Bump fxamacker/cbor to 2.x

[kostko]

The recently made some minor API changes which should be easy to integrate. The release also brings performance and memory use improvements.

Next release (v2.1) will also add support for CBOR tags which may be useful.

[fxamacker]

Hi @kostko! I released v2.2 with a bugfix and additional security-related settings: MaxNestedLevels, MaxArrayElements, MaxKeyPairs, IndefLength (allow/forbid), and TagsMd (allow/forbid).

v2.2 also added CBOR byte string <--> Go byte array was added (slice was already supported.)

v2.2 is slightly faster than v2.1, but biggest gains came from v2.1 -- struct using keyasint tag (like COSE and CWT) is about 24-28% faster and 53-61% fewer allocs compared to v1.x and 2.0.

v2.1 added CBOR tags and option to detect duplicate map keys.

Please let me know if I can help!

[kostko]

Cool, we'll bump to v2 soon. I just filed fxamacker/cbor#178, @fxamacker what do you think about adding an option for stricter decoding?

[fxamacker]

Option(s) for stricter decoding is coming but I need to be cautious because "strict mode" and even the word "strict" is gone in the draft 7049bis.

I want to provide preset options for deterministic encoding/decoding that round trips without violating RFC 7049 and newer 7049bis. This would require stricter decoding.

BTW, 7049bis is moving away from using the word "canonical". It prefers bytewise-lexicographic sort instead of length-first sort (RFC 7049 canonical). I heard 7049bis might get approved "later this year".

https://tools.ietf.org/id/draft-ietf-cbor-7049bis-12.html