Floating point example is the wrong way around

[dan-robertson]

The example in the specification is for the float 29.9510, whose bits are 0x41ef9ba6. Jackson serializes this as [0x28] 0x04 0x0f 0x3e 0x37 0x26 (the 0x28 is the tag for a 32-bit float). The specification gives these bytes in the opposite order.

Here is the code of the jackson serializer for reference: f64 f32.

For completeness, the double precision float for -29.9510 has binary representation 0xc03df374bc6a7efa and is thus serialized as [0x29] 0x01 0x40 0x1e 0x7c 0x6e 0x4b 0x63 0x29 0x7d 0x7a.

[cowtowncoder]

Thank you for reporting this @dan-robertson . Will verify and fix as necessary.

[cowtowncoder]

Adding Smile codec tests first: FasterXML/jackson-dataformats-binary#609 -- confirms suggested encoding, different from spec explanation.

[cowtowncoder]

Proposed fix: #29.