
                     block_info: BlockInfo::new(),
-                    used_encoding: UsedEncoding::new(),
                 }),
                 _ => unreachable!(),
             }

rust/lance-encoding/src/encodings/logical/list.rs

@@ -20,7 +20,7 @@ use lance_core::{cache::FileMetadataCache, Error, Result};
 
 use crate::{
     buffer::LanceBuffer,
-    data::{BlockInfo, DataBlock, FixedWidthDataBlock, UsedEncoding},
+    data::{BlockInfo, DataBlock, FixedWidthDataBlock},
     decoder::{
         DecodeArrayTask, DecodeBatchScheduler, FieldScheduler, FilterExpression, ListPriorityRange,
         LogicalPageDecoder, MessageType, NextDecodeTask, PageEncoding, PriorityRange,
@@ -1083,7 +1083,6 @@ impl ListOffsetsEncoder {
             data: LanceBuffer::reinterpret_vec(offsets),
             num_values: num_offsets,
             block_info: BlockInfo::new(),
-            used_encoding: UsedEncoding::new(),
         });
         inner_encoder.encode(offsets_data, &DataType::UInt64, buffer_index)
     }

rust/lance-encoding/src/encodings/logical/primitive.rs

@@ -16,7 +16,7 @@ use lance_core::{datatypes::Field, utils::tokio::spawn_cpu, Result};
 
 use crate::{
     buffer::LanceBuffer,
-    data::{BlockInfo, DataBlock, DataBlockBuilder, FixedWidthDataBlock, UsedEncoding},
+    data::{BlockInfo, DataBlock, DataBlockBuilder, FixedWidthDataBlock},
     decoder::{
         BlockDecompressor, ColumnInfo, DecodeArrayTask, DecodePageTask, DecodedArray, DecodedPage,
         DecompressorStrategy, FieldScheduler, FilterExpression, LoadedPage, LogicalPageDecoder,
@@ -1643,7 +1643,6 @@ impl PrimitiveStructuralEncoder {
                 bits_per_value: 16,
                 num_values,
                 block_info: BlockInfo::new(),
-                used_encoding: UsedEncoding::new(),
             });
             let levels_field = Field::new_arrow("", DataType::UInt16, false)?;
             // Pick a block compressor
@@ -1663,7 +1662,6 @@ impl PrimitiveStructuralEncoder {
                     bits_per_value: 16,
                     num_values: chunk_num_values,
                     block_info: BlockInfo::new(),
-                    used_encoding: UsedEncoding::new(),
                 });
                 let compressed_levels = compressor.compress(chunk_levels_block)?;
                 off += level_bytes;

rust/lance-encoding/src/encodings/physical/basic.rs

@@ -8,7 +8,7 @@ use futures::{future::BoxFuture, FutureExt};
 use log::trace;
 
 use crate::{
-    data::{AllNullDataBlock, BlockInfo, DataBlock, NullableDataBlock, UsedEncoding},
+    data::{AllNullDataBlock, BlockInfo, DataBlock, NullableDataBlock},
     decoder::{PageScheduler, PrimitivePageDecoder},
     encoder::{ArrayEncoder, EncodedArray},
     format::ProtobufUtils,
@@ -166,7 +166,6 @@ impl PrimitivePageDecoder for BasicPageDecoder {
                     data: Box::new(values),
                     nulls: validity.data,
                     block_info: BlockInfo::new(),
-                    used_encoding: UsedEncoding::new(),
                 }))
             }
             DataNullStatus::All => Ok(DataBlock::AllNull(AllNullDataBlock {
@@ -218,7 +217,6 @@ impl ArrayEncoder for BasicEncoder {
                     data: Box::new(encoded_values.data),
                     nulls: nullable.nulls,
                     block_info: BlockInfo::new(),
-                    used_encoding: UsedEncoding::new(),
                 });
                 Ok(EncodedArray {
                     data: encoded,

rust/lance-encoding/src/encodings/physical/binary.rs

@@ -16,7 +16,7 @@ use crate::encodings::logical::primitive::PrimitiveFieldDecoder;
 
 use crate::buffer::LanceBuffer;
 use crate::data::{
-    BlockInfo, DataBlock, FixedWidthDataBlock, NullableDataBlock, UsedEncoding, VariableWidthBlock,
+    BlockInfo, DataBlock, FixedWidthDataBlock, NullableDataBlock, VariableWidthBlock,
 };
 use crate::format::ProtobufUtils;
 use crate::{
@@ -327,14 +327,12 @@ impl PrimitivePageDecoder for BinaryPageDecoder {
             num_values: num_rows,
             offsets: LanceBuffer::from(offsets_buffer),
             block_info: BlockInfo::new(),
-            used_encodings: UsedEncoding::new(),
         });
         if let Some(validity) = validity_buffer {
             Ok(DataBlock::Nullable(NullableDataBlock {
                 data: Box::new(string_data),
                 nulls: LanceBuffer::from(validity),
                 block_info: BlockInfo::new(),
-                used_encoding: UsedEncoding::new(),
             }))
         } else {
             Ok(string_data)
@@ -372,7 +370,6 @@ impl BinaryEncoder {
                 num_values,
                 offsets: LanceBuffer::reinterpret_vec(vec![0_u32; num_values as usize + 1]),
                 block_info: BlockInfo::new(),
-                used_encodings: UsedEncoding::new(),
             }
         } else {
             VariableWidthBlock {
@@ -381,7 +378,6 @@ impl BinaryEncoder {
                 num_values,
                 offsets: LanceBuffer::reinterpret_vec(vec![0_u64; num_values as usize + 1]),
                 block_info: BlockInfo::new(),
-                used_encodings: UsedEncoding::new(),
             }
         }
     }
@@ -422,7 +418,6 @@ fn get_indices_from_string_arrays(
                 data: LanceBuffer::empty(),
                 num_values: 0,
                 block_info: BlockInfo::new(),
-                used_encoding: UsedEncoding::new(),
             }),
             0,
         );
@@ -452,7 +447,6 @@ fn get_indices_from_string_arrays(
         data: LanceBuffer::reinterpret_vec(indices),
         num_values: num_rows as u64,
         block_info: BlockInfo::new(),
-        used_encoding: UsedEncoding::new(),
     });
     (indices, null_adjustment)
 }
@@ -505,7 +499,6 @@ impl ArrayEncoder for BinaryEncoder {
             data: data.data,
             num_values: data.num_values,
             block_info: BlockInfo::new(),
-            used_encodings: UsedEncoding::new(),
         });
 
         let bytes_buffer_index = *buffer_index;

rust/lance-encoding/src/encodings/physical/bitmap.rs

@@ -12,7 +12,7 @@ use log::trace;
 
 use crate::{
     buffer::LanceBuffer,
-    data::{BlockInfo, DataBlock, FixedWidthDataBlock, UsedEncoding},
+    data::{BlockInfo, DataBlock, FixedWidthDataBlock},
     decoder::{PageScheduler, PrimitivePageDecoder},
     EncodingsIo,
 };
@@ -119,7 +119,6 @@ impl PrimitivePageDecoder for BitmapDecoder {
             bits_per_value: 1,
             num_values: num_rows,
             block_info: BlockInfo::new(),
-            used_encoding: UsedEncoding::new(),
         }))
     }
 }

rust/lance-encoding/src/encodings/physical/bitpack.rs

@@ -20,7 +20,7 @@ use lance_arrow::DataTypeExt;
 use lance_core::{Error, Result};
 
 use crate::buffer::LanceBuffer;
-use crate::data::{BlockInfo, DataBlock, FixedWidthDataBlock, UsedEncoding};
+use crate::data::{BlockInfo, DataBlock, FixedWidthDataBlock};
 use crate::decoder::{PageScheduler, PrimitivePageDecoder};
 use crate::encoder::{ArrayEncoder, EncodedArray};
 use crate::format::ProtobufUtils;
@@ -159,7 +159,6 @@ impl ArrayEncoder for BitpackedArrayEncoder {
             data: LanceBuffer::Owned(dst_buffer),
             num_values: unpacked.num_values,
             block_info: BlockInfo::new(),
-            used_encoding: UsedEncoding::new(),
         });
 
         let bitpacked_buffer_index = *buffer_index;
@@ -489,7 +488,6 @@ impl PrimitivePageDecoder for BitpackedPageDecoder {
             bits_per_value: self.uncompressed_bits_per_value,
             num_values: num_rows,
             block_info: BlockInfo::new(),
-            used_encoding: UsedEncoding::new(),
         }))
     }
 }