Re: [Cbor] draft-ietf-cbor-network-addresses: tag validity; deterministic encoding

[Brian E Carpenter <[e-mail]>]

On 15-Oct-21 17:50, Carsten Bormann wrote:
> On 15. Oct 2021, at 06:35, Brian E Carpenter <[e-mail]> wrote:
>>
>> So, my ugly code is in the functions build5254() and detag5254() in https://github.com/becarpenter/graspy/blob/master/pfxm4.py
> 
> Nice.
> 
> After line 122, there needs to be the Python equivalent of a
> 
> prefix.sub!(/\x00*\z/, ‘’)


I took an executive decision to throw an exception for covert bits, rather than silently zero them, and in fact changed all my error returns to ValueError exceptions. I believe that's more Pythonesque.
  
> (Why not mix floating point calculations and a regex in the same five lines of code :-)

Because it makes my head hurt, that's why not.
> 
> Before line 146, you’d need to check whether the size really is 
`asize`.

Yes, or that the type really is 'bytes' since Python won't check that for 
me.

> I think the while loop in 154 needs to be before 151.

I don't think it matters. Either way, the zero bytes get concatenated.
  
> If v[0] == 128 and prefix is 16 bytes, what does prefix[v[0]//8] yield in Python?

IndexError: index out of range

Good catch. In my app, that is a case that would never happen.

I've fixed my code, but will let my test app run for a few hours before I 
post the update to github.

    Brian