Re: [Cbor] Invalid Tag Values

[Carsten Bormann <[e-mail]>]

Hi Kio,

> I'm having trouble understanding the rationale of intentionally invalid tags as documented here[1]
> 
> Under what circumstances would an encoder produce these tags

Never.  They are explicitly reserved so that an implementation has a selection of tag numbers that it can use internally, such as in order to mark the absence of a tag.

> and what should a decoder do when encountering them?

Mark the data item as invalid.
I don’t think a more rigorous response (rejecting the data item the same way that a non-well-formed input would be) would be out of line, as the tag numbers are explicitly marked as “not to occur in interchange”.
(RFC 8949 just doesn’t have a category between non-well-formed and invalid that we could use here.)

> Since I'm working on that part of the decoder[2] anyway I thought I'd better check.

Thank you a lot for doing this!

If you have suggestions how the text in -notable-tags could be made even more explicit about this, please tell us.

Grüße, Carsten