pymongo's bson takes too much time to do simple json encondig!

[Aliane Abdelouahab]

Doing benchmarks with this:

tornado     serialize: 1.225  deserialize: 2.117  total: 3.342

ujson       serialize: 0.935  deserialize: 0.413  total: 1.348

simplejson  serialize: 1.944  deserialize: 0.794  total: 2.738

mongodb bson serialize: 17.280  deserialize: 3.438  total: 20.718

stdlib json serialize: 1.079  deserialize: 2.053  total: 3.132

If the json is too big, the operation will never end!

[Bernie Hackett]

Yes, json_util is not fast. It uses python's json module to convert JSON strings to python dictionaries, then iterates over those dictionaries to turn the mongo extended JSON format into PyMongo specific objects. This is slow. If you don't need your mongo extended JSON in PyMongo format, just use python's json module or simplejson.

[Bernie Hackett]

In fact, your benchmark doesn't need json_util at all. All json_util does in your case is add a bunch of overhead looking for mongo extended JSON types.

[Aliane Abdelouahab]

ah, got it, thank you ^_^