This is an attempt to fix#360. The algorithm produces
decimal representations which are guaranteed to roundtrip
and in ~99.8% actually produces the shortest possible
representation. So this is a nice compromise between using
a precision of digits10 and max_digits10.
Note 1:
The implementation only works for IEEE single/double precision
numbers. So the old implementation is kept for compatibility
with non-IEEE implementations and 'long double'.
Note 2:
If number_float_t is 'float', not all serialized numbers can
be recovered using strtod (strtof works, though). (There is
exactly one such number and the result is off by 1 ulp.)
This can be avoided by changing the implementation (the fix
is trivial), but then the resulting decimal numbers are not
exactly short.
We had a lot of issues with failing roundtrips (i.e., parse errors from serializations) in case string were stored in the library that were not UTF-8 encoded. This PR adds an exception in this case.
The example in the documentation on how to "force" an array of arrays was wrong since the first release. Fixed the documentation and added checks for the README unit tests.
Hey, thanks for this great library which I've used in many occasions now. I know it's not much, but I wanted to contribute at least a tiny bit back to you by this PR :-) Keep up the good work!
