E_INPUT_001 — Unknown compression type¶
A compression name passed to a recorder / writer is not one FLOX supports. Accepted values:
| Value | Effect |
|---|---|
"none" / "" |
No compression (default) |
"lz4" |
LZ4 frame compression |
How to fix¶
The name is a constructor argument on the two Python types that own a
binary-log writer — DataWriter (offline writing) and
BinaryLogRecorderHook (recording a live engine). Pass one of the two
accepted spellings:
Node.js never raises this code. new flox.DataWriter(dir, maxSegmentMb,
exchangeId) takes no compression argument and always writes
uncompressed; new flox.BinaryLogRecorderHook(dir, maxSegmentMb,
exchangeId, compression) does take one, but rejects a bad value with a
plain TypeError (compression must be "none" or "lz4").
Common causes¶
- Typo:
"lz4hc"/"zstd"/"gzip"— onlylz4is currently built in. Adding more compressors is a separate task. "lz4"is accepted by the argument check regardless of how the engine was built. If it was configured with-DFLOX_ENABLE_LZ4=OFF(the default is ON, with a vendored fallback), the compressor is compiled out and silently produces zero bytes instead of raising here.- The same code is also raised by
MergedTapeReaderfor an empty tape list, which is an unrelated input error with its own message ("paths list is empty").