Encoding

[boucadair]

I really like that there are only very few metadata in the document.

I would suggest that the encoding part is separated from the actual definition of the metadata (moved to an appendix). The 3-bit proposal can be listed as an implementation example, but I don't think that actually matters.

[sridharangirish91]

I really like that there are only very few metadata in the document.

I would suggest that the encoding part is separated from the actual definition of the metadata (moved to an appendix). The 3-bit proposal can be listed as an implementation example, but I don't think that actually matters.

Thanks Med.

Given the document focuses on standardizing (or using consistent) metadata across the board as well as proposing explicit approach in sharing granular information, wouldn't it be more streamlined and necessary for it to be the way it is where we read "the parameter -> what it means -> why its needed -> how it looks" and then both interpretations (simple as well as granular) of real-world examples (as many scenarios as we could cover without overdoing all possible applications)?

[boucadair]

I agree that this would be ideal.

From where I sit, I think that standardization the definition and intended usage is by itself a great contribution. The "how it looks" part depends on the signaling protocol. For example, there are encoding constraints that are imposed by the sig (QUIC, MASQUE, TCP, SCOKS, Convert, etc.).

Also, consider the 3-bit info in the draft. This is an artificial limitation at this stage because more bits may be needed if a new information is defined.

We may consider using an abstract data modelling (e.g., CDDL (rfc8610), YANG Abstract Data Structure (RFC8791)) to ease the mapping with specific sig protocol encoding.

[danwing]

CDDL is a good idea.