[Cbor] Re: cbor.me issue with indefinite form round-trips

[Carsten Bormann <[e-mail]>]

Hi Brian,

> On 13. Aug 2024, at 20:39, Sipos, Brian J. <[e-mail]> wrote:
> 
> All,
> I can’t find any specific issue tracker associated with the https://cbor.me site, so I’m posting in this list because I know some maintainers also read this list.

They do :-)

> Please let me know if there is a specific issue tracker that I should be submitting to instead.

The site weakly points to the cbor.io <http://cbor.io/> issue tracker; maybe the weakness of that reference should be fixed at some point.

>  I recently ran into an issue where the decoder does not properly indicate EDN for indefinite length containers even though the encoder does properly handle them. This means the tool will not fully round-trip indefinite form of arrays and maps which makes dealing with some structures more complicated than it seems like it should be.
>  For example, these inputs will encode properly but then going back to decoding will remove the indefinite-length indicating underscore:
> [_1,2]
> {_1:2}
>  I think this tool is great to have and would like to see this little discrepancy fixed.

The site pretty much still runs on its 2013 code base (which will turn eleven in a couple of days…).
The plan was to wait for draft-ietf-cbor-edn-literals to be approved and then to replace the current guts by ones based on the ABNF in that document.
(Those guts are already available in part in the edn-abnf gem so we have a prototype to work with, but integration into cbor.me <http://cbor.me/> is an outstanding step.)
I had planned to do this earlier, and I was now hoping I could finish this in August while many other things are in a lull in Europe.
I now have no idea when we can finish the current phase of (insert your euphemism here).

But, yes, the plan is to get there.

Grüße, Carsten