Support for new format of CCA top level token to match CMW changes

[SimonFrost-Arm]

Support for updated (breaking) change in CCA top level token made to conform with updated CMW definition.
Change to spec can be found in RMM 1.1-alp14.

[yogeshbdeshpande]

The changes LGTM, just one question:

I noticed the RMM spec referred in the README.md still uses
cca-token = #6.399(cca-token-collection) ;

Is there a plan to change the spec? as currently the spec seems out of sync with the proposed changes.

[SimonFrost-Arm]

Is there a plan to change the spec? as currently the spec seems out of sync with the proposed changes.
The change will be in the upcoming 1.1alp14

[yogeshbdeshpande]

LGTM!

[setrofim]

Please squash the commits, add Signed-off-by footer, and update the commit message to align with Conventional Commits.

Apart from that, LGTM.

[thomas-fossati]

I think we can be a bit more conservative on the decoding side.
The fact that the CBOR tags are different allows us to support the EAT collection in parallel with the CMW collection pretty straightforwardly.

Something like:

 var (
        tags = []encoding.CBORTagEntry{
                {Type: CBORCollection{}, Tag: 399},
                {Type: CMWCollection{}, Tag: 903},
        }

and then

type CMWCollection struct {
	PlatformTokenColl CMWCollectionEntry `cbor:"44234,keyasint"`
	RealmTokenColl    CMWCollectionEntry `cbor:"44241,keyasint"`

Or do that manually, by first decoding the top-level tag, and then dispatching to the right collection decoder:

  var t cbor.RawTag

  if err = t.UnmarshalCBOR(b); err != nil {
          return fmt.Errorf("unmarshal top-level CBOR Tag: %w", err)
  }
  
  switch t.Number {
  case 903: // do new
  case 399: // do legacy
  }

This would give senders the ability to align with the new format at their own pace.

[SimonFrost-Arm]

Latest commit adds support for reading old (399) tokens as well as the full CMW tokens

[yogeshbdeshpande]

Latest commit adds support for reading old (399) tokens as well as the full CMW tokens

Apologies for a late reply, but wanted to check, is this complexity really required, at this stage supporting two types of collection ?

The products using token type are not yet out in the field, so if we can unilaterally change that is a better fit for now ?
Just my two cents!

[thomas-fossati]

Latest commit adds support for reading old (399) tokens as well as the full CMW tokens

Apologies for a late reply, but wanted to check, is this complexity really required, at this stage supporting two types of collection ?

Yes, there is running code, in testing labs and in CIs, that’s better not brutally break :-)

[SimonFrost-Arm]

Merged separately to cleanup messy commit history