Trade-offs between cbor and cbor2

[cheqianh]

Hello!

I am new to cbor and recently looking into cbor and cbor2, and having a question about what benefits cbor2 brings that cbor does not have?

I experimented on their dumps/loads APIs with some sample data (E.g. a 64 bits integer, an alphabet string). In short, cbor uses less memory and usually runs faster.

My guess is that cbor2 implements RFC 8949 while cbor is still targeting RFC 7049? Is there any benefits cbor2 brings that cbor does not have? Please feel free to correct me or point out a documentation that I've missed.

[cheqianh]

One example,

Benchmark:

import time
import cbor
import cbor2

o = 9233720363654371807

start = time.time()
for i in range(1000000):
    cbor.dumps(o)  
print(time.time() - start)

Results cbor	cbor2
0.85502s	2.9855s

[agronholm]

cbor has not been maintained since 2016. It also segfaults on self referential structures, and does not implement any tag support. In short, it cuts corners which, I believe, is why it's faster.

[cheqianh]

Alright, thanks for the response!