I vaguely remembered that we had the preferred encoding for CRIs (both for general use and in CoRAL) as a byte string that encodes a CBOR stream. Current text and examples always talk of CBOR arrays.
Was that something we consciously decided on? I remember that it did play a role in the discussion of devices' ability to skip over CRIs they don't need, and to a smaller extent concerning easy comparability. (It doesn't help that I found nothing pertaining to streams in the earlier notes).
(By the way, me thinking of encoded CRIs as byte strings may explain some of last Friday's confusion when I suggested CURIE style templating based on the byte representation -- that only makes sense (but is still ugly) when done on byte strings).
Not trying to argue for any version here yet, just trying to understand where I diverged.
I vaguely remembered that we had the preferred encoding for CRIs (both for general use and in CoRAL) as a byte string that encodes a CBOR stream. Current text and examples always talk of CBOR arrays.
Was that something we consciously decided on? I remember that it did play a role in the discussion of devices' ability to skip over CRIs they don't need, and to a smaller extent concerning easy comparability. (It doesn't help that I found nothing pertaining to streams in the earlier notes).
(By the way, me thinking of encoded CRIs as byte strings may explain some of last Friday's confusion when I suggested CURIE style templating based on the byte representation -- that only makes sense (but is still ugly) when done on byte strings).
Not trying to argue for any version here yet, just trying to understand where I diverged.