add impls

return iterator from get_vals method. This will allow to save on unnecessary vec allocations.

examples/custom_collector.rs

@@ -7,10 +7,11 @@
 // Of course, you can have a look at the tantivy's built-in collectors
 // such as the `CountCollector` for more examples.
 
+use fastfield_codecs::Column;
 // ---
 // Importing tantivy...
 use tantivy::collector::{Collector, SegmentCollector};
-use tantivy::fastfield::{DynamicFastFieldReader, FastFieldReader};
+use tantivy::fastfield::DynamicFastFieldReader;
 use tantivy::query::QueryParser;
 use tantivy::schema::{Field, Schema, FAST, INDEXED, TEXT};
 use tantivy::{doc, Index, Score, SegmentReader};
@@ -103,7 +104,7 @@ impl SegmentCollector for StatsSegmentCollector {
     type Fruit = Option<Stats>;
 
     fn collect(&mut self, doc: u32, _score: Score) {
-        let value = self.fast_field_reader.get(doc) as f64;
+        let value = self.fast_field_reader.get_val(doc as u64) as f64;
         self.stats.count += 1;
         self.stats.sum += value;
         self.stats.squared_sum += value * value;

examples/warmer.rs

@@ -2,8 +2,8 @@ use std::cmp::Reverse;
 use std::collections::{HashMap, HashSet};
 use std::sync::{Arc, RwLock, Weak};
 
+use fastfield_codecs::Column;
 use tantivy::collector::TopDocs;
-use tantivy::fastfield::FastFieldReader;
 use tantivy::query::QueryParser;
 use tantivy::schema::{Field, Schema, FAST, TEXT};
 use tantivy::{
@@ -52,7 +52,7 @@ impl Warmer for DynamicPriceColumn {
             let product_id_reader = segment.fast_fields().u64(self.field)?;
             let product_ids: Vec<ProductId> = segment
                 .doc_ids_alive()
-                .map(|doc| product_id_reader.get(doc))
+                .map(|doc| product_id_reader.get_val(doc as u64))
                 .collect();
             let mut prices_it = self.price_fetcher.fetch_prices(&product_ids).into_iter();
             let mut price_vals: Vec<Price> = Vec::new();

fastfield_codecs/benches/bench.rs

@@ -4,9 +4,9 @@ extern crate test;
 
 #[cfg(test)]
 mod tests {
-    use fastfield_codecs::bitpacked::{BitpackedReader, BitpackedSerializer};
-    use fastfield_codecs::blockwise_linear::{BlockwiseLinearReader, BlockwiseLinearSerializer};
-    use fastfield_codecs::linear::{LinearReader, LinearSerializer};
+    use fastfield_codecs::bitpacked::BitpackedCodec;
+    use fastfield_codecs::blockwise_linear::BlockwiseLinearCodec;
+    use fastfield_codecs::linear::LinearCodec;
     use fastfield_codecs::*;
 
     fn get_data() -> Vec<u64> {
@@ -25,16 +25,10 @@ mod tests {
     fn value_iter() -> impl Iterator<Item = u64> {
         0..20_000
     }
-    fn bench_get<
-        S: FastFieldCodecSerializer,
-        R: FastFieldCodecDeserializer + FastFieldDataAccess,
-    >(
-        b: &mut Bencher,
-        data: &[u64],
-    ) {
+    fn bench_get<Codec: FastFieldCodec>(b: &mut Bencher, data: &[u64]) {
         let mut bytes = vec![];
-        S::serialize(&mut bytes, &data).unwrap();
-        let reader = R::open_from_bytes(OwnedBytes::new(bytes)).unwrap();
+        Codec::serialize(&mut bytes, &data).unwrap();
+        let reader = Codec::open_from_bytes(OwnedBytes::new(bytes)).unwrap();
         b.iter(|| {
             let mut sum = 0u64;
             for pos in value_iter() {
@@ -45,10 +39,11 @@ mod tests {
             sum
         });
     }
-    fn bench_create<S: FastFieldCodecSerializer>(b: &mut Bencher, data: &[u64]) {
-        let mut bytes = vec![];
+    fn bench_create<Codec: FastFieldCodec>(b: &mut Bencher, data: &[u64]) {
+        let mut bytes = Vec::new();
         b.iter(|| {
-            S::serialize(&mut bytes, &data).unwrap();
+            bytes.clear();
+            Codec::serialize(&mut bytes, &data).unwrap();
         });
     }
 
@@ -57,32 +52,32 @@ mod tests {
     #[bench]
     fn bench_fastfield_bitpack_create(b: &mut Bencher) {
         let data: Vec<_> = get_data();
-        bench_create::<BitpackedSerializer>(b, &data);
+        bench_create::<BitpackedCodec>(b, &data);
     }
     #[bench]
     fn bench_fastfield_linearinterpol_create(b: &mut Bencher) {
         let data: Vec<_> = get_data();
-        bench_create::<LinearSerializer>(b, &data);
+        bench_create::<LinearCodec>(b, &data);
     }
     #[bench]
     fn bench_fastfield_multilinearinterpol_create(b: &mut Bencher) {
         let data: Vec<_> = get_data();
-        bench_create::<BlockwiseLinearSerializer>(b, &data);
+        bench_create::<BlockwiseLinearCodec>(b, &data);
     }
     #[bench]
     fn bench_fastfield_bitpack_get(b: &mut Bencher) {
         let data: Vec<_> = get_data();
-        bench_get::<BitpackedSerializer, BitpackedReader>(b, &data);
+        bench_get::<BitpackedCodec>(b, &data);
     }
     #[bench]
     fn bench_fastfield_linearinterpol_get(b: &mut Bencher) {
         let data: Vec<_> = get_data();
-        bench_get::<LinearSerializer, LinearReader>(b, &data);
+        bench_get::<LinearCodec>(b, &data);
     }
     #[bench]
     fn bench_fastfield_multilinearinterpol_get(b: &mut Bencher) {
         let data: Vec<_> = get_data();
-        bench_get::<BlockwiseLinearSerializer, BlockwiseLinearReader>(b, &data);
+        bench_get::<BlockwiseLinearCodec>(b, &data);
     }
     pub fn stats_from_vec(data: &[u64]) -> FastFieldStats {
         let min_value = data.iter().cloned().min().unwrap_or(0);

fastfield_codecs/src/bitpacked.rs

@@ -4,9 +4,7 @@ use common::BinarySerializable;
 use ownedbytes::OwnedBytes;
 use tantivy_bitpacker::{compute_num_bits, BitPacker, BitUnpacker};
 
-use crate::{
-    FastFieldCodecDeserializer, FastFieldCodecSerializer, FastFieldCodecType, FastFieldDataAccess,
-};
+use crate::{Column, FastFieldCodec, FastFieldCodecType};
 
 /// Depending on the field type, a different
 /// fast field is required.
@@ -14,32 +12,12 @@ use crate::{
 pub struct BitpackedReader {
     data: OwnedBytes,
     bit_unpacker: BitUnpacker,
-    pub min_value_u64: u64,
-    pub max_value_u64: u64,
-    pub num_vals: u64,
+    min_value_u64: u64,
+    max_value_u64: u64,
+    num_vals: u64,
 }
 
-impl FastFieldCodecDeserializer for BitpackedReader {
-    /// Opens a fast field given a file.
-    fn open_from_bytes(bytes: OwnedBytes) -> io::Result<Self> {
-        let footer_offset = bytes.len() - 24;
-        let (data, mut footer) = bytes.split(footer_offset);
-        let min_value = u64::deserialize(&mut footer)?;
-        let amplitude = u64::deserialize(&mut footer)?;
-        let num_vals = u64::deserialize(&mut footer)?;
-        let max_value = min_value + amplitude;
-        let num_bits = compute_num_bits(amplitude);
-        let bit_unpacker = BitUnpacker::new(num_bits);
-        Ok(BitpackedReader {
-            data,
-            bit_unpacker,
-            min_value_u64: min_value,
-            max_value_u64: max_value,
-            num_vals,
-        })
-    }
-}
-impl FastFieldDataAccess for BitpackedReader {
+impl Column for BitpackedReader {
     #[inline]
     fn get_val(&self, doc: u64) -> u64 {
         self.min_value_u64 + self.bit_unpacker.get(doc, &self.data)
@@ -111,12 +89,33 @@ impl<'a, W: Write> BitpackedSerializerLegacy<'a, W> {
     }
 }
 
-pub struct BitpackedSerializer {}
+pub struct BitpackedCodec;
 
-impl FastFieldCodecSerializer for BitpackedSerializer {
+impl FastFieldCodec for BitpackedCodec {
     /// The CODEC_TYPE is an enum value used for serialization.
     const CODEC_TYPE: FastFieldCodecType = FastFieldCodecType::Bitpacked;
 
+    type Reader = BitpackedReader;
+
+    /// Opens a fast field given a file.
+    fn open_from_bytes(bytes: OwnedBytes) -> io::Result<Self::Reader> {
+        let footer_offset = bytes.len() - 24;
+        let (data, mut footer) = bytes.split(footer_offset);
+        let min_value = u64::deserialize(&mut footer)?;
+        let amplitude = u64::deserialize(&mut footer)?;
+        let num_vals = u64::deserialize(&mut footer)?;
+        let max_value = min_value + amplitude;
+        let num_bits = compute_num_bits(amplitude);
+        let bit_unpacker = BitUnpacker::new(num_bits);
+        Ok(BitpackedReader {
+            data,
+            bit_unpacker,
+            min_value_u64: min_value,
+            max_value_u64: max_value,
+            num_vals,
+        })
+    }
+
     /// Serializes data with the BitpackedFastFieldSerializer.
     ///
     /// The serializer in fact encode the values by bitpacking
@@ -125,10 +124,7 @@ impl FastFieldCodecSerializer for BitpackedSerializer {
     /// It requires a `min_value` and a `max_value` to compute
     /// compute the minimum number of bits required to encode
     /// values.
-    fn serialize(
-        write: &mut impl Write,
-        fastfield_accessor: &dyn FastFieldDataAccess,
-    ) -> io::Result<()> {
+    fn serialize(write: &mut impl Write, fastfield_accessor: &dyn Column) -> io::Result<()> {
         let mut serializer = BitpackedSerializerLegacy::open(
             write,
             fastfield_accessor.min_value(),
@@ -142,29 +138,27 @@ impl FastFieldCodecSerializer for BitpackedSerializer {
 
         Ok(())
     }
-    fn is_applicable(_fastfield_accessor: &impl FastFieldDataAccess) -> bool {
-        true
-    }
-    fn estimate(fastfield_accessor: &impl FastFieldDataAccess) -> f32 {
+
+    fn estimate(fastfield_accessor: &impl Column) -> Option<f32> {
         let amplitude = fastfield_accessor.max_value() - fastfield_accessor.min_value();
         let num_bits = compute_num_bits(amplitude);
         let num_bits_uncompressed = 64;
-        num_bits as f32 / num_bits_uncompressed as f32
+        Some(num_bits as f32 / num_bits_uncompressed as f32)
     }
 }
 
 #[cfg(test)]
 mod tests {
     use super::*;
-    use crate::tests::get_codec_test_data_sets;
+    use crate::tests::get_codec_test_datasets;
 
     fn create_and_validate(data: &[u64], name: &str) {
-        crate::tests::create_and_validate::<BitpackedSerializer, BitpackedReader>(data, name);
+        crate::tests::create_and_validate::<BitpackedCodec>(data, name);
     }
 
     #[test]
     fn test_with_codec_data_sets() {
-        let data_sets = get_codec_test_data_sets();
+        let data_sets = get_codec_test_datasets();
         for (mut data, name) in data_sets {
             create_and_validate(&data, name);
             data.reverse();

fastfield_codecs/src/blockwise_linear.rs

@@ -18,9 +18,7 @@ use ownedbytes::OwnedBytes;
 use tantivy_bitpacker::{compute_num_bits, BitPacker, BitUnpacker};
 
 use crate::linear::{get_calculated_value, get_slope};
-use crate::{
-    FastFieldCodecDeserializer, FastFieldCodecSerializer, FastFieldCodecType, FastFieldDataAccess,
-};
+use crate::{Column, FastFieldCodec, FastFieldCodecType};
 
 const CHUNK_SIZE: u64 = 512;
 
@@ -148,18 +146,7 @@ fn get_interpolation_function(doc: u64, interpolations: &[Function]) -> &Functio
     &interpolations[get_interpolation_position(doc)]
 }
 
-impl FastFieldCodecDeserializer for BlockwiseLinearReader {
-    /// Opens a fast field given a file.
-    fn open_from_bytes(bytes: OwnedBytes) -> io::Result<Self> {
-        let footer_len: u32 = (&bytes[bytes.len() - 4..]).deserialize()?;
-        let footer_offset = bytes.len() - 4 - footer_len as usize;
-        let (data, mut footer) = bytes.split(footer_offset);
-        let footer = BlockwiseLinearFooter::deserialize(&mut footer)?;
-        Ok(BlockwiseLinearReader { data, footer })
-    }
-}
-
-impl FastFieldDataAccess for BlockwiseLinearReader {
+impl Column for BlockwiseLinearReader {
     #[inline]
     fn get_val(&self, idx: u64) -> u64 {
         let interpolation = get_interpolation_function(idx, &self.footer.interpolations);
@@ -191,15 +178,24 @@ impl FastFieldDataAccess for BlockwiseLinearReader {
 }
 
 /// Same as LinearSerializer, but working on chunks of CHUNK_SIZE elements.
-pub struct BlockwiseLinearSerializer {}
+pub struct BlockwiseLinearCodec;
 
-impl FastFieldCodecSerializer for BlockwiseLinearSerializer {
+impl FastFieldCodec for BlockwiseLinearCodec {
     const CODEC_TYPE: FastFieldCodecType = FastFieldCodecType::BlockwiseLinear;
+
+    type Reader = BlockwiseLinearReader;
+
+    /// Opens a fast field given a file.
+    fn open_from_bytes(bytes: OwnedBytes) -> io::Result<Self::Reader> {
+        let footer_len: u32 = (&bytes[bytes.len() - 4..]).deserialize()?;
+        let footer_offset = bytes.len() - 4 - footer_len as usize;
+        let (data, mut footer) = bytes.split(footer_offset);
+        let footer = BlockwiseLinearFooter::deserialize(&mut footer)?;
+        Ok(BlockwiseLinearReader { data, footer })
+    }
+
     /// Creates a new fast field serializer.
-    fn serialize(
-        write: &mut impl Write,
-        fastfield_accessor: &dyn FastFieldDataAccess,
-    ) -> io::Result<()> {
+    fn serialize(write: &mut impl Write, fastfield_accessor: &dyn Column) -> io::Result<()> {
         assert!(fastfield_accessor.min_value() <= fastfield_accessor.max_value());
 
         let first_val = fastfield_accessor.get_val(0);
@@ -290,29 +286,24 @@ impl FastFieldCodecSerializer for BlockwiseLinearSerializer {
         Ok(())
     }
 
-    fn is_applicable(fastfield_accessor: &impl FastFieldDataAccess) -> bool {
-        if fastfield_accessor.num_vals() < 5_000 {
-            return false;
+    /// estimation for linear interpolation is hard because, you don't know
+    /// where the local maxima are for the deviation of the calculated value and
+    /// the offset is also unknown.
+    fn estimate(fastfield_accessor: &impl Column) -> Option<f32> {
+        if fastfield_accessor.num_vals() < 10 * CHUNK_SIZE {
+            return None;
         }
+
         // On serialization the offset is added to the actual value.
         // We need to make sure this won't run into overflow calculation issues.
         // For this we take the maximum theroretical offset and add this to the max value.
         // If this doesn't overflow the algorithm should be fine
         let theorethical_maximum_offset =
             fastfield_accessor.max_value() - fastfield_accessor.min_value();
-        if fastfield_accessor
+        fastfield_accessor
             .max_value()
-            .checked_add(theorethical_maximum_offset)
-            .is_none()
-        {
-            return false;
-        }
-        true
-    }
-    /// estimation for linear interpolation is hard because, you don't know
-    /// where the local maxima are for the deviation of the calculated value and
-    /// the offset is also unknown.
-    fn estimate(fastfield_accessor: &impl FastFieldDataAccess) -> f32 {
+            .checked_add(theorethical_maximum_offset)?;
+
         let first_val_in_first_block = fastfield_accessor.get_val(0);
         let last_elem_in_first_chunk = CHUNK_SIZE.min(fastfield_accessor.num_vals());
         let last_val_in_first_block =
@@ -351,7 +342,7 @@ impl FastFieldCodecSerializer for BlockwiseLinearSerializer {
             // function metadata per block
             + 29 * (fastfield_accessor.num_vals() / CHUNK_SIZE);
         let num_bits_uncompressed = 64 * fastfield_accessor.num_vals();
-        num_bits as f32 / num_bits_uncompressed as f32
+        Some(num_bits as f32 / num_bits_uncompressed as f32)
     }
 }
 
@@ -366,12 +357,10 @@ fn distance<T: Sub<Output = T> + Ord>(x: T, y: T) -> T {
 #[cfg(test)]
 mod tests {
     use super::*;
-    use crate::tests::get_codec_test_data_sets;
+    use crate::tests::get_codec_test_datasets;
 
-    fn create_and_validate(data: &[u64], name: &str) -> (f32, f32) {
-        crate::tests::create_and_validate::<BlockwiseLinearSerializer, BlockwiseLinearReader>(
-            data, name,
-        )
+    fn create_and_validate(data: &[u64], name: &str) -> Option<(f32, f32)> {
+        crate::tests::create_and_validate::<BlockwiseLinearCodec>(data, name)
     }
 
     const HIGHEST_BIT: u64 = 1 << 63;
@@ -385,7 +374,7 @@ mod tests {
             .map(i64_to_u64)
             .collect::<Vec<_>>();
         let (estimate, actual_compression) =
-            create_and_validate(&data, "simple monotonically large i64");
+            create_and_validate(&data, "simple monotonically large i64").unwrap();
         assert!(actual_compression < 0.2);
         assert!(estimate < 0.20);
         assert!(estimate > 0.15);
@@ -396,7 +385,7 @@ mod tests {
     fn test_compression() {
         let data = (10..=6_000_u64).collect::<Vec<_>>();
         let (estimate, actual_compression) =
-            create_and_validate(&data, "simple monotonically large");
+            create_and_validate(&data, "simple monotonically large").unwrap();
         assert!(actual_compression < 0.2);
         assert!(estimate < 0.20);
         assert!(estimate > 0.15);
@@ -405,7 +394,7 @@ mod tests {
 
     #[test]
     fn test_with_codec_data_sets() {
-        let data_sets = get_codec_test_data_sets();
+        let data_sets = get_codec_test_datasets();
         for (mut data, name) in data_sets {
             create_and_validate(&data, name);
             data.reverse();

fastfield_codecs/src/column.rs

@@ -0,0 +1,48 @@
+use std::ops::Range;
+
+use crate::ColumnIter;
+
+pub trait Column<T = u64> {
+    /// Return the value associated to the given idx.
+    ///
+    /// This accessor should return as fast as possible.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `idx` is greater than the column length.
+    fn get_val(&self, idx: u64) -> T;
+
+    /// Returns an iterator over given doc range.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `range.end()` is greater than
+    /// the segment's `maxdoc`.
+    #[inline]
+    fn get_range(&self, range: Range<u64>) -> ColumnIter<'_, Self, T>
+    where
+        Self: Sized,
+    {
+        ColumnIter::new(self, range)
+    }
+
+    /// Returns the minimum value for this fast field.
+    ///
+    /// The min value does not take in account of possible
+    /// deleted document, and should be considered as a lower bound
+    /// of the actual minimum value.
+    fn min_value(&self) -> T;
+
+    /// Returns the maximum value for this fast field.
+    ///
+    /// The max value does not take in account of possible
+    /// deleted document, and should be considered as an upper bound
+    /// of the actual maximum value
+    fn max_value(&self) -> T;
+
+    fn num_vals(&self) -> u64;
+    /// Returns a iterator over the data
+    fn iter<'a>(&'a self) -> Box<dyn Iterator<Item = T> + 'a> {
+        Box::new((0..self.num_vals()).map(|idx| self.get_val(idx)))
+    }
+}

fastfield_codecs/src/lib.rs

@@ -4,6 +4,9 @@ extern crate more_asserts;
 
 use std::io;
 use std::io::Write;
+use std::iter::FusedIterator;
+use std::marker::PhantomData;
+use std::ops::Range;
 
 use common::BinarySerializable;
 use ownedbytes::OwnedBytes;
@@ -12,22 +15,9 @@ pub mod bitpacked;
 pub mod blockwise_linear;
 pub mod linear;
 
-pub trait FastFieldCodecDeserializer: Sized {
-    /// Reads the metadata and returns the CodecReader
-    fn open_from_bytes(bytes: OwnedBytes) -> std::io::Result<Self>
-    where Self: FastFieldDataAccess;
-}
+mod column;
 
-pub trait FastFieldDataAccess {
-    fn get_val(&self, doc: u64) -> u64;
-    fn min_value(&self) -> u64;
-    fn max_value(&self) -> u64;
-    fn num_vals(&self) -> u64;
-    /// Returns a iterator over the data
-    fn iter<'a>(&'a self) -> Box<dyn Iterator<Item = u64> + 'a> {
-        Box::new((0..self.num_vals()).map(|idx| self.get_val(idx)))
-    }
-}
+pub use self::column::Column;
 
 #[derive(PartialEq, Eq, PartialOrd, Ord, Debug, Clone, Copy)]
 #[repr(u8)]
@@ -69,29 +59,30 @@ impl FastFieldCodecType {
 
 /// The FastFieldSerializerEstimate trait is required on all variants
 /// of fast field compressions, to decide which one to choose.
-pub trait FastFieldCodecSerializer {
+pub trait FastFieldCodec {
     /// A codex needs to provide a unique name and id, which is
     /// used for debugging and de/serialization.
     const CODEC_TYPE: FastFieldCodecType;
 
-    /// Check if the Codec is able to compress the data
-    fn is_applicable(fastfield_accessor: &impl FastFieldDataAccess) -> bool;
+    type Reader: Column<u64>;
 
-    /// Returns an estimate of the compression ratio.
-    /// The baseline is uncompressed 64bit data.
-    ///
-    /// It could make sense to also return a value representing
-    /// computational complexity.
-    fn estimate(fastfield_accessor: &impl FastFieldDataAccess) -> f32;
+    /// Reads the metadata and returns the CodecReader
+    fn open_from_bytes(bytes: OwnedBytes) -> io::Result<Self::Reader>;
 
     /// Serializes the data using the serializer into write.
     ///
     /// The fastfield_accessor iterator should be preferred over using fastfield_accessor for
     /// performance reasons.
-    fn serialize(
-        write: &mut impl Write,
-        fastfield_accessor: &dyn FastFieldDataAccess,
-    ) -> io::Result<()>;
+    fn serialize(write: &mut impl Write, fastfield_accessor: &dyn Column<u64>) -> io::Result<()>;
+
+    /// Returns an estimate of the compression ratio.
+    /// If the codec is not applicable, returns `None`.
+    ///
+    /// The baseline is uncompressed 64bit data.
+    ///
+    /// It could make sense to also return a value representing
+    /// computational complexity.
+    fn estimate(fastfield_accessor: &impl Column) -> Option<f32>;
 }
 
 #[derive(Debug, Clone)]
@@ -102,7 +93,7 @@ pub struct FastFieldStats {
     pub num_vals: u64,
 }
 
-impl<'a> FastFieldDataAccess for &'a [u64] {
+impl<'a> Column for &'a [u64] {
     fn get_val(&self, position: u64) -> u64 {
         self[position as usize]
     }
@@ -124,7 +115,57 @@ impl<'a> FastFieldDataAccess for &'a [u64] {
     }
 }
 
-impl FastFieldDataAccess for Vec<u64> {
+pub struct ColumnIter<'a, C: Column<I>, I> {
+    column: &'a C,
+    range: Range<u64>,
+    _phantom: PhantomData<I>,
+}
+
+impl<'a, C: Column<I>, I> ColumnIter<'a, C, I> {
+    #[inline]
+    pub fn new(col: &'a C, range: Range<u64>) -> Self {
+        Self {
+            column: col,
+            range,
+            _phantom: PhantomData,
+        }
+    }
+}
+
+impl<'a, C: Column<I>, I> Iterator for ColumnIter<'a, C, I> {
+    type Item = I;
+
+    #[inline]
+    fn next(&mut self) -> Option<Self::Item> {
+        Some(self.column.get_val(self.range.next()?))
+    }
+
+    #[inline]
+    fn fold<Acc, G>(self, init: Acc, mut g: G) -> Acc
+    where
+        G: FnMut(Acc, Self::Item) -> Acc,
+    {
+        self.range
+            .fold(init, move |acc, idx| g(acc, self.column.get_val(idx)))
+    }
+
+    #[inline]
+    fn size_hint(&self) -> (usize, Option<usize>) {
+        let size = (self.range.end - self.range.start) as usize;
+        (size, Some(size))
+    }
+}
+
+impl<'a, C: Column<I>, I> ExactSizeIterator for ColumnIter<'a, C, I> {
+    #[inline]
+    fn len(&self) -> usize {
+        let size = (self.range.end - self.range.start) as usize;
+        size as usize
+    }
+}
+impl<'a, C: Column<I>, I> FusedIterator for ColumnIter<'a, C, I> {}
+
+impl Column for Vec<u64> {
     fn get_val(&self, position: u64) -> u64 {
         self[position as usize]
     }
@@ -149,61 +190,55 @@ mod tests {
     use proptest::arbitrary::any;
     use proptest::proptest;
 
-    use crate::bitpacked::{BitpackedReader, BitpackedSerializer};
-    use crate::blockwise_linear::{BlockwiseLinearReader, BlockwiseLinearSerializer};
-    use crate::linear::{LinearReader, LinearSerializer};
+    use crate::bitpacked::BitpackedCodec;
+    use crate::blockwise_linear::BlockwiseLinearCodec;
+    use crate::linear::LinearCodec;
 
-    pub fn create_and_validate<
-        S: FastFieldCodecSerializer,
-        R: FastFieldCodecDeserializer + FastFieldDataAccess,
-    >(
+    pub fn create_and_validate<Codec: FastFieldCodec>(
         data: &[u64],
         name: &str,
-    ) -> (f32, f32) {
-        if !S::is_applicable(&data) {
-            return (f32::MAX, 0.0);
-        }
-        let estimation = S::estimate(&data);
+    ) -> Option<(f32, f32)> {
+        let estimation = Codec::estimate(&data)?;
+
         let mut out: Vec<u8> = Vec::new();
-        S::serialize(&mut out, &data).unwrap();
+        Codec::serialize(&mut out, &data).unwrap();
 
         let actual_compression = out.len() as f32 / (data.len() as f32 * 8.0);
 
-        let reader = R::open_from_bytes(OwnedBytes::new(out)).unwrap();
+        let reader = Codec::open_from_bytes(OwnedBytes::new(out)).unwrap();
         assert_eq!(reader.num_vals(), data.len() as u64);
-        for (doc, orig_val) in data.iter().enumerate() {
+        for (doc, orig_val) in data.iter().copied().enumerate() {
             let val = reader.get_val(doc as u64);
-            if val != *orig_val {
-                panic!(
-                    "val {val:?} does not match orig_val {orig_val:?}, in data set {name}, data \
-                     {data:?}",
-                );
-            }
+            assert_eq!(
+                val, orig_val,
+                "val `{val}` does not match orig_val {orig_val:?}, in data set {name}, data \
+                 `{data:?}`",
+            );
         }
-        (estimation, actual_compression)
+        Some((estimation, actual_compression))
     }
 
     proptest! {
         #[test]
         fn test_proptest_small(data in proptest::collection::vec(any::<u64>(), 1..10)) {
-            create_and_validate::<LinearSerializer, LinearReader>(&data, "proptest linearinterpol");
-            create_and_validate::<BlockwiseLinearSerializer, BlockwiseLinearReader>(&data, "proptest multilinearinterpol");
-            create_and_validate::<BitpackedSerializer, BitpackedReader>(&data, "proptest bitpacked");
+            create_and_validate::<LinearCodec>(&data, "proptest linearinterpol");
+            create_and_validate::<BlockwiseLinearCodec>(&data, "proptest multilinearinterpol");
+            create_and_validate::<BitpackedCodec>(&data, "proptest bitpacked");
         }
 
         #[test]
         fn test_proptest_large(data in proptest::collection::vec(any::<u64>(), 1..6000)) {
-            create_and_validate::<LinearSerializer, LinearReader>(&data, "proptest linearinterpol");
-            create_and_validate::<BlockwiseLinearSerializer, BlockwiseLinearReader>(&data, "proptest multilinearinterpol");
-            create_and_validate::<BitpackedSerializer, BitpackedReader>(&data, "proptest bitpacked");
+            create_and_validate::<LinearCodec>(&data, "proptest linearinterpol");
+            create_and_validate::<BlockwiseLinearCodec>(&data, "proptest multilinearinterpol");
+            create_and_validate::<BitpackedCodec>(&data, "proptest bitpacked");
         }
 
     }
 
-    pub fn get_codec_test_data_sets() -> Vec<(Vec<u64>, &'static str)> {
+    pub fn get_codec_test_datasets() -> Vec<(Vec<u64>, &'static str)> {
         let mut data_and_names = vec![];
 
-        let data = (10..=20_u64).collect::<Vec<_>>();
+        let data = (10..=10_000_u64).collect::<Vec<_>>();
         data_and_names.push((data, "simple monotonically increasing"));
 
         data_and_names.push((
@@ -213,35 +248,38 @@ mod tests {
         data_and_names.push((vec![5, 50, 3, 13, 1, 1000, 35], "rand small"));
         data_and_names.push((vec![10], "single value"));
 
+        data_and_names.push((
+            vec![1572656989877777, 1170935903116329, 720575940379279, 0],
+            "overflow error",
+        ));
+
         data_and_names
     }
 
-    fn test_codec<
-        S: FastFieldCodecSerializer,
-        R: FastFieldDataAccess + FastFieldCodecDeserializer,
-    >() {
-        let codec_name = format!("{:?}", S::CODEC_TYPE);
-        for (data, dataset_name) in get_codec_test_data_sets() {
-            let (estimate, actual) = crate::tests::create_and_validate::<S, R>(&data, dataset_name);
-            let result = if estimate == f32::MAX {
-                "Disabled".to_string()
-            } else {
+    fn test_codec<C: FastFieldCodec>() {
+        let codec_name = format!("{:?}", C::CODEC_TYPE);
+        for (data, dataset_name) in get_codec_test_datasets() {
+            let estimate_actual_opt: Option<(f32, f32)> =
+                crate::tests::create_and_validate::<C>(&data, dataset_name);
+            let result = if let Some((estimate, actual)) = estimate_actual_opt {
                 format!("Estimate `{estimate}` Actual `{actual}`")
+            } else {
+                "Disabled".to_string()
             };
             println!("Codec {codec_name}, DataSet {dataset_name}, {result}");
         }
     }
     #[test]
     fn test_codec_bitpacking() {
-        test_codec::<BitpackedSerializer, BitpackedReader>();
+        test_codec::<BitpackedCodec>();
     }
     #[test]
     fn test_codec_interpolation() {
-        test_codec::<LinearSerializer, LinearReader>();
+        test_codec::<LinearCodec>();
     }
     #[test]
     fn test_codec_multi_interpolation() {
-        test_codec::<BlockwiseLinearSerializer, BlockwiseLinearReader>();
+        test_codec::<BlockwiseLinearCodec>();
     }
 
     use super::*;
@@ -250,37 +288,37 @@ mod tests {
     fn estimation_good_interpolation_case() {
         let data = (10..=20000_u64).collect::<Vec<_>>();
 
-        let linear_interpol_estimation = LinearSerializer::estimate(&data);
+        let linear_interpol_estimation = LinearCodec::estimate(&data).unwrap();
         assert_le!(linear_interpol_estimation, 0.01);
 
-        let multi_linear_interpol_estimation = BlockwiseLinearSerializer::estimate(&data);
+        let multi_linear_interpol_estimation = BlockwiseLinearCodec::estimate(&data).unwrap();
         assert_le!(multi_linear_interpol_estimation, 0.2);
         assert_le!(linear_interpol_estimation, multi_linear_interpol_estimation);
 
-        let bitpacked_estimation = BitpackedSerializer::estimate(&data);
+        let bitpacked_estimation = BitpackedCodec::estimate(&data).unwrap();
         assert_le!(linear_interpol_estimation, bitpacked_estimation);
     }
     #[test]
     fn estimation_test_bad_interpolation_case() {
         let data = vec![200, 10, 10, 10, 10, 1000, 20];
 
-        let linear_interpol_estimation = LinearSerializer::estimate(&data);
+        let linear_interpol_estimation = LinearCodec::estimate(&data).unwrap();
         assert_le!(linear_interpol_estimation, 0.32);
 
-        let bitpacked_estimation = BitpackedSerializer::estimate(&data);
+        let bitpacked_estimation = BitpackedCodec::estimate(&data).unwrap();
         assert_le!(bitpacked_estimation, linear_interpol_estimation);
     }
     #[test]
     fn estimation_test_bad_interpolation_case_monotonically_increasing() {
-        let mut data = (200..=20000_u64).collect::<Vec<_>>();
+        let mut data: Vec<u64> = (200..=20000_u64).collect();
         data.push(1_000_000);
 
         // in this case the linear interpolation can't in fact not be worse than bitpacking,
         // but the estimator adds some threshold, which leads to estimated worse behavior
-        let linear_interpol_estimation = LinearSerializer::estimate(&data);
+        let linear_interpol_estimation = LinearCodec::estimate(&data).unwrap();
         assert_le!(linear_interpol_estimation, 0.35);
 
-        let bitpacked_estimation = BitpackedSerializer::estimate(&data);
+        let bitpacked_estimation = BitpackedCodec::estimate(&data).unwrap();
         assert_le!(bitpacked_estimation, 0.32);
         assert_le!(bitpacked_estimation, linear_interpol_estimation);
     }

fastfield_codecs/src/linear.rs

@@ -5,9 +5,7 @@ use common::{BinarySerializable, FixedSize};
 use ownedbytes::OwnedBytes;
 use tantivy_bitpacker::{compute_num_bits, BitPacker, BitUnpacker};
 
-use crate::{
-    FastFieldCodecDeserializer, FastFieldCodecSerializer, FastFieldCodecType, FastFieldDataAccess,
-};
+use crate::{Column, FastFieldCodec, FastFieldCodecType};
 
 /// Depending on the field type, a different
 /// fast field is required.
@@ -59,25 +57,7 @@ impl FixedSize for LinearFooter {
     const SIZE_IN_BYTES: usize = 56;
 }
 
-impl FastFieldCodecDeserializer for LinearReader {
-    /// Opens a fast field given a file.
-    fn open_from_bytes(bytes: OwnedBytes) -> io::Result<Self> {
-        let footer_offset = bytes.len() - LinearFooter::SIZE_IN_BYTES;
-        let (data, mut footer) = bytes.split(footer_offset);
-        let footer = LinearFooter::deserialize(&mut footer)?;
-        let slope = get_slope(footer.first_val, footer.last_val, footer.num_vals);
-        let num_bits = compute_num_bits(footer.relative_max_value);
-        let bit_unpacker = BitUnpacker::new(num_bits);
-        Ok(LinearReader {
-            data,
-            bit_unpacker,
-            footer,
-            slope,
-        })
-    }
-}
-
-impl FastFieldDataAccess for LinearReader {
+impl Column for LinearReader {
     #[inline]
     fn get_val(&self, doc: u64) -> u64 {
         let calculated_value = get_calculated_value(self.footer.first_val, doc, self.slope);
@@ -100,7 +80,7 @@ impl FastFieldDataAccess for LinearReader {
 
 /// Fastfield serializer, which tries to guess values by linear interpolation
 /// and stores the difference bitpacked.
-pub struct LinearSerializer {}
+pub struct LinearCodec;
 
 #[inline]
 pub(crate) fn get_slope(first_val: u64, last_val: u64, num_vals: u64) -> f32 {
@@ -135,20 +115,35 @@ fn diff(val1: u64, val2: u64) -> f64 {
 #[inline]
 pub fn get_calculated_value(first_val: u64, pos: u64, slope: f32) -> u64 {
     if slope < 0.0 {
-        first_val - (pos as f32 * -slope) as u64
+        first_val.saturating_sub((pos as f32 * -slope) as u64)
     } else {
-        first_val + (pos as f32 * slope) as u64
+        first_val.saturating_add((pos as f32 * slope) as u64)
     }
 }
 
-impl FastFieldCodecSerializer for LinearSerializer {
+impl FastFieldCodec for LinearCodec {
     const CODEC_TYPE: FastFieldCodecType = FastFieldCodecType::Linear;
 
+    type Reader = LinearReader;
+
+    /// Opens a fast field given a file.
+    fn open_from_bytes(bytes: OwnedBytes) -> io::Result<Self::Reader> {
+        let footer_offset = bytes.len() - LinearFooter::SIZE_IN_BYTES;
+        let (data, mut footer) = bytes.split(footer_offset);
+        let footer = LinearFooter::deserialize(&mut footer)?;
+        let slope = get_slope(footer.first_val, footer.last_val, footer.num_vals);
+        let num_bits = compute_num_bits(footer.relative_max_value);
+        let bit_unpacker = BitUnpacker::new(num_bits);
+        Ok(LinearReader {
+            data,
+            bit_unpacker,
+            footer,
+            slope,
+        })
+    }
+
     /// Creates a new fast field serializer.
-    fn serialize(
-        write: &mut impl Write,
-        fastfield_accessor: &dyn FastFieldDataAccess,
-    ) -> io::Result<()> {
+    fn serialize(write: &mut impl Write, fastfield_accessor: &dyn Column) -> io::Result<()> {
         assert!(fastfield_accessor.min_value() <= fastfield_accessor.max_value());
 
         let first_val = fastfield_accessor.get_val(0);
@@ -194,29 +189,25 @@ impl FastFieldCodecSerializer for LinearSerializer {
         footer.serialize(write)?;
         Ok(())
     }
-    fn is_applicable(fastfield_accessor: &impl FastFieldDataAccess) -> bool {
+
+    /// estimation for linear interpolation is hard because, you don't know
+    /// where the local maxima for the deviation of the calculated value are and
+    /// the offset to shift all values to >=0 is also unknown.
+    fn estimate(fastfield_accessor: &impl Column) -> Option<f32> {
         if fastfield_accessor.num_vals() < 3 {
-            return false; // disable compressor for this case
+            return None; // disable compressor for this case
         }
+
         // On serialisation the offset is added to the actual value.
         // We need to make sure this won't run into overflow calculation issues.
         // For this we take the maximum theroretical offset and add this to the max value.
         // If this doesn't overflow the algorithm should be fine
         let theorethical_maximum_offset =
             fastfield_accessor.max_value() - fastfield_accessor.min_value();
-        if fastfield_accessor
+        fastfield_accessor
             .max_value()
-            .checked_add(theorethical_maximum_offset)
-            .is_none()
-        {
-            return false;
-        }
-        true
-    }
-    /// estimation for linear interpolation is hard because, you don't know
-    /// where the local maxima for the deviation of the calculated value are and
-    /// the offset to shift all values to >=0 is also unknown.
-    fn estimate(fastfield_accessor: &impl FastFieldDataAccess) -> f32 {
+            .checked_add(theorethical_maximum_offset)?;
+
         let first_val = fastfield_accessor.get_val(0);
         let last_val = fastfield_accessor.get_val(fastfield_accessor.num_vals() as u64 - 1);
         let slope = get_slope(first_val, last_val, fastfield_accessor.num_vals());
@@ -248,7 +239,7 @@ impl FastFieldCodecSerializer for LinearSerializer {
             * fastfield_accessor.num_vals()
             + LinearFooter::SIZE_IN_BYTES as u64;
         let num_bits_uncompressed = 64 * fastfield_accessor.num_vals();
-        num_bits as f32 / num_bits_uncompressed as f32
+        Some(num_bits as f32 / num_bits_uncompressed as f32)
     }
 }
 
@@ -264,10 +255,10 @@ fn distance<T: Sub<Output = T> + Ord>(x: T, y: T) -> T {
 #[cfg(test)]
 mod tests {
     use super::*;
-    use crate::tests::get_codec_test_data_sets;
+    use crate::tests::get_codec_test_datasets;
 
-    fn create_and_validate(data: &[u64], name: &str) -> (f32, f32) {
-        crate::tests::create_and_validate::<LinearSerializer, LinearReader>(data, name)
+    fn create_and_validate(data: &[u64], name: &str) -> Option<(f32, f32)> {
+        crate::tests::create_and_validate::<LinearCodec>(data, name)
     }
 
     #[test]
@@ -294,15 +285,15 @@ mod tests {
     fn test_compression() {
         let data = (10..=6_000_u64).collect::<Vec<_>>();
         let (estimate, actual_compression) =
-            create_and_validate(&data, "simple monotonically large");
+            create_and_validate(&data, "simple monotonically large").unwrap();
 
         assert!(actual_compression < 0.01);
         assert!(estimate < 0.01);
     }
 
     #[test]
-    fn test_with_codec_data_sets() {
-        let data_sets = get_codec_test_data_sets();
+    fn test_with_codec_datasets() {
+        let data_sets = get_codec_test_datasets();
         for (mut data, name) in data_sets {
             create_and_validate(&data, name);
             data.reverse();
@@ -319,6 +310,13 @@ mod tests {
 
         create_and_validate(&data, "large amplitude");
     }
+
+    #[test]
+    fn overflow_error_test() {
+        let data = vec![1572656989877777, 1170935903116329, 720575940379279, 0];
+        create_and_validate(&data, "overflow test");
+    }
+
     #[test]
     fn linear_interpol_fast_concave_data() {
         let data = vec![0, 1, 2, 5, 8, 10, 20, 50];
@@ -339,9 +337,10 @@ mod tests {
     #[test]
     fn linear_interpol_fast_field_rand() {
         for _ in 0..5000 {
-            let mut data = (0..50).map(|_| rand::random::<u64>()).collect::<Vec<_>>();
+            let mut data = (0..10_000)
+                .map(|_| rand::random::<u64>())
+                .collect::<Vec<_>>();
             create_and_validate(&data, "random");
-
             data.reverse();
             create_and_validate(&data, "random");
         }

fastfield_codecs/src/main.rs

@@ -1,8 +1,9 @@
 #[macro_use]
 extern crate prettytable;
-use fastfield_codecs::blockwise_linear::BlockwiseLinearSerializer;
-use fastfield_codecs::linear::LinearSerializer;
-use fastfield_codecs::{FastFieldCodecSerializer, FastFieldCodecType, FastFieldStats};
+use fastfield_codecs::bitpacked::BitpackedCodec;
+use fastfield_codecs::blockwise_linear::BlockwiseLinearCodec;
+use fastfield_codecs::linear::LinearCodec;
+use fastfield_codecs::{FastFieldCodec, FastFieldCodecType, FastFieldStats};
 use prettytable::{Cell, Row, Table};
 
 fn main() {
@@ -12,37 +13,30 @@ fn main() {
     table.add_row(row!["", "Compression Ratio", "Compression Estimation"]);
 
     for (data, data_set_name) in get_codec_test_data_sets() {
-        let mut results = vec![];
-        let res = serialize_with_codec::<LinearSerializer>(&data);
-        results.push(res);
-        let res = serialize_with_codec::<BlockwiseLinearSerializer>(&data);
-        results.push(res);
-        let res = serialize_with_codec::<fastfield_codecs::bitpacked::BitpackedSerializer>(&data);
-        results.push(res);
-
-        // let best_estimation_codec = results
-        //.iter()
-        //.min_by(|res1, res2| res1.partial_cmp(&res2).unwrap())
-        //.unwrap();
+        let results: Vec<(f32, f32, FastFieldCodecType)> = [
+            serialize_with_codec::<LinearCodec>(&data),
+            serialize_with_codec::<BlockwiseLinearCodec>(&data),
+            serialize_with_codec::<BlockwiseLinearCodec>(&data),
+            serialize_with_codec::<BitpackedCodec>(&data),
+        ]
+        .into_iter()
+        .flatten()
+        .collect();
         let best_compression_ratio_codec = results
             .iter()
-            .min_by(|res1, res2| res1.partial_cmp(res2).unwrap())
+            .min_by(|&res1, &res2| res1.partial_cmp(res2).unwrap())
             .cloned()
             .unwrap();
 
         table.add_row(Row::new(vec![Cell::new(data_set_name).style_spec("Bbb")]));
-        for (is_applicable, est, comp, codec_type) in results {
-            let (est_cell, ratio_cell) = if !is_applicable {
-                ("Codec Disabled".to_string(), "".to_string())
-            } else {
-                (est.to_string(), comp.to_string())
-            };
+        for (est, comp, codec_type) in results {
+            let est_cell = est.to_string();
+            let ratio_cell = comp.to_string();
             let style = if comp == best_compression_ratio_codec.1 {
                 "Fb"
             } else {
                 ""
             };
-
             table.add_row(Row::new(vec![
                 Cell::new(&format!("{codec_type:?}")).style_spec("bFg"),
                 Cell::new(&ratio_cell).style_spec(style),
@@ -89,19 +83,14 @@ pub fn get_codec_test_data_sets() -> Vec<(Vec<u64>, &'static str)> {
     data_and_names
 }
 
-pub fn serialize_with_codec<S: FastFieldCodecSerializer>(
+pub fn serialize_with_codec<C: FastFieldCodec>(
     data: &[u64],
-) -> (bool, f32, f32, FastFieldCodecType) {
-    let is_applicable = S::is_applicable(&data);
-    if !is_applicable {
-        return (false, 0.0, 0.0, S::CODEC_TYPE);
-    }
-    let estimation = S::estimate(&data);
-    let mut out = vec![];
-    S::serialize(&mut out, &data).unwrap();
-
+) -> Option<(f32, f32, FastFieldCodecType)> {
+    let estimation = C::estimate(&data)?;
+    let mut out = Vec::new();
+    C::serialize(&mut out, &data).unwrap();
     let actual_compression = out.len() as f32 / (data.len() * 8) as f32;
-    (true, estimation, actual_compression, S::CODEC_TYPE)
+    Some((estimation, actual_compression, C::CODEC_TYPE))
 }
 
 pub fn stats_from_vec(data: &[u64]) -> FastFieldStats {

src/aggregation/bucket/histogram/histogram.rs

@@ -1,6 +1,7 @@
 use std::cmp::Ordering;
 use std::fmt::Display;
 
+use fastfield_codecs::Column;
 use itertools::Itertools;
 use serde::{Deserialize, Serialize};
 
@@ -14,7 +15,7 @@ use crate::aggregation::intermediate_agg_result::{
     IntermediateAggregationResults, IntermediateBucketResult, IntermediateHistogramBucketEntry,
 };
 use crate::aggregation::segment_agg_result::SegmentAggregationResultsCollector;
-use crate::fastfield::{DynamicFastFieldReader, FastFieldReader};
+use crate::fastfield::DynamicFastFieldReader;
 use crate::schema::Type;
 use crate::{DocId, TantivyError};
 
@@ -331,10 +332,10 @@ impl SegmentHistogramCollector {
             .expect("unexpected fast field cardinatility");
         let mut iter = doc.chunks_exact(4);
         for docs in iter.by_ref() {
-            let val0 = self.f64_from_fastfield_u64(accessor.get(docs[0]));
-            let val1 = self.f64_from_fastfield_u64(accessor.get(docs[1]));
-            let val2 = self.f64_from_fastfield_u64(accessor.get(docs[2]));
-            let val3 = self.f64_from_fastfield_u64(accessor.get(docs[3]));
+            let val0 = self.f64_from_fastfield_u64(accessor.get_val(docs[0] as u64));
+            let val1 = self.f64_from_fastfield_u64(accessor.get_val(docs[1] as u64));
+            let val2 = self.f64_from_fastfield_u64(accessor.get_val(docs[2] as u64));
+            let val3 = self.f64_from_fastfield_u64(accessor.get_val(docs[3] as u64));
 
             let bucket_pos0 = get_bucket_num(val0);
             let bucket_pos1 = get_bucket_num(val1);
@@ -370,8 +371,8 @@ impl SegmentHistogramCollector {
                 &bucket_with_accessor.sub_aggregation,
             )?;
         }
-        for doc in iter.remainder() {
-            let val = f64_from_fastfield_u64(accessor.get(*doc), &self.field_type);
+        for &doc in iter.remainder() {
+            let val = f64_from_fastfield_u64(accessor.get_val(doc as u64), &self.field_type);
             if !bounds.contains(val) {
                 continue;
             }
@@ -382,7 +383,7 @@ impl SegmentHistogramCollector {
                 self.buckets[bucket_pos].key,
                 get_bucket_val(val, self.interval, self.offset) as f64
             );
-            self.increment_bucket(bucket_pos, *doc, &bucket_with_accessor.sub_aggregation)?;
+            self.increment_bucket(bucket_pos, doc, &bucket_with_accessor.sub_aggregation)?;
         }
         if force_flush {
             if let Some(sub_aggregations) = self.sub_aggregations.as_mut() {

src/aggregation/bucket/range.rs

@@ -1,6 +1,7 @@
 use std::fmt::Debug;
 use std::ops::Range;
 
+use fastfield_codecs::Column;
 use fnv::FnvHashMap;
 use serde::{Deserialize, Serialize};
 
@@ -12,7 +13,6 @@ use crate::aggregation::intermediate_agg_result::{
 };
 use crate::aggregation::segment_agg_result::{BucketCount, SegmentAggregationResultsCollector};
 use crate::aggregation::{f64_from_fastfield_u64, f64_to_fastfield_u64, Key, SerializedKey};
-use crate::fastfield::FastFieldReader;
 use crate::schema::Type;
 use crate::{DocId, TantivyError};
 
@@ -264,10 +264,10 @@ impl SegmentRangeCollector {
             .as_single()
             .expect("unexpected fast field cardinatility");
         for docs in iter.by_ref() {
-            let val1 = accessor.get(docs[0]);
-            let val2 = accessor.get(docs[1]);
-            let val3 = accessor.get(docs[2]);
-            let val4 = accessor.get(docs[3]);
+            let val1 = accessor.get_val(docs[0] as u64);
+            let val2 = accessor.get_val(docs[1] as u64);
+            let val3 = accessor.get_val(docs[2] as u64);
+            let val4 = accessor.get_val(docs[3] as u64);
             let bucket_pos1 = self.get_bucket_pos(val1);
             let bucket_pos2 = self.get_bucket_pos(val2);
             let bucket_pos3 = self.get_bucket_pos(val3);
@@ -278,10 +278,10 @@ impl SegmentRangeCollector {
             self.increment_bucket(bucket_pos3, docs[2], &bucket_with_accessor.sub_aggregation)?;
             self.increment_bucket(bucket_pos4, docs[3], &bucket_with_accessor.sub_aggregation)?;
         }
-        for doc in iter.remainder() {
-            let val = accessor.get(*doc);
+        for &doc in iter.remainder() {
+            let val = accessor.get_val(doc as u64);
             let bucket_pos = self.get_bucket_pos(val);
-            self.increment_bucket(bucket_pos, *doc, &bucket_with_accessor.sub_aggregation)?;
+            self.increment_bucket(bucket_pos, doc, &bucket_with_accessor.sub_aggregation)?;
         }
         if force_flush {
             for bucket in &mut self.buckets {

src/aggregation/bucket/term_agg.rs

@@ -242,13 +242,13 @@ impl TermBuckets {
 
     fn increment_bucket(
         &mut self,
-        term_ids: &[u64],
+        term_ids: impl Iterator<Item = u64>,
         doc: DocId,
         sub_aggregation: &AggregationsWithAccessor,
         bucket_count: &BucketCount,
         blueprint: &Option<SegmentAggregationResultsCollector>,
     ) -> crate::Result<()> {
-        for &term_id in term_ids {
+        for term_id in term_ids {
             let entry = self.entries.entry(term_id as u32).or_insert_with(|| {
                 bucket_count.add_count(1);
 
@@ -432,39 +432,30 @@ impl SegmentTermCollector {
             .as_multi()
             .expect("unexpected fast field cardinatility");
         let mut iter = doc.chunks_exact(4);
-        let mut vals1 = vec![];
-        let mut vals2 = vec![];
-        let mut vals3 = vec![];
-        let mut vals4 = vec![];
         for docs in iter.by_ref() {
-            accessor.get_vals(docs[0], &mut vals1);
-            accessor.get_vals(docs[1], &mut vals2);
-            accessor.get_vals(docs[2], &mut vals3);
-            accessor.get_vals(docs[3], &mut vals4);
-
             self.term_buckets.increment_bucket(
-                &vals1,
+                accessor.get_vals(docs[0]),
                 docs[0],
                 &bucket_with_accessor.sub_aggregation,
                 &bucket_with_accessor.bucket_count,
                 &self.blueprint,
             )?;
             self.term_buckets.increment_bucket(
-                &vals2,
+                accessor.get_vals(docs[1]),
                 docs[1],
                 &bucket_with_accessor.sub_aggregation,
                 &bucket_with_accessor.bucket_count,
                 &self.blueprint,
             )?;
             self.term_buckets.increment_bucket(
-                &vals3,
+                accessor.get_vals(docs[2]),
                 docs[2],
                 &bucket_with_accessor.sub_aggregation,
                 &bucket_with_accessor.bucket_count,
                 &self.blueprint,
             )?;
             self.term_buckets.increment_bucket(
-                &vals4,
+                accessor.get_vals(docs[3]),
                 docs[3],
                 &bucket_with_accessor.sub_aggregation,
                 &bucket_with_accessor.bucket_count,
@@ -472,10 +463,8 @@ impl SegmentTermCollector {
             )?;
         }
         for &doc in iter.remainder() {
-            accessor.get_vals(doc, &mut vals1);
-
             self.term_buckets.increment_bucket(
-                &vals1,
+                accessor.get_vals(doc),
                 doc,
                 &bucket_with_accessor.sub_aggregation,
                 &bucket_with_accessor.bucket_count,
@@ -1334,11 +1323,15 @@ mod bench {
             max_bucket_count: 1_000_001u32,
         };
         b.iter(|| {
-            for &val in &vals {
-                collector
-                    .increment_bucket(&[val], 0, &aggregations_with_accessor, &bucket_count, &None)
-                    .unwrap();
-            }
+            collector
+                .increment_bucket(
+                    vals.iter().cloned(),
+                    0,
+                    &aggregations_with_accessor,
+                    &bucket_count,
+                    &None,
+                )
+                .unwrap();
         })
     }
 

src/aggregation/metric/average.rs

@@ -1,9 +1,10 @@
 use std::fmt::Debug;
 
+use fastfield_codecs::Column;
 use serde::{Deserialize, Serialize};
 
 use crate::aggregation::f64_from_fastfield_u64;
-use crate::fastfield::{DynamicFastFieldReader, FastFieldReader};
+use crate::fastfield::DynamicFastFieldReader;
 use crate::schema::Type;
 use crate::DocId;
 
@@ -60,10 +61,10 @@ impl SegmentAverageCollector {
     pub(crate) fn collect_block(&mut self, doc: &[DocId], field: &DynamicFastFieldReader<u64>) {
         let mut iter = doc.chunks_exact(4);
         for docs in iter.by_ref() {
-            let val1 = field.get(docs[0]);
-            let val2 = field.get(docs[1]);
-            let val3 = field.get(docs[2]);
-            let val4 = field.get(docs[3]);
+            let val1 = field.get_val(docs[0] as u64);
+            let val2 = field.get_val(docs[1] as u64);
+            let val3 = field.get_val(docs[2] as u64);
+            let val4 = field.get_val(docs[3] as u64);
             let val1 = f64_from_fastfield_u64(val1, &self.field_type);
             let val2 = f64_from_fastfield_u64(val2, &self.field_type);
             let val3 = f64_from_fastfield_u64(val3, &self.field_type);
@@ -73,8 +74,8 @@ impl SegmentAverageCollector {
             self.data.collect(val3);
             self.data.collect(val4);
         }
-        for doc in iter.remainder() {
-            let val = field.get(*doc);
+        for &doc in iter.remainder() {
+            let val = field.get_val(doc as u64);
             let val = f64_from_fastfield_u64(val, &self.field_type);
             self.data.collect(val);
         }

src/aggregation/metric/stats.rs

@@ -1,7 +1,8 @@
+use fastfield_codecs::Column;
 use serde::{Deserialize, Serialize};
 
 use crate::aggregation::f64_from_fastfield_u64;
-use crate::fastfield::{DynamicFastFieldReader, FastFieldReader};
+use crate::fastfield::DynamicFastFieldReader;
 use crate::schema::Type;
 use crate::{DocId, TantivyError};
 
@@ -166,10 +167,10 @@ impl SegmentStatsCollector {
     pub(crate) fn collect_block(&mut self, doc: &[DocId], field: &DynamicFastFieldReader<u64>) {
         let mut iter = doc.chunks_exact(4);
         for docs in iter.by_ref() {
-            let val1 = field.get(docs[0]);
-            let val2 = field.get(docs[1]);
-            let val3 = field.get(docs[2]);
-            let val4 = field.get(docs[3]);
+            let val1 = field.get_val(docs[0] as u64);
+            let val2 = field.get_val(docs[1] as u64);
+            let val3 = field.get_val(docs[2] as u64);
+            let val4 = field.get_val(docs[3] as u64);
             let val1 = f64_from_fastfield_u64(val1, &self.field_type);
             let val2 = f64_from_fastfield_u64(val2, &self.field_type);
             let val3 = f64_from_fastfield_u64(val3, &self.field_type);
@@ -179,8 +180,8 @@ impl SegmentStatsCollector {
             self.stats.collect(val3);
             self.stats.collect(val4);
         }
-        for doc in iter.remainder() {
-            let val = field.get(*doc);
+        for &doc in iter.remainder() {
+            let val = field.get_val(doc as u64);
             let val = f64_from_fastfield_u64(val, &self.field_type);
             self.stats.collect(val);
         }

src/collector/filter_collector_wrapper.rs

@@ -11,8 +11,10 @@
 // Importing tantivy...
 use std::marker::PhantomData;
 
+use fastfield_codecs::Column;
+
 use crate::collector::{Collector, SegmentCollector};
-use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue};
+use crate::fastfield::{DynamicFastFieldReader, FastValue};
 use crate::schema::Field;
 use crate::{Score, SegmentReader, TantivyError};
 
@@ -174,7 +176,7 @@ where
     type Fruit = TSegmentCollector::Fruit;
 
     fn collect(&mut self, doc: u32, score: Score) {
-        let value = self.fast_field_reader.get(doc);
+        let value = self.fast_field_reader.get_val(doc as u64);
         if (self.predicate)(value) {
             self.segment_collector.collect(doc, score)
         }

src/collector/histogram_collector.rs

@@ -1,7 +1,8 @@
 use fastdivide::DividerU64;
+use fastfield_codecs::Column;
 
 use crate::collector::{Collector, SegmentCollector};
-use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue};
+use crate::fastfield::{DynamicFastFieldReader, FastValue};
 use crate::schema::{Field, Type};
 use crate::{DocId, Score};
 
@@ -91,7 +92,7 @@ impl SegmentCollector for SegmentHistogramCollector {
     type Fruit = Vec<u64>;
 
     fn collect(&mut self, doc: DocId, _score: Score) {
-        let value = self.ff_reader.get(doc);
+        let value = self.ff_reader.get_val(doc as u64);
         self.histogram_computer.add_value(value);
     }
 

src/collector/tests.rs

@@ -1,7 +1,9 @@
+use fastfield_codecs::Column;
+
 use super::*;
 use crate::collector::{Count, FilterCollector, TopDocs};
 use crate::core::SegmentReader;
-use crate::fastfield::{BytesFastFieldReader, DynamicFastFieldReader, FastFieldReader};
+use crate::fastfield::{BytesFastFieldReader, DynamicFastFieldReader};
 use crate::query::{AllQuery, QueryParser};
 use crate::schema::{Field, Schema, FAST, TEXT};
 use crate::time::format_description::well_known::Rfc3339;
@@ -197,7 +199,7 @@ impl SegmentCollector for FastFieldSegmentCollector {
     type Fruit = Vec<u64>;
 
     fn collect(&mut self, doc: DocId, _score: Score) {
-        let val = self.reader.get(doc);
+        let val = self.reader.get_val(doc as u64);
         self.vals.push(val);
     }
 

src/collector/top_score_collector.rs

@@ -2,6 +2,8 @@ use std::collections::BinaryHeap;
 use std::fmt;
 use std::marker::PhantomData;
 
+use fastfield_codecs::Column;
+
 use super::Collector;
 use crate::collector::custom_score_top_collector::CustomScoreTopCollector;
 use crate::collector::top_collector::{ComparableDoc, TopCollector, TopSegmentCollector};
@@ -9,7 +11,7 @@ use crate::collector::tweak_score_top_collector::TweakedScoreTopCollector;
 use crate::collector::{
     CustomScorer, CustomSegmentScorer, ScoreSegmentTweaker, ScoreTweaker, SegmentCollector,
 };
-use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue};
+use crate::fastfield::{DynamicFastFieldReader, FastValue};
 use crate::query::Weight;
 use crate::schema::Field;
 use crate::{DocAddress, DocId, Score, SegmentOrdinal, SegmentReader, TantivyError};
@@ -134,7 +136,7 @@ struct ScorerByFastFieldReader {
 
 impl CustomSegmentScorer<u64> for ScorerByFastFieldReader {
     fn score(&mut self, doc: DocId) -> u64 {
-        self.ff_reader.get(doc)
+        self.ff_reader.get_val(doc as u64)
     }
 }
 
@@ -407,7 +409,7 @@ impl TopDocs {
     /// # use tantivy::query::QueryParser;
     /// use tantivy::SegmentReader;
     /// use tantivy::collector::TopDocs;
-    /// use tantivy::fastfield::FastFieldReader;
+    /// use tantivy::fastfield::Column;
     /// use tantivy::schema::Field;
     ///
     /// fn create_schema() -> Schema {
@@ -456,7 +458,7 @@ impl TopDocs {
     ///
     ///             // We can now define our actual scoring function
     ///             move |doc: DocId, original_score: Score| {
-    ///                 let popularity: u64 = popularity_reader.get(doc);
+    ///                 let popularity: u64 = popularity_reader.get_val(doc as u64);
     ///                 // Well.. For the sake of the example we use a simple logarithm
     ///                 // function.
     ///                 let popularity_boost_score = ((2u64 + popularity) as Score).log2();
@@ -515,7 +517,7 @@ impl TopDocs {
     /// use tantivy::SegmentReader;
     /// use tantivy::collector::TopDocs;
     /// use tantivy::schema::Field;
-    /// use tantivy::fastfield::FastFieldReader;
+    /// use fastfield_codecs::Column;
     ///
     /// # fn create_schema() -> Schema {
     /// #    let mut schema_builder = Schema::builder();
@@ -567,8 +569,8 @@ impl TopDocs {
     ///
     ///             // We can now define our actual scoring function
     ///             move |doc: DocId| {
-    ///                 let popularity: u64 = popularity_reader.get(doc);
-    ///                 let boosted: u64 = boosted_reader.get(doc);
+    ///                 let popularity: u64 = popularity_reader.get_val(doc as u64);
+    ///                 let boosted: u64 = boosted_reader.get_val(doc as u64);
     ///                 // Score do not have to be `f64` in tantivy.
     ///                 // Here we return a couple to get lexicographical order
     ///                 // for free.

src/fastfield/bytes/reader.rs

@@ -1,5 +1,7 @@
+use fastfield_codecs::Column;
+
 use crate::directory::{FileSlice, OwnedBytes};
-use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, MultiValueLength};
+use crate::fastfield::{DynamicFastFieldReader, MultiValueLength};
 use crate::DocId;
 
 /// Reader for byte array fast fields
@@ -28,8 +30,9 @@ impl BytesFastFieldReader {
     }
 
     fn range(&self, doc: DocId) -> (usize, usize) {
-        let start = self.idx_reader.get(doc) as usize;
-        let stop = self.idx_reader.get(doc + 1) as usize;
+        let idx = doc as u64;
+        let start = self.idx_reader.get_val(idx) as usize;
+        let stop = self.idx_reader.get_val(idx + 1) as usize;
         (start, stop)
     }
 

src/fastfield/facet_reader.rs

@@ -76,7 +76,8 @@ impl FacetReader {
 
     /// Return the list of facet ordinals associated to a document.
     pub fn facet_ords(&self, doc: DocId, output: &mut Vec<u64>) {
-        self.term_ords.get_vals(doc, output);
+        output.clear();
+        output.extend(self.term_ords.get_vals(doc))
     }
 }
 

src/fastfield/gcd.rs

@@ -3,7 +3,7 @@ use std::num::NonZeroU64;
 
 use common::BinarySerializable;
 use fastdivide::DividerU64;
-use fastfield_codecs::{FastFieldCodecDeserializer, FastFieldDataAccess};
+use fastfield_codecs::{Column, FastFieldCodec};
 use ownedbytes::OwnedBytes;
 
 pub const GCD_DEFAULT: u64 = 1;
@@ -12,50 +12,70 @@ pub const GCD_DEFAULT: u64 = 1;
 ///
 /// Holds the data and the codec to the read the data.
 #[derive(Clone)]
-pub struct GCDFastFieldCodec<CodecReader> {
-    gcd: u64,
-    min_value: u64,
-    num_vals: u64,
+pub struct GCDReader<CodecReader: Column> {
+    gcd_params: GCDParams,
     reader: CodecReader,
 }
 
-impl<C: FastFieldDataAccess + FastFieldCodecDeserializer + Clone> FastFieldCodecDeserializer
-    for GCDFastFieldCodec<C>
-{
-    fn open_from_bytes(bytes: OwnedBytes) -> std::io::Result<Self> {
-        let footer_offset = bytes.len() - 24;
-        let (body, mut footer) = bytes.split(footer_offset);
-        let gcd = u64::deserialize(&mut footer)?;
-        let min_value = u64::deserialize(&mut footer)?;
-        let num_vals = u64::deserialize(&mut footer)?;
-        let reader = C::open_from_bytes(body)?;
-        Ok(GCDFastFieldCodec {
+#[derive(Debug, Clone, Copy)]
+struct GCDParams {
+    gcd: u64,
+    min_value: u64,
+    num_vals: u64,
+}
+
+impl GCDParams {
+    pub fn eval(&self, val: u64) -> u64 {
+        self.min_value + self.gcd * val
+    }
+}
+
+impl BinarySerializable for GCDParams {
+    fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
+        self.gcd.serialize(writer)?;
+        self.min_value.serialize(writer)?;
+        self.num_vals.serialize(writer)?;
+        Ok(())
+    }
+
+    fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
+        let gcd: u64 = u64::deserialize(reader)?;
+        let min_value: u64 = u64::deserialize(reader)?;
+        let num_vals: u64 = u64::deserialize(reader)?;
+        Ok(Self {
             gcd,
             min_value,
             num_vals,
-            reader,
         })
     }
 }
 
-impl<C: FastFieldDataAccess + Clone> FastFieldDataAccess for GCDFastFieldCodec<C> {
+pub fn open_gcd_from_bytes<WrappedCodec: FastFieldCodec>(
+    bytes: OwnedBytes,
+) -> io::Result<GCDReader<WrappedCodec::Reader>> {
+    let footer_offset = bytes.len() - 24;
+    let (body, mut footer) = bytes.split(footer_offset);
+    let gcd_params = GCDParams::deserialize(&mut footer)?;
+    let reader: WrappedCodec::Reader = WrappedCodec::open_from_bytes(body)?;
+    Ok(GCDReader { gcd_params, reader })
+}
+
+impl<C: Column + Clone> Column for GCDReader<C> {
     #[inline]
     fn get_val(&self, doc: u64) -> u64 {
-        let mut data = self.reader.get_val(doc);
-        data *= self.gcd;
-        data += self.min_value;
-        data
+        let val = self.reader.get_val(doc);
+        self.gcd_params.eval(val)
     }
 
     fn min_value(&self) -> u64 {
-        self.min_value + self.reader.min_value() * self.gcd
+        self.gcd_params.eval(self.reader.min_value())
     }
 
     fn max_value(&self) -> u64 {
-        self.min_value + self.reader.max_value() * self.gcd
+        self.gcd_params.eval(self.reader.max_value())
     }
     fn num_vals(&self) -> u64 {
-        self.num_vals
+        self.gcd_params.num_vals
     }
 }
 
@@ -117,6 +137,7 @@ mod tests {
     use std::time::{Duration, SystemTime};
 
     use common::HasLen;
+    use fastfield_codecs::Column;
 
     use crate::directory::{CompositeFile, RamDirectory, WritePtr};
     use crate::fastfield::gcd::compute_gcd;
@@ -124,7 +145,7 @@ mod tests {
     use crate::fastfield::tests::{FIELD, FIELDI64, SCHEMA, SCHEMAI64};
     use crate::fastfield::{
         find_gcd, CompositeFastFieldSerializer, DynamicFastFieldReader, FastFieldCodecType,
-        FastFieldReader, FastFieldsWriter, ALL_CODECS,
+        FastFieldsWriter, ALL_CODECS,
     };
     use crate::schema::{Cardinality, Schema};
     use crate::{DateOptions, DatePrecision, DateTime, Directory};
@@ -168,9 +189,9 @@ mod tests {
         let file = composite_file.open_read(*FIELD).unwrap();
         let fast_field_reader = DynamicFastFieldReader::<i64>::open(file)?;
 
-        assert_eq!(fast_field_reader.get(0), -4000i64);
-        assert_eq!(fast_field_reader.get(1), -3000i64);
-        assert_eq!(fast_field_reader.get(2), -2000i64);
+        assert_eq!(fast_field_reader.get_val(0), -4000i64);
+        assert_eq!(fast_field_reader.get_val(1), -3000i64);
+        assert_eq!(fast_field_reader.get_val(2), -2000i64);
         assert_eq!(fast_field_reader.max_value(), (num_vals as i64 - 5) * 1000);
         assert_eq!(fast_field_reader.min_value(), -4000i64);
         let file = directory.open_read(path).unwrap();
@@ -189,7 +210,7 @@ mod tests {
     #[test]
     fn test_fastfield_gcd_i64() -> crate::Result<()> {
         for &code_type in ALL_CODECS {
-            test_fastfield_gcd_i64_with_codec(code_type, 5005)?;
+            test_fastfield_gcd_i64_with_codec(code_type, 5500)?;
         }
         Ok(())
     }
@@ -209,9 +230,9 @@ mod tests {
         let composite_file = CompositeFile::open(&file)?;
         let file = composite_file.open_read(*FIELD).unwrap();
         let fast_field_reader = DynamicFastFieldReader::<u64>::open(file)?;
-        assert_eq!(fast_field_reader.get(0), 1000u64);
-        assert_eq!(fast_field_reader.get(1), 2000u64);
-        assert_eq!(fast_field_reader.get(2), 3000u64);
+        assert_eq!(fast_field_reader.get_val(0), 1000u64);
+        assert_eq!(fast_field_reader.get_val(1), 2000u64);
+        assert_eq!(fast_field_reader.get_val(2), 3000u64);
         assert_eq!(fast_field_reader.max_value(), num_vals as u64 * 1000);
         assert_eq!(fast_field_reader.min_value(), 1000u64);
         let file = directory.open_read(path).unwrap();
@@ -230,7 +251,7 @@ mod tests {
     #[test]
     fn test_fastfield_gcd_u64() -> crate::Result<()> {
         for &code_type in ALL_CODECS {
-            test_fastfield_gcd_u64_with_codec(code_type, 5005)?;
+            test_fastfield_gcd_u64_with_codec(code_type, 5500)?;
         }
         Ok(())
     }
@@ -238,9 +259,9 @@ mod tests {
     #[test]
     pub fn test_fastfield2() {
         let test_fastfield = DynamicFastFieldReader::<u64>::from(vec![100, 200, 300]);
-        assert_eq!(test_fastfield.get(0), 100);
-        assert_eq!(test_fastfield.get(1), 200);
-        assert_eq!(test_fastfield.get(2), 300);
+        assert_eq!(test_fastfield.get_val(0), 100);
+        assert_eq!(test_fastfield.get_val(1), 200);
+        assert_eq!(test_fastfield.get_val(2), 300);
     }
 
     #[test]
@@ -305,9 +326,9 @@ mod tests {
         let len = file.len();
         let test_fastfield = DynamicFastFieldReader::<DateTime>::open(file)?;
 
-        assert_eq!(test_fastfield.get(0), time1.truncate(precision));
-        assert_eq!(test_fastfield.get(1), time2.truncate(precision));
-        assert_eq!(test_fastfield.get(2), time3.truncate(precision));
+        assert_eq!(test_fastfield.get_val(0), time1.truncate(precision));
+        assert_eq!(test_fastfield.get_val(1), time2.truncate(precision));
+        assert_eq!(test_fastfield.get_val(2), time3.truncate(precision));
         Ok(len)
     }
 

src/fastfield/mod.rs

@@ -26,12 +26,12 @@ pub use self::alive_bitset::{intersect_alive_bitsets, write_alive_bitset, AliveB
 pub use self::bytes::{BytesFastFieldReader, BytesFastFieldWriter};
 pub use self::error::{FastFieldNotAvailableError, Result};
 pub use self::facet_reader::FacetReader;
-pub(crate) use self::gcd::{find_gcd, GCDFastFieldCodec, GCD_DEFAULT};
+pub(crate) use self::gcd::{find_gcd, GCDReader, GCD_DEFAULT};
 pub use self::multivalued::{MultiValuedFastFieldReader, MultiValuedFastFieldWriter};
-pub use self::reader::{DynamicFastFieldReader, FastFieldReader};
+pub use self::reader::DynamicFastFieldReader;
 pub use self::readers::FastFieldReaders;
 pub(crate) use self::readers::{type_and_cardinality, FastType};
-pub use self::serializer::{CompositeFastFieldSerializer, FastFieldDataAccess, FastFieldStats};
+pub use self::serializer::{Column, CompositeFastFieldSerializer, FastFieldStats};
 pub use self::writer::{FastFieldsWriter, IntFastFieldWriter};
 use crate::schema::{Cardinality, FieldType, Type, Value};
 use crate::{DateTime, DocId};
@@ -298,9 +298,9 @@ mod tests {
     #[test]
     pub fn test_fastfield() {
         let test_fastfield = DynamicFastFieldReader::<u64>::from(vec![100, 200, 300]);
-        assert_eq!(test_fastfield.get(0), 100);
-        assert_eq!(test_fastfield.get(1), 200);
-        assert_eq!(test_fastfield.get(2), 300);
+        assert_eq!(test_fastfield.get_val(0u64), 100);
+        assert_eq!(test_fastfield.get_val(1u64), 200);
+        assert_eq!(test_fastfield.get_val(2u64), 300);
     }
 
     #[test]
@@ -330,9 +330,9 @@ mod tests {
         let composite_file = CompositeFile::open(&file)?;
         let file = composite_file.open_read(*FIELD).unwrap();
         let fast_field_reader = DynamicFastFieldReader::<u64>::open(file)?;
-        assert_eq!(fast_field_reader.get(0), 13u64);
-        assert_eq!(fast_field_reader.get(1), 14u64);
-        assert_eq!(fast_field_reader.get(2), 2u64);
+        assert_eq!(fast_field_reader.get_val(0), 13u64);
+        assert_eq!(fast_field_reader.get_val(1), 14u64);
+        assert_eq!(fast_field_reader.get_val(2), 2u64);
         Ok(())
     }
 
@@ -362,15 +362,15 @@ mod tests {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
             let fast_field_reader = DynamicFastFieldReader::<u64>::open(data)?;
-            assert_eq!(fast_field_reader.get(0), 4u64);
-            assert_eq!(fast_field_reader.get(1), 14_082_001u64);
-            assert_eq!(fast_field_reader.get(2), 3_052u64);
-            assert_eq!(fast_field_reader.get(3), 9002u64);
-            assert_eq!(fast_field_reader.get(4), 15_001u64);
-            assert_eq!(fast_field_reader.get(5), 777u64);
-            assert_eq!(fast_field_reader.get(6), 1_002u64);
-            assert_eq!(fast_field_reader.get(7), 1_501u64);
-            assert_eq!(fast_field_reader.get(8), 215u64);
+            assert_eq!(fast_field_reader.get_val(0), 4u64);
+            assert_eq!(fast_field_reader.get_val(1), 14_082_001u64);
+            assert_eq!(fast_field_reader.get_val(2), 3_052u64);
+            assert_eq!(fast_field_reader.get_val(3), 9002u64);
+            assert_eq!(fast_field_reader.get_val(4), 15_001u64);
+            assert_eq!(fast_field_reader.get_val(5), 777u64);
+            assert_eq!(fast_field_reader.get_val(6), 1_002u64);
+            assert_eq!(fast_field_reader.get_val(7), 1_501u64);
+            assert_eq!(fast_field_reader.get_val(8), 215u64);
         }
         Ok(())
     }
@@ -399,7 +399,7 @@ mod tests {
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
             let fast_field_reader = DynamicFastFieldReader::<u64>::open(data)?;
             for doc in 0..10_000 {
-                assert_eq!(fast_field_reader.get(doc), 100_000u64);
+                assert_eq!(fast_field_reader.get_val(doc), 100_000u64);
             }
         }
         Ok(())
@@ -430,10 +430,10 @@ mod tests {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
             let fast_field_reader = DynamicFastFieldReader::<u64>::open(data)?;
-            assert_eq!(fast_field_reader.get(0), 0u64);
+            assert_eq!(fast_field_reader.get_val(0), 0u64);
             for doc in 1..10_001 {
                 assert_eq!(
-                    fast_field_reader.get(doc),
+                    fast_field_reader.get_val(doc),
                     5_000_000_000_000_000_000u64 + doc as u64 - 1u64
                 );
             }
@@ -475,10 +475,9 @@ mod tests {
             assert_eq!(fast_field_reader.min_value(), -100i64);
             assert_eq!(fast_field_reader.max_value(), 9_999i64);
             for (doc, i) in (-100i64..10_000i64).enumerate() {
-                assert_eq!(fast_field_reader.get(doc as u32), i);
+                assert_eq!(fast_field_reader.get_val(doc as u64), i);
             }
-            let mut buffer = vec![0i64; 100];
-            fast_field_reader.get_range(53, &mut buffer[..]);
+            let buffer: Vec<i64> = fast_field_reader.get_range(53..154).collect();
             for i in 0..100 {
                 assert_eq!(buffer[i], -100i64 + 53i64 + i as i64);
             }
@@ -511,7 +510,7 @@ mod tests {
             let fast_fields_composite = CompositeFile::open(&file).unwrap();
             let data = fast_fields_composite.open_read(i64_field).unwrap();
             let fast_field_reader = DynamicFastFieldReader::<i64>::open(data)?;
-            assert_eq!(fast_field_reader.get(0u32), 0i64);
+            assert_eq!(fast_field_reader.get_val(0), 0i64);
         }
         Ok(())
     }
@@ -551,7 +550,7 @@ mod tests {
             let fast_field_reader = DynamicFastFieldReader::<u64>::open(data)?;
 
             for a in 0..n {
-                assert_eq!(fast_field_reader.get(a as u32), permutation[a as usize]);
+                assert_eq!(fast_field_reader.get_val(a as u64), permutation[a as usize]);
             }
         }
         Ok(())
@@ -607,9 +606,7 @@ mod tests {
         let mut all = vec![];
 
         for doc in docs {
-            let mut out = vec![];
-            ff.get_vals(doc, &mut out);
-            all.extend(out);
+            all.extend(ff.get_vals(doc));
         }
         all
     }
@@ -654,8 +651,7 @@ mod tests {
                 vec![1, 0, 0, 0, 1, 2]
             );
 
-            let mut out = vec![];
-            text_fast_field.get_vals(3, &mut out);
+            let out = text_fast_field.get_vals(3u32).collect::<Vec<_>>();
             assert_eq!(out, vec![0, 1]);
 
             let inverted_index = segment_reader.inverted_index(text_field)?;
@@ -840,22 +836,20 @@ mod tests {
         let fast_fields = segment_reader.fast_fields();
         let date_fast_field = fast_fields.date(date_field).unwrap();
         let dates_fast_field = fast_fields.dates(multi_date_field).unwrap();
-        let mut dates = vec![];
         {
-            assert_eq!(date_fast_field.get(0u32).into_timestamp_micros(), 1i64);
-            dates_fast_field.get_vals(0u32, &mut dates);
+            assert_eq!(date_fast_field.get_val(0).into_timestamp_micros(), 1i64);
+            let dates = dates_fast_field.get_vals(0u32).collect::<Vec<_>>();
             assert_eq!(dates.len(), 2);
             assert_eq!(dates[0].into_timestamp_micros(), 2i64);
             assert_eq!(dates[1].into_timestamp_micros(), 3i64);
         }
         {
-            assert_eq!(date_fast_field.get(1u32).into_timestamp_micros(), 4i64);
-            dates_fast_field.get_vals(1u32, &mut dates);
-            assert!(dates.is_empty());
+            assert_eq!(date_fast_field.get_val(1).into_timestamp_micros(), 4i64);
+            assert!(dates_fast_field.get_vals(1u32).next().is_none());
         }
         {
-            assert_eq!(date_fast_field.get(2u32).into_timestamp_micros(), 0i64);
-            dates_fast_field.get_vals(2u32, &mut dates);
+            assert_eq!(date_fast_field.get_val(2).into_timestamp_micros(), 0i64);
+            let dates = dates_fast_field.get_vals(2u32).collect::<Vec<_>>();
             assert_eq!(dates.len(), 2);
             assert_eq!(dates[0].into_timestamp_micros(), 5i64);
             assert_eq!(dates[1].into_timestamp_micros(), 6i64);
@@ -866,10 +860,10 @@ mod tests {
     #[test]
     pub fn test_fastfield_bool() {
         let test_fastfield = DynamicFastFieldReader::<bool>::from(vec![true, false, true, false]);
-        assert_eq!(test_fastfield.get(0), true);
-        assert_eq!(test_fastfield.get(1), false);
-        assert_eq!(test_fastfield.get(2), true);
-        assert_eq!(test_fastfield.get(3), false);
+        assert_eq!(test_fastfield.get_val(0), true);
+        assert_eq!(test_fastfield.get_val(1), false);
+        assert_eq!(test_fastfield.get_val(2), true);
+        assert_eq!(test_fastfield.get_val(3), false);
     }
 
     #[test]
@@ -900,10 +894,10 @@ mod tests {
         let composite_file = CompositeFile::open(&file)?;
         let file = composite_file.open_read(field).unwrap();
         let fast_field_reader = DynamicFastFieldReader::<bool>::open(file)?;
-        assert_eq!(fast_field_reader.get(0), true);
-        assert_eq!(fast_field_reader.get(1), false);
-        assert_eq!(fast_field_reader.get(2), true);
-        assert_eq!(fast_field_reader.get(3), false);
+        assert_eq!(fast_field_reader.get_val(0), true);
+        assert_eq!(fast_field_reader.get_val(1), false);
+        assert_eq!(fast_field_reader.get_val(2), true);
+        assert_eq!(fast_field_reader.get_val(3), false);
 
         Ok(())
     }
@@ -937,8 +931,8 @@ mod tests {
         let file = composite_file.open_read(field).unwrap();
         let fast_field_reader = DynamicFastFieldReader::<bool>::open(file)?;
         for i in 0..25 {
-            assert_eq!(fast_field_reader.get(i * 2), true);
-            assert_eq!(fast_field_reader.get(i * 2 + 1), false);
+            assert_eq!(fast_field_reader.get_val(i * 2), true);
+            assert_eq!(fast_field_reader.get_val(i * 2 + 1), false);
         }
 
         Ok(())
@@ -970,7 +964,7 @@ mod tests {
         let composite_file = CompositeFile::open(&file)?;
         let file = composite_file.open_read(field).unwrap();
         let fast_field_reader = DynamicFastFieldReader::<bool>::open(file)?;
-        assert_eq!(fast_field_reader.get(0), false);
+        assert_eq!(fast_field_reader.get_val(0), false);
 
         Ok(())
     }
@@ -981,13 +975,74 @@ mod bench {
     use std::collections::HashMap;
     use std::path::Path;
 
+    use fastfield_codecs::Column;
     use test::{self, Bencher};
 
     use super::tests::{generate_permutation, FIELD, SCHEMA};
     use super::*;
     use crate::directory::{CompositeFile, Directory, RamDirectory, WritePtr};
     use crate::fastfield::tests::generate_permutation_gcd;
-    use crate::fastfield::FastFieldReader;
+    use crate::schema::{NumericOptions, Schema};
+    use crate::Document;
+
+    fn multi_values(num_docs: usize, vals_per_doc: usize) -> Vec<Vec<u64>> {
+        let mut vals = vec![];
+        for _i in 0..num_docs {
+            let mut block = vec![];
+            for j in 0..vals_per_doc {
+                block.push(j as u64);
+            }
+            vals.push(block);
+        }
+
+        vals
+    }
+
+    #[bench]
+    fn bench_multi_value_fflookup(b: &mut Bencher) {
+        let num_docs = 100_000;
+
+        let path = Path::new("test");
+        let directory: RamDirectory = RamDirectory::create();
+        {
+            let options = NumericOptions::default().set_fast(Cardinality::MultiValues);
+            let mut schema_builder = Schema::builder();
+            let field = schema_builder.add_u64_field("field", options);
+            let schema = schema_builder.build();
+
+            let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
+            let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
+            for block in &multi_values(num_docs, 3) {
+                let mut doc = Document::new();
+                for val in block {
+                    doc.add_u64(field, *val);
+                }
+                fast_field_writers.add_document(&doc);
+            }
+            fast_field_writers
+                .serialize(&mut serializer, &HashMap::new(), None)
+                .unwrap();
+            serializer.close().unwrap();
+        }
+        let file = directory.open_read(&path).unwrap();
+        {
+            let fast_fields_composite = CompositeFile::open(&file).unwrap();
+            let data_idx = fast_fields_composite.open_read_with_idx(*FIELD, 0).unwrap();
+            let idx_reader = DynamicFastFieldReader::<u64>::open(data_idx).unwrap();
+
+            let data_vals = fast_fields_composite.open_read_with_idx(*FIELD, 1).unwrap();
+            let vals_reader = DynamicFastFieldReader::<u64>::open(data_vals).unwrap();
+            let fast_field_reader = MultiValuedFastFieldReader::open(idx_reader, vals_reader);
+            b.iter(|| {
+                let mut sum = 0u64;
+                for i in 0u32..num_docs as u32 {
+                    sum += fast_field_reader.get_vals(i).sum::<u64>();
+                }
+                sum
+            });
+        }
+    }
 
     #[bench]
     fn bench_intfastfield_linear_veclookup(b: &mut Bencher) {
@@ -1042,7 +1097,7 @@ mod bench {
                 let n = test::black_box(7000u32);
                 let mut a = 0u64;
                 for i in (0u32..n / 7).map(|val| val * 7) {
-                    a ^= fast_field_reader.get(i);
+                    a ^= fast_field_reader.get_val(i as u64);
                 }
                 a
             });
@@ -1074,8 +1129,8 @@ mod bench {
 
             b.iter(|| {
                 let mut a = 0u32;
-                for i in 0u32..permutation.len() as u32 {
-                    a = fast_field_reader.get(i) as u32;
+                for i in 0u64..permutation.len() as u64 {
+                    a = fast_field_reader.get_val(i) as u32;
                 }
                 a
             });
@@ -1108,7 +1163,7 @@ mod bench {
             b.iter(|| {
                 let mut a = 0u32;
                 for i in 0u32..permutation.len() as u32 {
-                    a = fast_field_reader.get(i) as u32;
+                    a = fast_field_reader.get_val(i as u64) as u32;
                 }
                 a
             });

src/fastfield/multivalued/mod.rs

@@ -36,19 +36,17 @@ mod tests {
 
         let searcher = index.reader()?.searcher();
         let segment_reader = searcher.segment_reader(0);
-        let mut vals = Vec::new();
         let multi_value_reader = segment_reader.fast_fields().u64s(field)?;
         {
-            multi_value_reader.get_vals(2, &mut vals);
+            let vals = multi_value_reader.get_vals(2u32).collect::<Vec<_>>();
             assert_eq!(&vals, &[4u64]);
         }
         {
-            multi_value_reader.get_vals(0, &mut vals);
+            let vals = multi_value_reader.get_vals(0u32).collect::<Vec<_>>();
             assert_eq!(&vals, &[1u64, 3u64]);
         }
         {
-            multi_value_reader.get_vals(1, &mut vals);
-            assert!(vals.is_empty());
+            assert!(multi_value_reader.get_vals(1u32).next().is_none());
         }
         Ok(())
     }
@@ -213,15 +211,13 @@ mod tests {
 
         let searcher = index.reader()?.searcher();
         let segment_reader = searcher.segment_reader(0);
-        let mut vals = Vec::new();
         let multi_value_reader = segment_reader.fast_fields().i64s(field).unwrap();
-        multi_value_reader.get_vals(2, &mut vals);
+        let vals = multi_value_reader.get_vals(2u32).collect::<Vec<_>>();
         assert_eq!(&vals, &[-4i64]);
-        multi_value_reader.get_vals(0, &mut vals);
+        let vals = multi_value_reader.get_vals(0u32).collect::<Vec<_>>();
         assert_eq!(&vals, &[1i64, 3i64]);
-        multi_value_reader.get_vals(1, &mut vals);
-        assert!(vals.is_empty());
-        multi_value_reader.get_vals(3, &mut vals);
+        assert!(multi_value_reader.get_vals(1u32).next().is_none());
+        let vals = multi_value_reader.get_vals(3u32).collect::<Vec<_>>();
         assert_eq!(&vals, &[-5i64, -20i64, 1i64]);
         Ok(())
     }
@@ -245,15 +241,13 @@ mod tests {
 
         let searcher = index.reader()?.searcher();
         let segment_reader = searcher.segment_reader(0);
-        let mut vals = Vec::new();
         let multi_value_reader = segment_reader.fast_fields().bools(bool_field).unwrap();
-        multi_value_reader.get_vals(2, &mut vals);
+        let vals = multi_value_reader.get_vals(2u32).collect::<Vec<_>>();
         assert_eq!(&vals, &[false]);
-        multi_value_reader.get_vals(0, &mut vals);
+        let vals = multi_value_reader.get_vals(0u32).collect::<Vec<_>>();
         assert_eq!(&vals, &[true, false]);
-        multi_value_reader.get_vals(1, &mut vals);
-        assert!(vals.is_empty());
-        multi_value_reader.get_vals(3, &mut vals);
+        assert!(multi_value_reader.get_vals(1u32).next().is_none());
+        let vals = multi_value_reader.get_vals(3u32).collect::<Vec<_>>();
         assert_eq!(&vals, &[true, true, false]);
         Ok(())
     }

src/fastfield/multivalued/reader.rs

@@ -1,6 +1,8 @@
 use std::ops::Range;
 
-use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue, MultiValueLength};
+use fastfield_codecs::Column;
+
+use crate::fastfield::{DynamicFastFieldReader, FastValue, MultiValueLength};
 use crate::DocId;
 
 /// Reader for a multivalued `u64` fast field.
@@ -16,7 +18,9 @@ pub struct MultiValuedFastFieldReader<Item: FastValue> {
     vals_reader: DynamicFastFieldReader<Item>,
 }
 
-impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
+impl<Item: FastValue> MultiValuedFastFieldReader<Item>
+where DynamicFastFieldReader<Item>: Column<Item>
+{
     pub(crate) fn open(
         idx_reader: DynamicFastFieldReader<u64>,
         vals_reader: DynamicFastFieldReader<Item>,
@@ -31,24 +35,17 @@ impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
     /// to the given document are `start..end`.
     #[inline]
     fn range(&self, doc: DocId) -> Range<u64> {
-        let start = self.idx_reader.get(doc);
-        let end = self.idx_reader.get(doc + 1);
+        let idx = doc as u64;
+        let start = self.idx_reader.get_val(idx);
+        let end = self.idx_reader.get_val(idx + 1);
         start..end
     }
 
     /// Returns the array of values associated to the given `doc`.
     #[inline]
-    fn get_vals_for_range(&self, range: Range<u64>, vals: &mut Vec<Item>) {
-        let len = (range.end - range.start) as usize;
-        vals.resize(len, Item::make_zero());
-        self.vals_reader.get_range(range.start, &mut vals[..]);
-    }
-
-    /// Returns the array of values associated to the given `doc`.
-    #[inline]
-    pub fn get_vals(&self, doc: DocId, vals: &mut Vec<Item>) {
+    pub fn get_vals(&self, doc: DocId) -> impl Iterator<Item = Item> + '_ {
         let range = self.range(doc);
-        self.get_vals_for_range(range, vals);
+        self.vals_reader.get_range(range)
     }
 
     /// Returns the minimum value for this fast field.

src/fastfield/reader.rs

@@ -3,58 +3,17 @@ use std::marker::PhantomData;
 use std::path::Path;
 
 use common::BinarySerializable;
-use fastfield_codecs::bitpacked::BitpackedReader;
-use fastfield_codecs::blockwise_linear::BlockwiseLinearReader;
-use fastfield_codecs::linear::LinearReader;
-use fastfield_codecs::{FastFieldCodecDeserializer, FastFieldCodecType, FastFieldDataAccess};
+use fastfield_codecs::bitpacked::{BitpackedCodec, BitpackedReader};
+use fastfield_codecs::blockwise_linear::{BlockwiseLinearCodec, BlockwiseLinearReader};
+use fastfield_codecs::linear::{LinearCodec, LinearReader};
+use fastfield_codecs::{Column, FastFieldCodec, FastFieldCodecType};
 
-use super::{FastValue, GCDFastFieldCodec};
+use super::gcd::open_gcd_from_bytes;
+use super::FastValue;
 use crate::directory::{CompositeFile, Directory, FileSlice, OwnedBytes, RamDirectory, WritePtr};
 use crate::error::DataCorruption;
-use crate::fastfield::{CompositeFastFieldSerializer, FastFieldsWriter};
+use crate::fastfield::{CompositeFastFieldSerializer, FastFieldsWriter, GCDReader};
 use crate::schema::{Schema, FAST};
-use crate::DocId;
-
-/// FastFieldReader is the trait to access fast field data.
-pub trait FastFieldReader<Item: FastValue>: Clone {
-    /// Return the value associated to the given document.
-    ///
-    /// This accessor should return as fast as possible.
-    ///
-    /// # Panics
-    ///
-    /// May panic if `doc` is greater than the segment
-    fn get(&self, doc: DocId) -> Item;
-
-    /// Fills an output buffer with the fast field values
-    /// associated with the `DocId` going from
-    /// `start` to `start + output.len()`.
-    ///
-    /// Regardless of the type of `Item`, this method works
-    /// - transmuting the output array
-    /// - extracting the `Item`s as if they were `u64`
-    /// - possibly converting the `u64` value to the right type.
-    ///
-    /// # Panics
-    ///
-    /// May panic if `start + output.len()` is greater than
-    /// the segment's `maxdoc`.
-    fn get_range(&self, start: u64, output: &mut [Item]);
-
-    /// Returns the minimum value for this fast field.
-    ///
-    /// The min value does not take in account of possible
-    /// deleted document, and should be considered as a lower bound
-    /// of the actual minimum value.
-    fn min_value(&self) -> Item;
-
-    /// Returns the maximum value for this fast field.
-    ///
-    /// The max value does not take in account of possible
-    /// deleted document, and should be considered as an upper bound
-    /// of the actual maximum value.
-    fn max_value(&self) -> Item;
-}
 
 #[derive(Clone)]
 /// DynamicFastFieldReader wraps different readers to access
@@ -68,11 +27,11 @@ pub enum DynamicFastFieldReader<Item: FastValue> {
     BlockwiseLinear(FastFieldReaderCodecWrapper<Item, BlockwiseLinearReader>),
 
     /// GCD and Bitpacked compressed fastfield data.
-    BitpackedGCD(FastFieldReaderCodecWrapper<Item, GCDFastFieldCodec<BitpackedReader>>),
+    BitpackedGCD(FastFieldReaderCodecWrapper<Item, GCDReader<BitpackedReader>>),
     /// GCD and Linear interpolated values + bitpacked
-    LinearGCD(FastFieldReaderCodecWrapper<Item, GCDFastFieldCodec<LinearReader>>),
+    LinearGCD(FastFieldReaderCodecWrapper<Item, GCDReader<LinearReader>>),
     /// GCD and Blockwise linear interpolated values + bitpacked
-    BlockwiseLinearGCD(FastFieldReaderCodecWrapper<Item, GCDFastFieldCodec<BlockwiseLinearReader>>),
+    BlockwiseLinearGCD(FastFieldReaderCodecWrapper<Item, GCDReader<BlockwiseLinearReader>>),
 }
 
 impl<Item: FastValue> DynamicFastFieldReader<Item> {
@@ -81,59 +40,39 @@ impl<Item: FastValue> DynamicFastFieldReader<Item> {
         mut bytes: OwnedBytes,
         codec_type: FastFieldCodecType,
     ) -> crate::Result<DynamicFastFieldReader<Item>> {
-        let reader = match codec_type {
-            FastFieldCodecType::Bitpacked => {
-                DynamicFastFieldReader::Bitpacked(FastFieldReaderCodecWrapper::<
-                    Item,
-                    BitpackedReader,
-                >::open_from_bytes(bytes)?)
-            }
-            FastFieldCodecType::Linear => DynamicFastFieldReader::Linear(
-                FastFieldReaderCodecWrapper::<Item, LinearReader>::open_from_bytes(bytes)?,
-            ),
-            FastFieldCodecType::BlockwiseLinear => {
-                DynamicFastFieldReader::BlockwiseLinear(FastFieldReaderCodecWrapper::<
-                    Item,
-                    BlockwiseLinearReader,
-                >::open_from_bytes(bytes)?)
-            }
-            FastFieldCodecType::Gcd => {
-                let codec_type = FastFieldCodecType::deserialize(&mut bytes)?;
-                match codec_type {
-                    FastFieldCodecType::Bitpacked => {
-                        DynamicFastFieldReader::BitpackedGCD(FastFieldReaderCodecWrapper::<
-                            Item,
-                            GCDFastFieldCodec<BitpackedReader>,
-                        >::open_from_bytes(
-                            bytes
-                        )?)
-                    }
-                    FastFieldCodecType::Linear => {
-                        DynamicFastFieldReader::LinearGCD(FastFieldReaderCodecWrapper::<
-                            Item,
-                            GCDFastFieldCodec<LinearReader>,
-                        >::open_from_bytes(
-                            bytes
-                        )?)
-                    }
-                    FastFieldCodecType::BlockwiseLinear => {
-                        DynamicFastFieldReader::BlockwiseLinearGCD(FastFieldReaderCodecWrapper::<
-                            Item,
-                            GCDFastFieldCodec<BlockwiseLinearReader>,
-                        >::open_from_bytes(
-                            bytes
-                        )?)
-                    }
-                    FastFieldCodecType::Gcd => {
-                        return Err(DataCorruption::comment_only(
+        let reader =
+            match codec_type {
+                FastFieldCodecType::Bitpacked => DynamicFastFieldReader::Bitpacked(
+                    BitpackedCodec::open_from_bytes(bytes)?.into(),
+                ),
+                FastFieldCodecType::Linear => {
+                    DynamicFastFieldReader::Linear(LinearCodec::open_from_bytes(bytes)?.into())
+                }
+                FastFieldCodecType::BlockwiseLinear => DynamicFastFieldReader::BlockwiseLinear(
+                    BlockwiseLinearCodec::open_from_bytes(bytes)?.into(),
+                ),
+                FastFieldCodecType::Gcd => {
+                    let codec_type = FastFieldCodecType::deserialize(&mut bytes)?;
+                    match codec_type {
+                        FastFieldCodecType::Bitpacked => DynamicFastFieldReader::BitpackedGCD(
+                            open_gcd_from_bytes::<BitpackedCodec>(bytes)?.into(),
+                        ),
+                        FastFieldCodecType::Linear => DynamicFastFieldReader::LinearGCD(
+                            open_gcd_from_bytes::<LinearCodec>(bytes)?.into(),
+                        ),
+                        FastFieldCodecType::BlockwiseLinear => {
+                            DynamicFastFieldReader::BlockwiseLinearGCD(
+                                open_gcd_from_bytes::<BlockwiseLinearCodec>(bytes)?.into(),
+                            )
+                        }
+                        FastFieldCodecType::Gcd => return Err(DataCorruption::comment_only(
                             "Gcd codec wrapped into another gcd codec. This combination is not \
                              allowed.",
                         )
-                        .into())
+                        .into()),
                     }
                 }
-            }
-        };
+            };
         Ok(reader)
     }
 
@@ -145,27 +84,16 @@ impl<Item: FastValue> DynamicFastFieldReader<Item> {
     }
 }
 
-impl<Item: FastValue> FastFieldReader<Item> for DynamicFastFieldReader<Item> {
+impl<Item: FastValue> Column<Item> for DynamicFastFieldReader<Item> {
     #[inline]
-    fn get(&self, doc: DocId) -> Item {
+    fn get_val(&self, idx: u64) -> Item {
         match self {
-            Self::Bitpacked(reader) => reader.get(doc),
-            Self::Linear(reader) => reader.get(doc),
-            Self::BlockwiseLinear(reader) => reader.get(doc),
-            Self::BitpackedGCD(reader) => reader.get(doc),
-            Self::LinearGCD(reader) => reader.get(doc),
-            Self::BlockwiseLinearGCD(reader) => reader.get(doc),
-        }
-    }
-    #[inline]
-    fn get_range(&self, start: u64, output: &mut [Item]) {
-        match self {
-            Self::Bitpacked(reader) => reader.get_range(start, output),
-            Self::Linear(reader) => reader.get_range(start, output),
-            Self::BlockwiseLinear(reader) => reader.get_range(start, output),
-            Self::BitpackedGCD(reader) => reader.get_range(start, output),
-            Self::LinearGCD(reader) => reader.get_range(start, output),
-            Self::BlockwiseLinearGCD(reader) => reader.get_range(start, output),
+            Self::Bitpacked(reader) => reader.get_val(idx),
+            Self::Linear(reader) => reader.get_val(idx),
+            Self::BlockwiseLinear(reader) => reader.get_val(idx),
+            Self::BitpackedGCD(reader) => reader.get_val(idx),
+            Self::LinearGCD(reader) => reader.get_val(idx),
+            Self::BlockwiseLinearGCD(reader) => reader.get_val(idx),
         }
     }
     fn min_value(&self) -> Item {
@@ -188,6 +116,17 @@ impl<Item: FastValue> FastFieldReader<Item> for DynamicFastFieldReader<Item> {
             Self::BlockwiseLinearGCD(reader) => reader.max_value(),
         }
     }
+
+    fn num_vals(&self) -> u64 {
+        match self {
+            Self::Bitpacked(reader) => reader.num_vals(),
+            Self::Linear(reader) => reader.num_vals(),
+            Self::BlockwiseLinear(reader) => reader.num_vals(),
+            Self::BitpackedGCD(reader) => reader.num_vals(),
+            Self::LinearGCD(reader) => reader.num_vals(),
+            Self::BlockwiseLinearGCD(reader) => reader.num_vals(),
+        }
+    }
 }
 
 /// Wrapper for accessing a fastfield.
@@ -199,61 +138,26 @@ pub struct FastFieldReaderCodecWrapper<Item: FastValue, CodecReader> {
     _phantom: PhantomData<Item>,
 }
 
-impl<Item: FastValue, C: FastFieldDataAccess + FastFieldCodecDeserializer>
-    FastFieldReaderCodecWrapper<Item, C>
+impl<Item: FastValue, CodecReader> From<CodecReader>
+    for FastFieldReaderCodecWrapper<Item, CodecReader>
 {
-    /// Opens a fast field given a file.
-    pub fn open(file: FileSlice) -> crate::Result<Self> {
-        let mut bytes = file.read_bytes()?;
-        let codec_code = bytes.read_u8();
-        let codec_type = FastFieldCodecType::from_code(codec_code).ok_or_else(|| {
-            DataCorruption::comment_only("Unknown codec code does not exist `{codec_code}`")
-        })?;
-        assert_eq!(
-            FastFieldCodecType::Bitpacked,
-            codec_type,
-            "Tried to open fast field as bitpacked encoded (id=1), but got serializer with \
-             different id"
-        );
-        Self::open_from_bytes(bytes)
-    }
-    /// Opens a fast field given the bytes.
-    pub fn open_from_bytes(bytes: OwnedBytes) -> crate::Result<Self> {
-        let reader = C::open_from_bytes(bytes)?;
-        Ok(FastFieldReaderCodecWrapper {
+    fn from(reader: CodecReader) -> Self {
+        FastFieldReaderCodecWrapper {
             reader,
             _phantom: PhantomData,
-        })
-    }
-
-    #[inline]
-    pub(crate) fn get_u64(&self, doc: u64) -> Item {
-        let data = self.reader.get_val(doc);
-        Item::from_u64(data)
-    }
-
-    /// Internally `multivalued` also use SingleValue Fast fields.
-    /// It works as follows... A first column contains the list of start index
-    /// for each document, a second column contains the actual values.
-    ///
-    /// The values associated to a given doc, are then
-    ///  `second_column[first_column.get(doc)..first_column.get(doc+1)]`.
-    ///
-    /// Which means single value fast field reader can be indexed internally with
-    /// something different from a `DocId`. For this use case, we want to use `u64`
-    /// values.
-    ///
-    /// See `get_range` for an actual documentation about this method.
-    pub(crate) fn get_range_u64(&self, start: u64, output: &mut [Item]) {
-        for (i, out) in output.iter_mut().enumerate() {
-            *out = self.get_u64(start + (i as u64));
         }
     }
 }
 
-impl<Item: FastValue, C: FastFieldDataAccess + FastFieldCodecDeserializer + Clone>
-    FastFieldReader<Item> for FastFieldReaderCodecWrapper<Item, C>
-{
+impl<Item: FastValue, D: Column> FastFieldReaderCodecWrapper<Item, D> {
+    #[inline]
+    pub(crate) fn get_u64(&self, idx: u64) -> Item {
+        let data = self.reader.get_val(idx);
+        Item::from_u64(data)
+    }
+}
+
+impl<Item: FastValue, C: Column + Clone> Column<Item> for FastFieldReaderCodecWrapper<Item, C> {
     /// Return the value associated to the given document.
     ///
     /// This accessor should return as fast as possible.
@@ -262,25 +166,8 @@ impl<Item: FastValue, C: FastFieldDataAccess + FastFieldCodecDeserializer + Clon
     ///
     /// May panic if `doc` is greater than the segment
     // `maxdoc`.
-    fn get(&self, doc: DocId) -> Item {
-        self.get_u64(u64::from(doc))
-    }
-
-    /// Fills an output buffer with the fast field values
-    /// associated with the `DocId` going from
-    /// `start` to `start + output.len()`.
-    ///
-    /// Regardless of the type of `Item`, this method works
-    /// - transmuting the output array
-    /// - extracting the `Item`s as if they were `u64`
-    /// - possibly converting the `u64` value to the right type.
-    ///
-    /// # Panics
-    ///
-    /// May panic if `start + output.len()` is greater than
-    /// the segment's `maxdoc`.
-    fn get_range(&self, start: u64, output: &mut [Item]) {
-        self.get_range_u64(start, output);
+    fn get_val(&self, idx: u64) -> Item {
+        self.get_u64(idx)
     }
 
     /// Returns the minimum value for this fast field.
@@ -300,6 +187,10 @@ impl<Item: FastValue, C: FastFieldDataAccess + FastFieldCodecDeserializer + Clon
     fn max_value(&self) -> Item {
         Item::from_u64(self.reader.max_value())
     }
+
+    fn num_vals(&self) -> u64 {
+        self.reader.num_vals()
+    }
 }
 
 impl<Item: FastValue> From<Vec<Item>> for DynamicFastFieldReader<Item> {

src/fastfield/serializer/mod.rs

@@ -3,11 +3,11 @@ use std::num::NonZeroU64;
 
 use common::{BinarySerializable, CountingWriter};
 use fastdivide::DividerU64;
-pub use fastfield_codecs::bitpacked::{BitpackedSerializer, BitpackedSerializerLegacy};
-use fastfield_codecs::blockwise_linear::BlockwiseLinearSerializer;
-use fastfield_codecs::linear::LinearSerializer;
+pub use fastfield_codecs::bitpacked::{BitpackedCodec, BitpackedSerializerLegacy};
+use fastfield_codecs::blockwise_linear::BlockwiseLinearCodec;
+use fastfield_codecs::linear::LinearCodec;
 use fastfield_codecs::FastFieldCodecType;
-pub use fastfield_codecs::{FastFieldCodecSerializer, FastFieldDataAccess, FastFieldStats};
+pub use fastfield_codecs::{Column, FastFieldCodec, FastFieldStats};
 
 use super::{find_gcd, ALL_CODECS, GCD_DEFAULT};
 use crate::directory::{CompositeWrite, WritePtr};
@@ -64,15 +64,13 @@ impl From<FastFieldCodecType> for FastFieldCodecEnableCheck {
 
 // use this, when this is merged and stabilized explicit_generic_args_with_impl_trait
 // https://github.com/rust-lang/rust/pull/86176
-fn codec_estimation<T: FastFieldCodecSerializer, A: FastFieldDataAccess>(
-    fastfield_accessor: &A,
+fn codec_estimation<C: FastFieldCodec, D: Column>(
+    fastfield_accessor: &D,
     estimations: &mut Vec<(f32, FastFieldCodecType)>,
 ) {
-    if !T::is_applicable(fastfield_accessor) {
-        return;
+    if let Some(ratio) = C::estimate(fastfield_accessor) {
+        estimations.push((ratio, C::CODEC_TYPE));
     }
-    let ratio = T::estimate(fastfield_accessor);
-    estimations.push((ratio, T::CODEC_TYPE));
 }
 
 impl CompositeFastFieldSerializer {
@@ -99,7 +97,7 @@ impl CompositeFastFieldSerializer {
     pub fn create_auto_detect_u64_fast_field(
         &mut self,
         field: Field,
-        fastfield_accessor: impl FastFieldDataAccess,
+        fastfield_accessor: impl Column,
     ) -> io::Result<()> {
         self.create_auto_detect_u64_fast_field_with_idx(field, fastfield_accessor, 0)
     }
@@ -119,7 +117,7 @@ impl CompositeFastFieldSerializer {
     pub fn create_auto_detect_u64_fast_field_with_idx(
         &mut self,
         field: Field,
-        fastfield_accessor: impl FastFieldDataAccess,
+        fastfield_accessor: impl Column,
         idx: usize,
     ) -> io::Result<()> {
         let min_value = fastfield_accessor.min_value();
@@ -138,7 +136,7 @@ impl CompositeFastFieldSerializer {
         }
 
         Self::write_header(field_write, FastFieldCodecType::Gcd)?;
-        struct GCDWrappedFFAccess<T: FastFieldDataAccess> {
+        struct GCDWrappedFFAccess<T: Column> {
             fastfield_accessor: T,
             base_value: u64,
             max_value: u64,
@@ -146,7 +144,7 @@ impl CompositeFastFieldSerializer {
             gcd: DividerU64,
         }
 
-        impl<T: FastFieldDataAccess> FastFieldDataAccess for GCDWrappedFFAccess<T> {
+        impl<T: Column> Column for GCDWrappedFFAccess<T> {
             fn get_val(&self, position: u64) -> u64 {
                 self.gcd
                     .divide(self.fastfield_accessor.get_val(position) - self.base_value)
@@ -199,18 +197,18 @@ impl CompositeFastFieldSerializer {
         codec_enable_checker: FastFieldCodecEnableCheck,
         field: Field,
         field_write: &mut CountingWriter<W>,
-        fastfield_accessor: impl FastFieldDataAccess,
+        fastfield_accessor: impl Column,
     ) -> io::Result<()> {
         let mut estimations = vec![];
 
         if codec_enable_checker.is_enabled(FastFieldCodecType::Bitpacked) {
-            codec_estimation::<BitpackedSerializer, _>(&fastfield_accessor, &mut estimations);
+            codec_estimation::<BitpackedCodec, _>(&fastfield_accessor, &mut estimations);
         }
         if codec_enable_checker.is_enabled(FastFieldCodecType::Linear) {
-            codec_estimation::<LinearSerializer, _>(&fastfield_accessor, &mut estimations);
+            codec_estimation::<LinearCodec, _>(&fastfield_accessor, &mut estimations);
         }
         if codec_enable_checker.is_enabled(FastFieldCodecType::BlockwiseLinear) {
-            codec_estimation::<BlockwiseLinearSerializer, _>(&fastfield_accessor, &mut estimations);
+            codec_estimation::<BlockwiseLinearCodec, _>(&fastfield_accessor, &mut estimations);
         }
         if let Some(broken_estimation) = estimations.iter().find(|estimation| estimation.0.is_nan())
         {
@@ -229,13 +227,13 @@ impl CompositeFastFieldSerializer {
         Self::write_header(field_write, codec_type)?;
         match codec_type {
             FastFieldCodecType::Bitpacked => {
-                BitpackedSerializer::serialize(field_write, &fastfield_accessor)?;
+                BitpackedCodec::serialize(field_write, &fastfield_accessor)?;
             }
             FastFieldCodecType::Linear => {
-                LinearSerializer::serialize(field_write, &fastfield_accessor)?;
+                LinearCodec::serialize(field_write, &fastfield_accessor)?;
             }
             FastFieldCodecType::BlockwiseLinear => {
-                BlockwiseLinearSerializer::serialize(field_write, &fastfield_accessor)?;
+                BlockwiseLinearCodec::serialize(field_write, &fastfield_accessor)?;
             }
             FastFieldCodecType::Gcd => {
                 return Err(io::Error::new(

src/fastfield/writer.rs

@@ -2,7 +2,7 @@ use std::collections::HashMap;
 use std::io;
 
 use common;
-use fastfield_codecs::FastFieldDataAccess;
+use fastfield_codecs::Column;
 use fnv::FnvHashMap;
 use tantivy_bitpacker::BlockedBitpacker;
 
@@ -384,7 +384,7 @@ struct WriterFastFieldAccessProvider<'map, 'bitp> {
     vals: &'bitp BlockedBitpacker,
     stats: FastFieldStats,
 }
-impl<'map, 'bitp> FastFieldDataAccess for WriterFastFieldAccessProvider<'map, 'bitp> {
+impl<'map, 'bitp> Column for WriterFastFieldAccessProvider<'map, 'bitp> {
     /// Return the value associated to the given doc.
     ///
     /// Whenever possible use the Iterator passed to the fastfield creation instead, for performance

src/indexer/doc_id_mapping.rs

@@ -143,8 +143,9 @@ pub(crate) fn get_doc_id_mapping_from_field(
 
 #[cfg(test)]
 mod tests_indexsorting {
+    use fastfield_codecs::Column;
+
     use crate::collector::TopDocs;
-    use crate::fastfield::FastFieldReader;
     use crate::indexer::doc_id_mapping::DocIdMapping;
     use crate::query::QueryParser;
     use crate::schema::{Schema, *};
@@ -464,21 +465,19 @@ mod tests_indexsorting {
         let my_number = index.schema().get_field("my_number").unwrap();
 
         let fast_field = fast_fields.u64(my_number).unwrap();
-        assert_eq!(fast_field.get(0u32), 10u64);
-        assert_eq!(fast_field.get(1u32), 20u64);
-        assert_eq!(fast_field.get(2u32), 30u64);
+        assert_eq!(fast_field.get_val(0), 10u64);
+        assert_eq!(fast_field.get_val(1), 20u64);
+        assert_eq!(fast_field.get_val(2), 30u64);
 
         let multi_numbers = index.schema().get_field("multi_numbers").unwrap();
         let multifield = fast_fields.u64s(multi_numbers).unwrap();
-        let mut vals = vec![];
-        multifield.get_vals(0u32, &mut vals);
+        let vals = multifield.get_vals(0u32).collect::<Vec<_>>();
         assert_eq!(vals, &[] as &[u64]);
-        let mut vals = vec![];
-        multifield.get_vals(1u32, &mut vals);
+
+        let vals = multifield.get_vals(1u32).collect::<Vec<_>>();
         assert_eq!(vals, &[5, 6]);
 
-        let mut vals = vec![];
-        multifield.get_vals(2u32, &mut vals);
+        let vals = multifield.get_vals(2u32).collect::<Vec<_>>();
         assert_eq!(vals, &[3]);
         Ok(())
     }

src/indexer/index_writer.rs

@@ -777,6 +777,7 @@ impl Drop for IndexWriter {
 mod tests {
     use std::collections::{HashMap, HashSet};
 
+    use fastfield_codecs::Column;
     use proptest::prelude::*;
     use proptest::prop_oneof;
     use proptest::strategy::Strategy;
@@ -785,7 +786,6 @@ mod tests {
     use crate::collector::TopDocs;
     use crate::directory::error::LockError;
     use crate::error::*;
-    use crate::fastfield::FastFieldReader;
     use crate::indexer::NoMergePolicy;
     use crate::query::{QueryParser, TermQuery};
     use crate::schema::{
@@ -1327,7 +1327,7 @@ mod tests {
         let fast_field_reader = segment_reader.fast_fields().u64(id_field)?;
         let in_order_alive_ids: Vec<u64> = segment_reader
             .doc_ids_alive()
-            .map(|doc| fast_field_reader.get(doc))
+            .map(|doc| fast_field_reader.get_val(doc as u64))
             .collect();
         assert_eq!(&in_order_alive_ids[..], &[9, 8, 7, 6, 5, 4, 1, 0]);
         Ok(())
@@ -1493,7 +1493,7 @@ mod tests {
                 let ff_reader = segment_reader.fast_fields().u64(id_field).unwrap();
                 segment_reader
                     .doc_ids_alive()
-                    .map(move |doc| ff_reader.get(doc))
+                    .map(move |doc| ff_reader.get_val(doc as u64))
             })
             .collect();
 
@@ -1504,7 +1504,7 @@ mod tests {
                 let ff_reader = segment_reader.fast_fields().u64(id_field).unwrap();
                 segment_reader
                     .doc_ids_alive()
-                    .map(move |doc| ff_reader.get(doc))
+                    .map(move |doc| ff_reader.get_val(doc as u64))
             })
             .collect();
 
@@ -1535,13 +1535,11 @@ mod tests {
             let ff_reader = segment_reader.fast_fields().u64s(multi_numbers).unwrap();
             let bool_ff_reader = segment_reader.fast_fields().bools(multi_bools).unwrap();
             for doc in segment_reader.doc_ids_alive() {
-                let mut vals = vec![];
-                ff_reader.get_vals(doc, &mut vals);
+                let vals = ff_reader.get_vals(doc).collect::<Vec<_>>();
                 assert_eq!(vals.len(), 2);
                 assert_eq!(vals[0], vals[1]);
 
-                let mut bool_vals = vec![];
-                bool_ff_reader.get_vals(doc, &mut bool_vals);
+                let bool_vals = bool_ff_reader.get_vals(doc).collect::<Vec<_>>();
                 assert_eq!(bool_vals.len(), 2);
                 assert_ne!(bool_vals[0], bool_vals[1]);
 
@@ -1622,7 +1620,7 @@ mod tests {
                 facet_reader
                     .facet_from_ord(facet_ords[0], &mut facet)
                     .unwrap();
-                let id = ff_reader.get(doc_id);
+                let id = ff_reader.get_val(doc_id as u64);
                 let facet_expected = Facet::from(&("/cola/".to_string() + &id.to_string()));
 
                 assert_eq!(facet, facet_expected);

src/indexer/merger.rs

@@ -4,14 +4,13 @@ use std::sync::Arc;
 
 use itertools::Itertools;
 use measure_time::debug_time;
-use tantivy_bitpacker::minmax;
 
 use crate::core::{Segment, SegmentReader};
 use crate::docset::{DocSet, TERMINATED};
 use crate::error::DataCorruption;
 use crate::fastfield::{
-    AliveBitSet, CompositeFastFieldSerializer, DynamicFastFieldReader, FastFieldDataAccess,
-    FastFieldReader, FastFieldStats, MultiValueLength, MultiValuedFastFieldReader,
+    AliveBitSet, Column, CompositeFastFieldSerializer, DynamicFastFieldReader, FastFieldStats,
+    MultiValueLength, MultiValuedFastFieldReader,
 };
 use crate::fieldnorm::{FieldNormReader, FieldNormReaders, FieldNormsSerializer, FieldNormsWriter};
 use crate::indexer::doc_id_mapping::{expect_field_id_for_sort_field, SegmentDocIdMapping};
@@ -88,7 +87,7 @@ pub struct IndexMerger {
 }
 
 fn compute_min_max_val(
-    u64_reader: &impl FastFieldReader<u64>,
+    u64_reader: &impl Column<u64>,
     segment_reader: &SegmentReader,
 ) -> Option<(u64, u64)> {
     if segment_reader.max_doc() == 0 {
@@ -102,11 +101,11 @@ fn compute_min_max_val(
     }
     // some deleted documents,
     // we need to recompute the max / min
-    minmax(
-        segment_reader
-            .doc_ids_alive()
-            .map(|doc_id| u64_reader.get(doc_id)),
-    )
+    segment_reader
+        .doc_ids_alive()
+        .map(|doc_id| u64_reader.get_val(doc_id as u64))
+        .minmax()
+        .into_option()
 }
 
 struct TermOrdinalMapping {
@@ -376,13 +375,13 @@ impl IndexMerger {
             fast_field_readers: &'a Vec<DynamicFastFieldReader<u64>>,
             stats: FastFieldStats,
         }
-        impl<'a> FastFieldDataAccess for SortedDocIdFieldAccessProvider<'a> {
+        impl<'a> Column for SortedDocIdFieldAccessProvider<'a> {
             fn get_val(&self, doc: u64) -> u64 {
                 let DocAddress {
                     doc_id,
                     segment_ord,
                 } = self.doc_id_mapping.get_old_doc_addr(doc as u32);
-                self.fast_field_readers[segment_ord as usize].get(doc_id)
+                self.fast_field_readers[segment_ord as usize].get_val(doc_id as u64)
             }
 
             fn iter(&self) -> Box<dyn Iterator<Item = u64> + '_> {
@@ -392,7 +391,7 @@ impl IndexMerger {
                         .map(|old_doc_addr| {
                             let fast_field_reader =
                                 &self.fast_field_readers[old_doc_addr.segment_ord as usize];
-                            fast_field_reader.get(old_doc_addr.doc_id)
+                            fast_field_reader.get_val(old_doc_addr.doc_id as u64)
                         }),
                 )
             }
@@ -429,7 +428,7 @@ impl IndexMerger {
 
         let everything_is_in_order = reader_ordinal_and_field_accessors
             .into_iter()
-            .map(|reader| reader.1)
+            .map(|(_, col)| Arc::new(col))
             .tuple_windows()
             .all(|(field_accessor1, field_accessor2)| {
                 if sort_by_field.order.is_asc() {
@@ -444,7 +443,7 @@ impl IndexMerger {
     pub(crate) fn get_sort_field_accessor(
         reader: &SegmentReader,
         sort_by_field: &IndexSortByField,
-    ) -> crate::Result<impl FastFieldReader<u64>> {
+    ) -> crate::Result<impl Column> {
         let field_id = expect_field_id_for_sort_field(reader.schema(), sort_by_field)?; // for now expect fastfield, but not strictly required
         let value_accessor = reader.fast_fields().u64_lenient(field_id)?;
         Ok(value_accessor)
@@ -453,7 +452,7 @@ impl IndexMerger {
     pub(crate) fn get_reader_with_sort_field_accessor(
         &self,
         sort_by_field: &IndexSortByField,
-    ) -> crate::Result<Vec<(SegmentOrdinal, impl FastFieldReader<u64> + Clone)>> {
+    ) -> crate::Result<Vec<(SegmentOrdinal, impl Column)>> {
         let reader_ordinal_and_field_accessors = self
             .readers
             .iter()
@@ -506,8 +505,8 @@ impl IndexMerger {
             doc_id_reader_pair
                 .into_iter()
                 .kmerge_by(|a, b| {
-                    let val1 = a.2.get(a.0);
-                    let val2 = b.2.get(b.0);
+                    let val1 = a.2.get_val(a.0 as u64);
+                    let val2 = b.2.get_val(b.0 as u64);
                     if sort_by_field.order == Order::Asc {
                         val1 < val2
                     } else {
@@ -578,7 +577,8 @@ impl IndexMerger {
             offsets: &'a [u64],
             stats: FastFieldStats,
         }
-        impl<'a> FastFieldDataAccess for FieldIndexAccessProvider<'a> {
+        impl<'a> Column for FieldIndexAccessProvider<'a> {
+            #[inline]
             fn get_val(&self, doc: u64) -> u64 {
                 self.offsets[doc as usize]
             }
@@ -669,15 +669,13 @@ impl IndexMerger {
         {
             let mut serialize_vals =
                 fast_field_serializer.new_u64_fast_field_with_idx(field, 0u64, max_term_ord, 1)?;
-            let mut vals = Vec::with_capacity(100);
 
             for old_doc_addr in doc_id_mapping.iter_old_doc_addrs() {
                 let term_ordinal_mapping: &[TermOrdinal] =
                     term_ordinal_mappings.get_segment(old_doc_addr.segment_ord as usize);
 
                 let ff_reader = &fast_field_reader[old_doc_addr.segment_ord as usize];
-                ff_reader.get_vals(old_doc_addr.doc_id, &mut vals);
-                for &prev_term_ord in &vals {
+                for prev_term_ord in ff_reader.get_vals(old_doc_addr.doc_id) {
                     let new_term_ord = term_ordinal_mapping[prev_term_ord as usize];
                     serialize_vals.add_val(new_term_ord)?;
                 }
@@ -730,8 +728,6 @@ impl IndexMerger {
         let mut max_value = u64::MIN;
         let mut num_vals = 0;
 
-        let mut vals = Vec::with_capacity(100);
-
         let mut ff_readers = Vec::new();
 
         // Our values are bitpacked and we need to know what should be
@@ -749,12 +745,11 @@ impl IndexMerger {
                      Please report.",
                 );
             for doc in reader.doc_ids_alive() {
-                ff_reader.get_vals(doc, &mut vals);
-                for &val in &vals {
+                for val in ff_reader.get_vals(doc) {
                     min_value = cmp::min(val, min_value);
                     max_value = cmp::max(val, max_value);
+                    num_vals += 1;
                 }
-                num_vals += vals.len();
             }
             ff_readers.push(ff_reader);
             // TODO optimize when no deletes
@@ -778,7 +773,7 @@ impl IndexMerger {
             offsets: Vec<u64>,
             stats: FastFieldStats,
         }
-        impl<'a> FastFieldDataAccess for SortedDocIdMultiValueAccessProvider<'a> {
+        impl<'a> Column for SortedDocIdMultiValueAccessProvider<'a> {
             fn get_val(&self, pos: u64) -> u64 {
                 // use the offsets index to find the doc_id which will contain the position.
                 // the offsets are strictly increasing so we can do a simple search on it.
@@ -797,11 +792,10 @@ impl IndexMerger {
                 let num_vals = self.fast_field_readers[old_doc_addr.segment_ord as usize]
                     .get_len(old_doc_addr.doc_id);
                 assert!(num_vals >= pos_in_values);
-                let mut vals = Vec::new();
                 self.fast_field_readers[old_doc_addr.segment_ord as usize]
-                    .get_vals(old_doc_addr.doc_id, &mut vals);
-
-                vals[pos_in_values as usize]
+                    .get_vals(old_doc_addr.doc_id)
+                    .nth(pos_in_values as usize)
+                    .expect("computation error in SortedDocIdMultiValueAccessProvider")
             }
 
             fn iter(&self) -> Box<dyn Iterator<Item = u64> + '_> {
@@ -811,9 +805,7 @@ impl IndexMerger {
                         .flat_map(|old_doc_addr| {
                             let ff_reader =
                                 &self.fast_field_readers[old_doc_addr.segment_ord as usize];
-                            let mut vals = Vec::new();
-                            ff_reader.get_vals(old_doc_addr.doc_id, &mut vals);
-                            vals.into_iter()
+                            ff_reader.get_vals(old_doc_addr.doc_id)
                         }),
                 )
             }
@@ -1200,6 +1192,7 @@ impl IndexMerger {
 #[cfg(test)]
 mod tests {
     use byteorder::{BigEndian, ReadBytesExt};
+    use fastfield_codecs::Column;
     use schema::FAST;
 
     use crate::collector::tests::{
@@ -1207,7 +1200,6 @@ mod tests {
     };
     use crate::collector::{Count, FacetCollector};
     use crate::core::Index;
-    use crate::fastfield::FastFieldReader;
     use crate::query::{AllQuery, BooleanQuery, Scorer, TermQuery};
     use crate::schema::{
         Cardinality, Document, Facet, FacetOptions, IndexRecordOption, NumericOptions, Term,
@@ -1976,49 +1968,32 @@ mod tests {
         }
         let reader = index.reader()?;
         let searcher = reader.searcher();
-        let mut vals: Vec<u64> = Vec::new();
 
         {
             let segment = searcher.segment_reader(0u32);
             let ff_reader = segment.fast_fields().u64s(int_field).unwrap();
 
-            ff_reader.get_vals(0, &mut vals);
-            assert_eq!(&vals, &[1, 2]);
-
-            ff_reader.get_vals(1, &mut vals);
-            assert_eq!(&vals, &[1, 2, 3]);
-
-            ff_reader.get_vals(2, &mut vals);
-            assert_eq!(&vals, &[4, 5]);
-
-            ff_reader.get_vals(3, &mut vals);
-            assert_eq!(&vals, &[1, 2]);
-
-            ff_reader.get_vals(4, &mut vals);
-            assert_eq!(&vals, &[1, 5]);
-
-            ff_reader.get_vals(5, &mut vals);
-            assert_eq!(&vals, &[3]);
-
-            ff_reader.get_vals(6, &mut vals);
-            assert_eq!(&vals, &[17]);
+            assert_eq!(&ff_reader.get_vals(0).collect::<Vec<_>>(), &[1, 2]);
+            assert_eq!(&ff_reader.get_vals(1).collect::<Vec<_>>(), &[1, 2, 3]);
+            assert_eq!(&ff_reader.get_vals(2).collect::<Vec<_>>(), &[4, 5]);
+            assert_eq!(&ff_reader.get_vals(3).collect::<Vec<_>>(), &[1, 2]);
+            assert_eq!(&ff_reader.get_vals(4).collect::<Vec<_>>(), &[1, 5]);
+            assert_eq!(&ff_reader.get_vals(5).collect::<Vec<_>>(), &[3]);
+            assert_eq!(&ff_reader.get_vals(6).collect::<Vec<_>>(), &[17]);
         }
 
         {
             let segment = searcher.segment_reader(1u32);
             let ff_reader = segment.fast_fields().u64s(int_field).unwrap();
-            ff_reader.get_vals(0, &mut vals);
-            assert_eq!(&vals, &[28, 27]);
+            assert_eq!(&ff_reader.get_vals(0).collect::<Vec<_>>(), &[28, 27]);
 
-            ff_reader.get_vals(1, &mut vals);
-            assert_eq!(&vals, &[1_000]);
+            assert_eq!(&ff_reader.get_vals(1).collect::<Vec<_>>(), &[1000]);
         }
 
         {
             let segment = searcher.segment_reader(2u32);
             let ff_reader = segment.fast_fields().u64s(int_field).unwrap();
-            ff_reader.get_vals(0, &mut vals);
-            assert_eq!(&vals, &[20]);
+            assert_eq!(&ff_reader.get_vals(0).collect::<Vec<_>>(), &[20]);
         }
 
         // Merging the segments
@@ -2035,35 +2010,16 @@ mod tests {
             let segment = searcher.segment_reader(0u32);
             let ff_reader = segment.fast_fields().u64s(int_field).unwrap();
 
-            ff_reader.get_vals(0, &mut vals);
-            assert_eq!(&vals, &[1, 2]);
-
-            ff_reader.get_vals(1, &mut vals);
-            assert_eq!(&vals, &[1, 2, 3]);
-
-            ff_reader.get_vals(2, &mut vals);
-            assert_eq!(&vals, &[4, 5]);
-
-            ff_reader.get_vals(3, &mut vals);
-            assert_eq!(&vals, &[1, 2]);
-
-            ff_reader.get_vals(4, &mut vals);
-            assert_eq!(&vals, &[1, 5]);
-
-            ff_reader.get_vals(5, &mut vals);
-            assert_eq!(&vals, &[3]);
-
-            ff_reader.get_vals(6, &mut vals);
-            assert_eq!(&vals, &[17]);
-
-            ff_reader.get_vals(7, &mut vals);
-            assert_eq!(&vals, &[28, 27]);
-
-            ff_reader.get_vals(8, &mut vals);
-            assert_eq!(&vals, &[1_000]);
-
-            ff_reader.get_vals(9, &mut vals);
-            assert_eq!(&vals, &[20]);
+            assert_eq!(&ff_reader.get_vals(0).collect::<Vec<_>>(), &[1, 2]);
+            assert_eq!(&ff_reader.get_vals(1).collect::<Vec<_>>(), &[1, 2, 3]);
+            assert_eq!(&ff_reader.get_vals(2).collect::<Vec<_>>(), &[4, 5]);
+            assert_eq!(&ff_reader.get_vals(3).collect::<Vec<_>>(), &[1, 2]);
+            assert_eq!(&ff_reader.get_vals(4).collect::<Vec<_>>(), &[1, 5]);
+            assert_eq!(&ff_reader.get_vals(5).collect::<Vec<_>>(), &[3]);
+            assert_eq!(&ff_reader.get_vals(6).collect::<Vec<_>>(), &[17]);
+            assert_eq!(&ff_reader.get_vals(7).collect::<Vec<_>>(), &[28, 27]);
+            assert_eq!(&ff_reader.get_vals(8).collect::<Vec<_>>(), &[1_000]);
+            assert_eq!(&ff_reader.get_vals(9).collect::<Vec<_>>(), &[20]);
         }
         Ok(())
     }

src/indexer/merger_sorted_index_test.rs

@@ -1,8 +1,10 @@
 #[cfg(test)]
 mod tests {
+    use fastfield_codecs::Column;
+
     use crate::collector::TopDocs;
     use crate::core::Index;
-    use crate::fastfield::{AliveBitSet, FastFieldReader, MultiValuedFastFieldReader};
+    use crate::fastfield::{AliveBitSet, MultiValuedFastFieldReader};
     use crate::query::QueryParser;
     use crate::schema::{
         self, BytesOptions, Cardinality, Facet, FacetOptions, IndexRecordOption, NumericOptions,
@@ -186,17 +188,17 @@ mod tests {
 
         let fast_fields = segment_reader.fast_fields();
         let fast_field = fast_fields.u64(int_field).unwrap();
-        assert_eq!(fast_field.get(5u32), 1u64);
-        assert_eq!(fast_field.get(4u32), 2u64);
-        assert_eq!(fast_field.get(3u32), 3u64);
+        assert_eq!(fast_field.get_val(5), 1u64);
+        assert_eq!(fast_field.get_val(4), 2u64);
+        assert_eq!(fast_field.get_val(3), 3u64);
         if force_disjunct_segment_sort_values {
-            assert_eq!(fast_field.get(2u32), 20u64);
-            assert_eq!(fast_field.get(1u32), 100u64);
+            assert_eq!(fast_field.get_val(2u64), 20u64);
+            assert_eq!(fast_field.get_val(1u64), 100u64);
         } else {
-            assert_eq!(fast_field.get(2u32), 10u64);
-            assert_eq!(fast_field.get(1u32), 20u64);
+            assert_eq!(fast_field.get_val(2u64), 10u64);
+            assert_eq!(fast_field.get_val(1u64), 20u64);
         }
-        assert_eq!(fast_field.get(0u32), 1_000u64);
+        assert_eq!(fast_field.get_val(0u64), 1_000u64);
 
         // test new field norm mapping
         {
@@ -373,17 +375,15 @@ mod tests {
 
         let fast_fields = segment_reader.fast_fields();
         let fast_field = fast_fields.u64(int_field).unwrap();
-        assert_eq!(fast_field.get(0u32), 1u64);
-        assert_eq!(fast_field.get(1u32), 2u64);
-        assert_eq!(fast_field.get(2u32), 3u64);
-        assert_eq!(fast_field.get(3u32), 10u64);
-        assert_eq!(fast_field.get(4u32), 20u64);
-        assert_eq!(fast_field.get(5u32), 1_000u64);
+        assert_eq!(fast_field.get_val(0), 1u64);
+        assert_eq!(fast_field.get_val(1), 2u64);
+        assert_eq!(fast_field.get_val(2), 3u64);
+        assert_eq!(fast_field.get_val(3), 10u64);
+        assert_eq!(fast_field.get_val(4), 20u64);
+        assert_eq!(fast_field.get_val(5), 1_000u64);
 
         let get_vals = |fast_field: &MultiValuedFastFieldReader<u64>, doc_id: u32| -> Vec<u64> {
-            let mut vals = vec![];
-            fast_field.get_vals(doc_id, &mut vals);
-            vals
+            fast_field.get_vals(doc_id).collect()
         };
         let fast_fields = segment_reader.fast_fields();
         let fast_field = fast_fields.u64s(multi_numbers).unwrap();
@@ -478,11 +478,11 @@ mod tests {
 #[cfg(all(test, feature = "unstable"))]
 mod bench_sorted_index_merge {
 
+    use fastfield_codecs::Column;
     use test::{self, Bencher};
 
     use crate::core::Index;
-    // use cratedoc_id, readerdoc_id_mappinglet vals = reader.fate::schema;
-    use crate::fastfield::{DynamicFastFieldReader, FastFieldReader};
+    use crate::fastfield::DynamicFastFieldReader;
     use crate::indexer::merger::IndexMerger;
     use crate::schema::{Cardinality, NumericOptions, Schema};
     use crate::{IndexSettings, IndexSortByField, IndexWriter, Order};
@@ -544,7 +544,7 @@ mod bench_sorted_index_merge {
             // add values in order of the new doc_ids
             let mut val = 0;
             for (doc_id, _reader, field_reader) in sorted_doc_ids {
-                val = field_reader.get(doc_id);
+                val = field_reader.get_val(doc_id as u64);
             }
 
             val

src/lib.rs

@@ -421,6 +421,7 @@ pub struct DocAddress {
 #[cfg(test)]
 pub mod tests {
     use common::{BinarySerializable, FixedSize};
+    use fastfield_codecs::Column;
     use rand::distributions::{Bernoulli, Uniform};
     use rand::rngs::StdRng;
     use rand::{Rng, SeedableRng};
@@ -429,7 +430,6 @@ pub mod tests {
     use crate::collector::tests::TEST_COLLECTOR_WITH_SCORE;
     use crate::core::SegmentReader;
     use crate::docset::{DocSet, TERMINATED};
-    use crate::fastfield::FastFieldReader;
     use crate::merge_policy::NoMergePolicy;
     use crate::query::BooleanQuery;
     use crate::schema::*;
@@ -1036,21 +1036,21 @@ pub mod tests {
             let fast_field_reader_opt = segment_reader.fast_fields().u64(fast_field_unsigned);
             assert!(fast_field_reader_opt.is_ok());
             let fast_field_reader = fast_field_reader_opt.unwrap();
-            assert_eq!(fast_field_reader.get(0), 4u64)
+            assert_eq!(fast_field_reader.get_val(0), 4u64)
         }
 
         {
             let fast_field_reader_res = segment_reader.fast_fields().i64(fast_field_signed);
             assert!(fast_field_reader_res.is_ok());
             let fast_field_reader = fast_field_reader_res.unwrap();
-            assert_eq!(fast_field_reader.get(0), 4i64)
+            assert_eq!(fast_field_reader.get_val(0), 4i64)
         }
 
         {
             let fast_field_reader_res = segment_reader.fast_fields().f64(fast_field_float);
             assert!(fast_field_reader_res.is_ok());
             let fast_field_reader = fast_field_reader_res.unwrap();
-            assert_eq!(fast_field_reader.get(0), 4f64)
+            assert_eq!(fast_field_reader.get_val(0), 4f64)
         }
         Ok(())
     }

src/snippet/mod.rs

@@ -639,10 +639,7 @@ Survey in 2016, 2017, and 2018."#;
     #[test]
     fn test_collapse_overlapped_ranges() {
         assert_eq!(&collapse_overlapped_ranges(&[0..1, 2..3,]), &[0..1, 2..3]);
-        assert_eq!(
-            collapse_overlapped_ranges(&vec![0..1, 1..2,]),
-            vec![0..1, 1..2]
-        );
+        assert_eq!(collapse_overlapped_ranges(&[0..1, 1..2,]), &[0..1, 1..2]);
         assert_eq!(collapse_overlapped_ranges(&[0..2, 1..2,]), vec![0..2]);
         assert_eq!(collapse_overlapped_ranges(&[0..2, 1..3,]), vec![0..3]);
         assert_eq!(collapse_overlapped_ranges(&[0..3, 1..2,]), vec![0..3]);