A compact and simple object notation. µ[mju:] stands for "micro".
File extension: .mu
MIME type: application/muon
Endianness: little-endian
Signature/Magic: optional, 8F B5 30 31 ["�µ01"] @ 0x0
Muon has some interesting properties (see presentation and docs):
- Every null-terminated
UTF8
string is at the same time a validmuon
object - Gaps in the
UTF8
encoding space (code units
) are used to encode things like[
]
{
}
etc. - Muon is self-describing and schemaless, just like
JSON
(unlikeProtobuf
andFlatBuffers
) - Compact (~10..50% smaller than
JSON
). On par or outperformsCBOR
,MsgPack
,UBJSON
- Unlimited size of objects and values
- Data is ready to be used in-place without pre-processing
- Supports raw binary data (values and
TypedArrays
) - Can optionally contain count of elements (and size in bytes) of all structures for efficient document processing, similar to
BSON
Future goals:
- Strict specification (little or no room for implementation-specific behavior / vendor-specific extensions)
python3 muon_py/json2mu.py ./data/AirlineDelays.min.json ./AirlineDelays.mu
python3 muon_py/mu2json.py ./AirlineDelays.mu > ./AirlineDelays.json
Run benchmarks:
python3 muon_py/extra/json-analyze.py ./data/*.json ./data/small/*.json
python3 muon_py/mu-benchmark.py ./data/*.json ./data/small/*.json
See more in documentation.
Disclaimer: the notation is still Work In Progress. If you have any ideas or comments, please feel free to post them here.