Re: [core] draft-bierman-core-yid-00.txt questions

[peter van der Stok <[e-mail]>]

Alexander Pelov schreef op 2016-08-25 18:14:
> Hi Andy,
> 
>> Le 25 août 2016 à 18:02, Andy Bierman <[e-mail]> a
>> écrit :
>> 
>> On Thu, Aug 25, 2016 at 8:54 AM, Michel Veillette
>> <[e-mail]> wrote:
>> 
>>> Hi Andy, Hi David
>>> 
>>> There is no requirement to have multiple tools or algorithms
>>> producing the same result.
>> 
>> Then it isn't auto-numbering.
>> It is just an offline tool to assign numbers to YANG statements.
> 
> It is a way to assign numbers… automatically. Auto-numbering.
> 
>> So there is no need to standardize SID at all.
>> There is just a list of [id, path] mappings in a registry entry,
>> which is what the YID draft proposes.  (centralized or distributed
>> list of mappings).
> 
> I think we agree here.

I am also convinced there is much overlap in the SID and YID drafts
The conclusions by Alexander are a bit fast for me.

What happened to the discussion about module number reservation against 
SID range reservation?
I don't think the strains on IANA (enormous or not) are an issue, but 
the strains on the module writers and what rules they are willing to 
follow.
For me it is the process of establishing identifiers, with a discussion 
about freedom of choice to companies, and scopes of inter-operability 
(world-wide, SDO wide,--), that is central here.

The YID draft proposes a subset of what can be
> achieved by the SID draft. The YID draft proposes a more complex
> process, which does not ensure interoperability, and which puts
> enormous strain on IANA, with no description of how a more generic
> system will work.
> 
> Everything you can do with YID you can do with SID. Not everything you
> can do with SID can be achieved with YID.
> SID is simpler.
> SID cares for ID exhaustion.
> SID has running code.
>