[yang-tooling] Re: SID file issues

[Vojtech Vilimek <[e-mail]>]

Hi Laurent,

On 2/17/26 09:20, Laurent Toutain wrote:
> Hi,
> 
> There is something I don't understand in the discussion about RPCs and 
> input/output. For me, SIDs are not here to represent a YANG module, but 
> rather a way to serialize information in a more compact way than JSON.
> 

Yes, SIDs are generally used for byte-compact representation of 
YANG-modeled data. SIDs are equivalent to the (absolute) node name.

> SIDs are just a way to point to leaves or sub-trees. If there is no RPC, 
> why impose SIDs for input and output? And if there are several RPCs, 
> several SIDs will be needed.

You can only use 'input' and 'output' statements as child nodes of 'rpc' 
or 'action'. As Andy pointed out, if you create e.g. empty 'rpc' it will 
contain 'input' and 'output' nodes implicitly. The problem with RFC 9595 
example .sid file is that the 'ietf-system' contains rpc with implicit 
'output' or 'input' nodes which are not assigned SIDs.

> 
> For instance, in the SID allocation I'm doing for SCHC, I don't allocate 
> SIDs for typedefs, since they do not appear in any serialization.
> 

The YANG typedef is similar to C typedef. It is only a shortcut for 
specific restricted type, a bundle of YANG statements reusable together 
(and identified by simple name).

> Am I wrong? What would be the consequences if some SIDs are not defined?
> 
> Thank you for your clarifications.
> 

Hope this helps!


Best regards,

> 
> Laurent
> 

Vojtech Vilimek
CZ.NIC z.s.p.o.