Fix encoding of strings of length 256 (or 2^16 or 2^32)

[plexus]

I hope this is the right place to submit a patch for the CBOR package found on MELPA.

There is currently an issue in cbor.el where strings with a length of exactly 256 are encoded as if they had a length of 0. This causes the UTF-8 bytes of the string to be interpreted as CBOR bytes, leading to a garbled message.

The greatest number that can be encoded in a single byte is 255, for a string of length 256 we need a second byte to encode the length.

• cbor playground 256 byte string (expected)
• cbor playground of current cbor.el output (actual)

[Titan-C]

Thank you for noticing this. It is actually a serious bug on all integer encoding. I added tests at each integer length type.

[plexus]

Thanks for the fix!