pysimdjson VS orjson

Its magic function wrapping comes at a cost, trading ease of use for runtime performance. When you have a single C++ function to call that will run for a "long" time, pybind all the way. But pysimdjson tends to call a single function very quickly, and the overhead of a single function call is orders of magnitude slower than with cython when being explit with types and signatures. Wrap a class in pybind11 and cython and compare the stack trace between the two, and the difference is startling.

Ex: https://github.com/TkTech/pysimdjson/issues/73

simdjson

You are aware of simdjson being available in python if you really need some json crunching, albeit json module in Python is implemented in C itself, so I don't think understand why do you think Python is slow there?

json: 113.79130696877837 ms

While `orjson`, is faster than `ujson`/`json` here, it's only ~6% faster (in this benchmark). `simdjson` and `msgspec` (my library, see https://jcristharif.com/msgspec/) are much faster due to them avoiding creating PyObjects for fields that are never used.

If spyql's query engine can determine the fields it will access statically before processing, you might find using `msgspec` for JSON gives a nice speedup (it'll also type check the JSON if you know the type of each field). If this information isn't known though, you may find using `pysimdjson` (https://pysimdjson.tkte.ch/) gives an easy speed boost, as it should be more of a drop-in for `orjson`.

I don't think JSON is really the problem - parsing 10MB of JSON is not so slow. For example, using Python's json.load takes about 800ms for a 47MB file on my system, using something like simdjson cuts that down to ~70ms.

orjson is the fastest JSON library available for python. It natively manages dataclass objects, datetime, numpy and UUID objects.

As we began venturing down that road, a few things emerged that would constitute significant changes to some of the core parts of Litestar, but there were two things in particular that started a chain reaction of changes by opening up further possibilities: The new DTOs and our switch from orjson to msgspec.

In version 1.45.0, we introduced msgspec as our serialization backend, replacing orjson. This had some immediate performance benefits, but that's not the main reason we made the switch.

Doesn't orjson do that already?

orjson

For most cases, you would want to go with python’s standard json library which removes dependencies on other libraries. On other hand you could try out ujsonwhich is simple replacement for python’s json library. If you want more speed and also want dataclass, datetime, numpy, and UUID instances and you are ready to deal with more complex code, then you can try your hands on orjson

Try https://github.com/ijl/orjson. Not exactly sure if that can help.