[BuildsOnTop Of Julien and Vinoo's work] [WIP] ALP pseudo decimal encoding#3444

Draft

prtkgaur wants to merge 24 commits intoapache:masterfrom

prtkgaur:gh540-alp-pseudoDecimal-encoding

Commits on Mar 11, 2026

ALP: Add AlpConstants with encoding tables and format constants
sfc-gh-pgaur
committed
ALP: Add AlpEncoderDecoder with core encode/decode logic
sfc-gh-pgaur
committed
ALP: Add AlpCompression for single-vector compress/decompress
sfc-gh-pgaur
committed
ALP: Add AlpSampler for sampling-based encoding preset generation
sfc-gh-pgaur
committed
ALP: Add AlpWrapper for page-level encode/decode with 7-byte header
sfc-gh-pgaur
committed
ALP: Add incremental AlpValuesWriter for float and double columns
sfc-gh-pgaur
committed
ALP: Add lazy AlpValuesReader for float and double columns
sfc-gh-pgaur
committed
ALP: Add cross-implementation tests and fix encode/decode to match C++
sfc-gh-pgaur
committed
ALP: Add encoding benchmark for float and double throughput
sfc-gh-pgaur
committed
ALP: Wire up Encoding.ALP in column Encoding enum and AlpValuesWriter
sfc-gh-pgaur
committed
ALP: Apply spotless formatting fixes
sfc-gh-pgaur
committed
ALP: Add cross-language interop test reading C++ ALP-encoded parquet files
sfc-gh-pgaur
committed
ALP: Add writer pipeline integration and bidirectional cross-language tests
sfc-gh-pgaur
committed
ALP: Add float32 cross-language tests and generator support
sfc-gh-pgaur
committed
ALP: Add codec-level and pipeline throughput benchmarks
sfc-gh-pgaur
committed
ALP: Add pipeline integration, reader buffer reuse, and validation
sfc-gh-pgaur
committed
ALP: Add encoding comparison benchmark (ALP vs ZSTD vs BSS+ZSTD)
sfc-gh-pgaur
committed
ALP: Pre-allocate long[] encoded buffer for double vector decodes
sfc-gh-pgaur
committed
ALP: Use unpack32Values for long bit unpacking
sfc-gh-pgaur
committed
ALP: Inline decode with hoisted multipliers for double decompression
sfc-gh-pgaur
committed
ALP: Use real Spotify dataset in codec throughput benchmark
sfc-gh-pgaur
committed
ALP: Remove duplicate CSV test data from parquet-column
sfc-gh-pgaur
committed
ALP: Gzip compress CSV test data and regenerate float parquet files
sfc-gh-pgaur
committed
ALP: Use real Spotify dataset in encoding comparison benchmark
sfc-gh-pgaur
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[BuildsOnTop Of Julien and Vinoo's work] [WIP] ALP pseudo decimal encoding#3444

[BuildsOnTop Of Julien and Vinoo's work] [WIP] ALP pseudo decimal encoding#3444
prtkgaur wants to merge 24 commits intoapache:masterfrom
prtkgaur:gh540-alp-pseudoDecimal-encoding

Commits on Mar 11, 2026

ALP: Add AlpConstants with encoding tables and format constants

ALP: Add AlpEncoderDecoder with core encode/decode logic

ALP: Add AlpCompression for single-vector compress/decompress

ALP: Add AlpSampler for sampling-based encoding preset generation

ALP: Add AlpWrapper for page-level encode/decode with 7-byte header

ALP: Add incremental AlpValuesWriter for float and double columns

ALP: Add lazy AlpValuesReader for float and double columns

ALP: Add cross-implementation tests and fix encode/decode to match C++

ALP: Add encoding benchmark for float and double throughput

ALP: Wire up Encoding.ALP in column Encoding enum and AlpValuesWriter

ALP: Apply spotless formatting fixes

ALP: Add cross-language interop test reading C++ ALP-encoded parquet files

ALP: Add writer pipeline integration and bidirectional cross-language tests

ALP: Add float32 cross-language tests and generator support

ALP: Add codec-level and pipeline throughput benchmarks

ALP: Add pipeline integration, reader buffer reuse, and validation

ALP: Add encoding comparison benchmark (ALP vs ZSTD vs BSS+ZSTD)

ALP: Pre-allocate long[] encoded buffer for double vector decodes

ALP: Use unpack32Values for long bit unpacking

ALP: Inline decode with hoisted multipliers for double decompression

ALP: Use real Spotify dataset in codec throughput benchmark

ALP: Remove duplicate CSV test data from parquet-column

ALP: Gzip compress CSV test data and regenerate float parquet files

ALP: Use real Spotify dataset in encoding comparison benchmark