It would be really great if Aeson was able to parse (and re-serialize) JSON generated with the Python json module, which has some extra (non-spec) features with regard to numbers:
import json
import math
print(
json.dumps(
{
"foo": 99,
"bar": math.nan,
"baz": -math.inf
}
)
)
which yields
{"foo": 99, "bar": NaN, "baz": -Infinity}
I think NaN, Infinity and -Infinity are the only extra tokens that occur.
For me personally this comes up when working with JSON generated by data-scientists, where Python is the main language and these values seem to occur often.