diff --git a/README.md b/README.md
index 24ab790..fa49847 100644
--- a/README.md
+++ b/README.md
@@ -18,16 +18,16 @@ Unless otherwise specified, all codecs support `&[u32]` only.
 ```text
 * BP32
 * Copy
+* FastBinaryPacking16
+* FastBinaryPacking32
 * FastBinaryPacking8
 * FastPFor128 (both `&[u32]` and `&[u64]`)
 * FastPFor256 (both `&[u32]` and `&[u64]`)
-* FastBinaryPacking16
-* FastBinaryPacking32
 * MaskedVByte
 * NewPFor
 * OptPFor
-* PFor2008
 * PFor
+* PFor2008
 * SimdBinaryPacking
 * SimdFastPFor128
 * SimdFastPFor256
diff --git a/src/codec.rs b/src/codec.rs
index 8bff34c..f56ebf3 100644
--- a/src/codec.rs
+++ b/src/codec.rs
@@ -1,6 +1,6 @@
 use bytemuck::{Pod, cast_slice};
 
-use crate::FastPForError;
+use crate::FastPForResult;
 
 /// Internal default for max decompressed length. Used by trait defaults and C++ FFI.
 #[inline]
@@ -23,13 +23,15 @@ pub(crate) fn default_max_decoded_len(compressed_words: usize) -> usize {
 ///
 /// # Implementing this trait
 ///
-/// ```rust,ignore
+/// ```
+/// # use fastpfor::{BlockCodec, FastPForResult};
+/// struct MyCodec;
 /// impl BlockCodec for MyCodec {
 ///     type Block = [u32; 256];
-///     fn encode_blocks(&self, blocks: &[[u32; 256]], out: &mut Vec<u32>)
-///         -> Result<(), FastPForError> { ... }
-///     fn decode_blocks(&self, input: &[u32], expected_len: Option<u32>,
-///         out: &mut Vec<u32>) -> Result<usize, FastPForError> { ... }
+///     fn encode_blocks(&mut self, blocks: &[[u32; 256]], out: &mut Vec<u32>)
+///         -> FastPForResult<()> { todo!() }
+///     fn decode_blocks(&mut self, input: &[u32], expected_len: Option<u32>,
+///         out: &mut Vec<u32>) -> FastPForResult<usize> { todo!() }
 /// }
 /// ```
 pub trait BlockCodec {
@@ -54,11 +56,7 @@ pub trait BlockCodec {
     ///
     /// No remainder is possible — the caller must split the input first using
     /// [`slice_to_blocks`] and handle any remainder separately.
-    fn encode_blocks(
-        &mut self,
-        blocks: &[Self::Block],
-        out: &mut Vec<u32>,
-    ) -> Result<(), FastPForError>;
+    fn encode_blocks(&mut self, blocks: &[Self::Block], out: &mut Vec<u32>) -> FastPForResult<()>;
 
     /// Decompress blocks from `input`, using the length stored in the header.
     ///
@@ -75,7 +73,7 @@ pub trait BlockCodec {
         input: &[u32],
         expected_len: Option<u32>,
         out: &mut Vec<u32>,
-    ) -> Result<usize, FastPForError>;
+    ) -> FastPForResult<usize>;
 
     /// Maximum decompressed element count for a given compressed input length.
     /// Reject `expected_len` values exceeding this to avoid allocation from bad data.
@@ -100,9 +98,9 @@ pub trait BlockCodec {
 #[cfg(feature = "cpp")]
 pub trait BlockCodec64 {
     /// Compress 64-bit integers into a 32-bit word stream.
-    fn encode64(&mut self, input: &[u64], out: &mut Vec<u32>) -> Result<(), FastPForError>;
+    fn encode64(&mut self, input: &[u64], out: &mut Vec<u32>) -> FastPForResult<()>;
     /// Decompress 64-bit integers from a 32-bit word stream.
-    fn decode64(&mut self, input: &[u32], out: &mut Vec<u64>) -> Result<(), FastPForError>;
+    fn decode64(&mut self, input: &[u32], out: &mut Vec<u64>) -> FastPForResult<()>;
 }
 
 /// Compresses and decompresses an arbitrary-length `&[u32]` slice.
@@ -113,7 +111,7 @@ pub trait BlockCodec64 {
 /// to produce an `AnyLenCodec`.
 pub trait AnyLenCodec {
     /// Compress an arbitrary-length slice of `u32` values.
-    fn encode(&mut self, input: &[u32], out: &mut Vec<u32>) -> Result<(), FastPForError>;
+    fn encode(&mut self, input: &[u32], out: &mut Vec<u32>) -> FastPForResult<()>;
 
     /// Maximum decompressed element count for a given compressed input length.
     /// Reject `expected_len` values exceeding this to avoid allocation from bad data.
@@ -140,7 +138,7 @@ pub trait AnyLenCodec {
         input: &[u32],
         out: &mut Vec<u32>,
         expected_len: Option<u32>,
-    ) -> Result<(), FastPForError>;
+    ) -> FastPForResult<()>;
 }
 
 /// Split a flat `&[u32]` into `(&[Blocks::Block], &[u32])` without copying.
@@ -154,7 +152,8 @@ pub trait AnyLenCodec {
 ///
 /// # Example
 ///
-/// ```rust,ignore
+/// ```ignore
+/// # use fastpfor::{slice_to_blocks, FastPForBlock256};
 /// let data: Vec<u32> = (0..600).collect(); // 2 × 256 + 88 remainder
 /// let (blocks, remainder) = slice_to_blocks::<FastPForBlock256>(&data);
 /// assert_eq!(blocks.len(), 2);    // 2 blocks of [u32; 256]
diff --git a/src/cpp/codecs.rs b/src/cpp/codecs.rs
index 2ac6750..2def24a 100644
--- a/src/cpp/codecs.rs
+++ b/src/cpp/codecs.rs
@@ -1,6 +1,6 @@
 use cxx::UniquePtr;
 
-use crate::FastPForError;
+use crate::FastPForResult;
 use crate::codec::{AnyLenCodec, BlockCodec64};
 use crate::cpp::ffi;
 use crate::cpp::wrappers::{
@@ -37,7 +37,7 @@ macro_rules! implement_cpp_codecs {
             }
 
             impl AnyLenCodec for $name {
-                fn encode(&mut self, input: &[u32], out: &mut Vec<u32>) -> Result<(), FastPForError> {
+                fn encode(&mut self, input: &[u32], out: &mut Vec<u32>) -> FastPForResult<()> {
                     encode32_to_vec_ffi(&self.0, input, out)
                 }
 
@@ -46,7 +46,7 @@ macro_rules! implement_cpp_codecs {
                     input: &[u32],
                     out: &mut Vec<u32>,
                     expected_len: Option<u32>,
-                ) -> Result<(), FastPForError> {
+                ) -> FastPForResult<()> {
                     decode32_anylen_ffi(&self.0, input, out, expected_len)
                 }
             }
@@ -161,10 +161,10 @@ macro_rules! implement_cpp_codecs_64 {
     ($($name:ident => $ffi:ident ,)*) => {
         $(
             impl BlockCodec64 for $name {
-                fn encode64(&mut self, input: &[u64], out: &mut Vec<u32>) -> Result<(), FastPForError> {
+                fn encode64(&mut self, input: &[u64], out: &mut Vec<u32>) -> FastPForResult<()> {
                     encode64_to_vec_ffi(&self.0, input, out)
                 }
-                fn decode64(&mut self, input: &[u32], out: &mut Vec<u64>) -> Result<(), FastPForError> {
+                fn decode64(&mut self, input: &[u32], out: &mut Vec<u64>) -> FastPForResult<()> {
                     decode64_to_vec_ffi(&self.0, input, out)
                 }
             }
@@ -192,7 +192,7 @@ pub(crate) mod tests {
     }
 
     /// C++ `fastpfor256_codec` returns `CompositeCodec<FastPFor<8>, VariableByte>` — already
-    /// any-length. Use it directly; do not wrap in Rust `CppComposite`.
+    /// any-length. Use it directly; do not wrap in Rust `CompositeCodec`.
     #[test]
     fn test_cpp_fastpfor256_composite_anylen() {
         let mut codec = CppFastPFor256::new();
diff --git a/src/cpp/wrappers.rs b/src/cpp/wrappers.rs
index 7d600a3..90a58ad 100644
--- a/src/cpp/wrappers.rs
+++ b/src/cpp/wrappers.rs
@@ -1,6 +1,6 @@
 use cxx::UniquePtr;
 
-use crate::FastPForError;
+use crate::FastPForResult;
 use crate::codec::default_max_decoded_len;
 use crate::cpp::ffi;
 use crate::helpers::AsUsize;
@@ -14,7 +14,7 @@ pub fn encode32_to_vec_ffi(
     codec: &UniquePtr<ffi::IntegerCODEC>,
     input: &[u32],
     out: &mut Vec<u32>,
-) -> Result<(), FastPForError> {
+) -> FastPForResult<()> {
     let capacity = input.len() * 2 + 1024;
     let start = out.len();
     out.resize(start + capacity, 0);
@@ -28,7 +28,7 @@ fn decode32_to_vec_ffi(
     input: &[u32],
     out: &mut Vec<u32>,
     capacity: usize,
-) -> Result<(), FastPForError> {
+) -> FastPForResult<()> {
     if !input.is_empty() {
         let start = out.len();
         out.resize(start + capacity, 0);
@@ -43,7 +43,7 @@ pub fn decode32_anylen_ffi(
     input: &[u32],
     out: &mut Vec<u32>,
     expected_len: Option<u32>,
-) -> Result<(), FastPForError> {
+) -> FastPForResult<()> {
     let max = default_max_decoded_len(input.len());
     let capacity = if let Some(n) = expected_len {
         n.is_valid_expected(max)?
@@ -64,7 +64,7 @@ pub fn encode64_to_vec_ffi(
     codec: &UniquePtr<ffi::IntegerCODEC>,
     input: &[u64],
     out: &mut Vec<u32>,
-) -> Result<(), FastPForError> {
+) -> FastPForResult<()> {
     let capacity = input.len() * 3 + 1024;
     let start = out.len();
     out.resize(start + capacity, 0);
@@ -77,7 +77,7 @@ pub fn decode64_to_vec_ffi(
     codec: &UniquePtr<ffi::IntegerCODEC>,
     input: &[u32],
     out: &mut Vec<u64>,
-) -> Result<(), FastPForError> {
+) -> FastPForResult<()> {
     if !input.is_empty() {
         // C++ decodeArray needs output buffer. Variable-byte can pack multiple values per word.
         let capacity = input.len().saturating_mul(4);
diff --git a/src/helpers.rs b/src/helpers.rs
index 8db9446..3c2a1d1 100644
--- a/src/helpers.rs
+++ b/src/helpers.rs
@@ -1,4 +1,4 @@
-use crate::FastPForError;
+use crate::{FastPForError, FastPForResult};
 
 /// Finds the greatest multiple of `factor` that is less than or equal to `value`.
 #[cfg_attr(feature = "cpp", allow(dead_code))]
@@ -18,7 +18,7 @@ pub trait AsUsize: Eq + Copy {
 
     #[inline]
     #[cfg(feature = "cpp")]
-    fn is_decoded_mismatch(self, expected: impl AsUsize) -> Result<(), FastPForError> {
+    fn is_decoded_mismatch(self, expected: impl AsUsize) -> FastPForResult<()> {
         let actual = self.as_usize();
         let expected = expected.as_usize();
         if self.as_usize() == expected {
@@ -31,7 +31,7 @@ pub trait AsUsize: Eq + Copy {
     /// Returns an error if `expected` exceeds `max`.
     #[inline]
     #[cfg(feature = "cpp")]
-    fn is_valid_expected(self, max: impl AsUsize) -> Result<usize, FastPForError> {
+    fn is_valid_expected(self, max: impl AsUsize) -> FastPForResult<usize> {
         let expected = self.as_usize();
         let max = max.as_usize();
         if expected > max {
@@ -69,12 +69,12 @@ impl AsUsize for u32 {
 
 #[cfg_attr(feature = "cpp", allow(dead_code))]
 pub trait GetWithErr<T> {
-    fn get_val(&self, pos: impl AsUsize) -> Result<T, FastPForError>;
+    fn get_val(&self, pos: impl AsUsize) -> FastPForResult<T>;
 }
 
 impl<T: Copy> GetWithErr<T> for &[T] {
     #[inline]
-    fn get_val(&self, pos: impl AsUsize) -> Result<T, FastPForError> {
+    fn get_val(&self, pos: impl AsUsize) -> FastPForResult<T> {
         self.get(pos.as_usize())
             .copied()
             .ok_or(FastPForError::NotEnoughData)
@@ -83,7 +83,7 @@ impl<T: Copy> GetWithErr<T> for &[T] {
 
 impl<T: Copy> GetWithErr<T> for Vec<T> {
     #[inline]
-    fn get_val(&self, pos: impl AsUsize) -> Result<T, FastPForError> {
+    fn get_val(&self, pos: impl AsUsize) -> FastPForResult<T> {
         self.as_slice().get_val(pos)
     }
 }
diff --git a/src/lib.rs b/src/lib.rs
index a0cd734..20c4948 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -2,18 +2,13 @@
 #![cfg_attr(docsrs, feature(doc_cfg))]
 #![doc = include_str!("../README.md")]
 
-#[cfg(not(any(feature = "cpp", feature = "rust",)))]
+#[cfg(not(any(feature = "cpp", feature = "rust")))]
 compile_error!("At least one of the features 'cpp' or 'rust' must be enabled");
 
 // Error types are always available regardless of which codec features are enabled.
 mod error;
 pub use error::{FastPForError, FastPForResult};
 
-// FIXME: need decide on the external API. Some ideas:
-//  - offer two sets of similar APIs - rust and cpp ffi
-//  - it will be possible to enable/disable each with a feature flag
-//  - introduce a new feature-agnostic API that will forward to either
-//  - if both are enabled, forward to the more stable (ffi probably)
 #[cfg(feature = "cpp")]
 /// Rust wrapper for the [`FastPFOR` C++ library](https://github.com/fast-pack/FastPFor)
 pub mod cpp;
diff --git a/src/rust/integer_compression/fastpfor.rs b/src/rust/integer_compression/fastpfor.rs
index fbf55e8..09a80b0 100644
--- a/src/rust/integer_compression/fastpfor.rs
+++ b/src/rust/integer_compression/fastpfor.rs
@@ -2,6 +2,7 @@ use std::array;
 use std::io::Cursor;
 use std::num::NonZeroU32;
 
+use bytemuck::cast_slice;
 use bytes::{Buf as _, BufMut as _, BytesMut};
 
 use crate::helpers::{GetWithErr, bits, greatest_multiple};
@@ -176,13 +177,13 @@ impl FastPFOR {
     /// - Writes header, packed data, metadata bytes, and exception values.
     ///
     /// # Arguments
-    /// * `thissize` - Must be multiple of `block_size`
-    /// * `input_offset` - Advanced by `thissize`
+    /// * `this_size` - Must be multiple of `block_size`
+    /// * `input_offset` - Advanced by `this_size`
     /// * `output_offset` - Advanced by compressed size
     fn encode_page(
         &mut self,
         input: &[u32],
-        thissize: u32,
+        this_size: u32,
         input_offset: &mut Cursor<u32>,
         output: &mut [u32],
         output_offset: &mut Cursor<u32>,
@@ -196,7 +197,7 @@ impl FastPFOR {
         self.bytes_container.clear();
 
         let mut tmp_input_offset = input_offset.position() as u32;
-        let final_input_offset = tmp_input_offset + thissize - self.block_size;
+        let final_input_offset = tmp_input_offset + this_size - self.block_size;
         while tmp_input_offset <= final_input_offset {
             self.best_bit_from_data(input, tmp_input_offset);
             self.bytes_container.put_u8(self.optimal_bits);
@@ -331,9 +332,9 @@ impl FastPFOR {
     /// unpacks regular values per block, patches in exceptions by position.
     ///
     /// # Arguments
-    /// * `thissize` - Expected decompressed integer count
+    /// * `this_size` - Expected decompressed integer count
     /// * `input_offset` - Advanced by bytes read
-    /// * `output_offset` - Advanced by `thissize`
+    /// * `output_offset` - Advanced by `this_size`
     #[expect(clippy::too_many_lines)]
     fn decode_page(
         &mut self,
@@ -341,7 +342,7 @@ impl FastPFOR {
         input_offset: &mut Cursor<u32>,
         output: &mut [u32],
         output_offset: &mut Cursor<u32>,
-        thissize: u32,
+        this_size: u32,
     ) -> FastPForResult<()> {
         let n = u32::try_from(input.len())
             .map_err(|_| FastPForError::InvalidInputLength(input.len()))?;
@@ -362,7 +363,7 @@ impl FastPFOR {
         // The C++ encoder uses a raw `memcpy` of bytes into the u32 output (no endian
         // conversion), and the decoder does a raw reinterpret_cast back -- both native byte
         // order. `cast_slice` is the exact Rust equivalent: a safe, zero-copy native view.
-        let input_bytes: &[u8] = bytemuck::cast_slice(input);
+        let input_bytes: &[u8] = cast_slice(input);
         let mut byte_pos = (inexcept as usize)
             .checked_mul(4)
             .filter(|&bp| bp <= input_bytes.len())
@@ -448,7 +449,7 @@ impl FastPFOR {
         let mut tmp_output_offset = output_offset.position() as u32;
         let mut tmp_input_offset = input_offset.position() as u32;
 
-        let run_end = thissize / self.block_size;
+        let run_end = this_size / self.block_size;
         for _ in 0..run_end {
             let bits = input_bytes.get_val(byte_pos)?;
             if bits > 32 {
diff --git a/tests/decode_error_paths.rs b/tests/decode_error_paths.rs
index 234556d..68c49e7 100644
--- a/tests/decode_error_paths.rs
+++ b/tests/decode_error_paths.rs
@@ -11,6 +11,7 @@
 
 use std::io::Cursor;
 
+use bytemuck::cast_slice_mut;
 use fastpfor::rust::{BLOCK_SIZE_128, DEFAULT_PAGE_SIZE, FastPFOR, Integer, Skippable};
 
 // ── helpers ──────────────────────────────────────────────────────────────────
@@ -229,7 +230,7 @@ fn decode_exception_partial_group_not_enough_data() {
 fn decode_block_b_too_large() {
     let (mut compressed, _) = compressed_with_exceptions();
     let start = meta_byte_start(&compressed);
-    let bytes: &mut [u8] = bytemuck::cast_slice_mut(&mut compressed);
+    let bytes: &mut [u8] = cast_slice_mut(&mut compressed);
     bytes[start] = 33; // overwrite best_b of block 0
     assert!(try_decode(&compressed).is_err());
 }
@@ -289,7 +290,7 @@ fn find_exception_block(bytes: &[u8], meta_start: usize) -> Option<(usize, usize
 fn decode_exception_maxbits_too_large() {
     let (mut compressed, _) = compressed_with_exceptions();
     let start = meta_byte_start(&compressed);
-    let bytes: &mut [u8] = bytemuck::cast_slice_mut(&mut compressed);
+    let bytes: &mut [u8] = cast_slice_mut(&mut compressed);
     if let Some((_, _, mb_off)) = find_exception_block(bytes, start) {
         bytes[mb_off] = 33;
     }
@@ -301,7 +302,7 @@ fn decode_exception_maxbits_too_large() {
 fn decode_exception_index_underflow() {
     let (mut compressed, _) = compressed_with_exceptions();
     let start = meta_byte_start(&compressed);
-    let bytes: &mut [u8] = bytemuck::cast_slice_mut(&mut compressed);
+    let bytes: &mut [u8] = cast_slice_mut(&mut compressed);
     if let Some((bb_off, _, mb_off)) = find_exception_block(bytes, start) {
         bytes[mb_off] = bytes[bb_off].saturating_sub(1); // maxbits < best_b
     }
@@ -313,7 +314,7 @@ fn decode_exception_index_underflow() {
 fn decode_exception_index_zero() {
     let (mut compressed, _) = compressed_with_exceptions();
     let start = meta_byte_start(&compressed);
-    let bytes: &mut [u8] = bytemuck::cast_slice_mut(&mut compressed);
+    let bytes: &mut [u8] = cast_slice_mut(&mut compressed);
     if let Some((bb_off, _, mb_off)) = find_exception_block(bytes, start) {
         bytes[mb_off] = bytes[bb_off]; // maxbits == best_b → index 0
     }
@@ -368,7 +369,7 @@ fn decode_index1_pos_out_of_block() {
     buf.truncate(out_off.position() as usize);
 
     let start = meta_byte_start(&buf);
-    let bytes: &mut [u8] = bytemuck::cast_slice_mut(&mut buf);
+    let bytes: &mut [u8] = cast_slice_mut(&mut buf);
     if let Some((bb_off, _, mb_off)) = find_exception_block(bytes, start) {
         if bytes[mb_off].wrapping_sub(bytes[bb_off]) == 1 && mb_off + 1 < bytes.len() {
             bytes[mb_off + 1] = 200; // position 200 >= block_size 128
@@ -432,7 +433,7 @@ fn decode_exception_pos_out_of_block() {
     buf.truncate(out_off.position() as usize);
 
     let start = meta_byte_start(&buf);
-    let bytes: &mut [u8] = bytemuck::cast_slice_mut(&mut buf);
+    let bytes: &mut [u8] = cast_slice_mut(&mut buf);
     if let Some((bb_off, _, mb_off)) = find_exception_block(bytes, start) {
         if bytes[mb_off].wrapping_sub(bytes[bb_off]) > 1 && mb_off + 1 < bytes.len() {
             bytes[mb_off + 1] = 200; // position 200 >= block_size 128