msg_post_iss_object.md

post_iss_object: store an immutable object

A post_iss_object request asks to store the data for an immutable object; CRADLE returns a reference id for the data to the client, and caches that id. It does not cache the data itself: that would require an immutable id, which Thinknode does not return in its response. As a consequence, another identical post_iss_object request would be served from the cache, but an iss_object request specifying the same reference id would still cause a cache miss, and would be resolved by Thinknode first.

Assuming that a recently posted object will soon be needed again, CRADLE could retrieve the immutable id from Thinknode in its processing the post_iss_object request; then it would have all needed information to cache the object data too.

Example message:

request_id: 9f0eb0a45fcf438db785e30b1ebb57be
content:
  post_iss_object:
    context_id: 0123456789abcdef0123456789abcdef
    encoding: msgpack
    object: (MessagePack data)
    schema: string

The fields are:

• request_id: identifies this request; set by the client
• context_id: see here
• encoding: will usually be msgpack
• object: data for the immutable object, encoded as per encoding
• schema: describes the structure of the object's data

Example response:

request_id: 9f0eb0a45fcf438db785e30b1ebb57be
content:
  post_iss_object_response:
    object_id: 5ac4e6f700c0fb7a4baeadcf659fa217

The fields are:

• request_id: copied from the request message
• object_id: reference id for the stored object

Cache key

The cache key is a hash depending on:

• The key type (post_iss_object)
• The API URL (e.g. https://mgh.thinknode.io/api/v1.0)
• The context id
• The schema
• The account name if the schema includes named types
• The object data

Cache hit

If CRADLE already received an identical post_iss_object request, and did not yet evict the corresponding object id from its cache, the interaction looks like

Cache miss

If the cache does not contain the object id for the immutable object, then CRADLE will post the object to Thinknode, and cache the object id that Thinknode returns: