Service type: collection

[alexcos20]

A publisher need to be able to create a "collection", which is a reference to a group of datasets. Immediate use-case: albums/playlists on Waves, although some other use-cases are going to emerge as well

Proposal:

Add a new type of service, called "collection".

The "files" key/value can be used in two ways:

• store an unecrypted array of dids. This makes the list public, and dApps can generate the list on the fly. In this case, datatoken is not needed, so we can store it as 0x00000 Example:

  {
  "services": [
  {
    "id": "1",
    "type": "collection",
    "files": ["did:op:123" , "did:op:125"],
    "datatokenAddress": "0x0000000000000000000000000000000000000000",
    "serviceEndpoint": "https://v4.provider.goerli.oceanprotocol.com",
    "timeout": 0
  }
  ]
  }

• store an encrypted array of dids as string. This makes the list private, and require payment of datatoken to get it. When using the usual download flow, provider will decrypt the array and send it back to the consumer Example:

  {
  "services": [
  {
    "id": "1",
    "type": "collection",
    "files": "2433423432432423",
    "datatokenAddress": "0x000000000000000000000000000012345656",
    "serviceEndpoint": "https://v4.provider.goerli.oceanprotocol.com",
    "timeout": 0
  }
  ]
  }

Why not use tags when publishing datasets (ie: mark them as being part of collection X) ? Because somebody else can use the same tag, breaking the list