This package implements its own custom LZ4 encoding and decoding. It would be nice to have this in ChunkCodecLibLz4 too. The current supported LZ4 formats in ChunkCodecLibLz4 are Numcodecs (for Zarr), HDF5, Frame, and Block, but it looks like ASDF has its own slightly different format.
IIUC this is defined in https://github.com/asdf-format/asdf/blob/14c40c83fc3cca7285b59be3976bd29e2ba35543/asdf/_compression.py#L65
I also found https://github.com/asdf-format/asdf-compression/blob/25ca4e8b1c42ed2c329ff3694601ec4dbf504f14/asdf_compression/lz4f/compressor.py#L5
Based on this, it looks like Lz4 frame should be using "lz4f" as the 4-byte key not "lz4\0" to be readable by python (but I haven't set up tests for this yet).
This package implements its own custom LZ4 encoding and decoding. It would be nice to have this in
ChunkCodecLibLz4too. The current supported LZ4 formats inChunkCodecLibLz4are Numcodecs (for Zarr), HDF5, Frame, and Block, but it looks like ASDF has its own slightly different format.IIUC this is defined in https://github.com/asdf-format/asdf/blob/14c40c83fc3cca7285b59be3976bd29e2ba35543/asdf/_compression.py#L65
I also found https://github.com/asdf-format/asdf-compression/blob/25ca4e8b1c42ed2c329ff3694601ec4dbf504f14/asdf_compression/lz4f/compressor.py#L5
Based on this, it looks like Lz4 frame should be using "lz4f" as the 4-byte key not "lz4\0" to be readable by python (but I haven't set up tests for this yet).