$ref do not work with bindings in channel in v3

[derberg]

issue might be with any bindings, not just channel, I did not check

spec says that in channels I can also have $ref

which makes sense as I want to have some reusable stuff in components. channelBindings

I would expect below example to work (studio direct link - btw I love new share button in studio:

asyncapi: '3.0.0'
info:
  title: MySlackBot API
  version: '1.0.0'
  description:  |
    The MySlackBot App manages popular messages in a workspace by monitoring message reaction data from Slack's Event API.
    
servers:
  ws:
    host: wss://wss-primary.slack.com/
    protocol: wss
    description: "Websocket URL generated to communicate with Slack"
channels:
  root:
    address: /
    messages:
      reactionListener: {}
      helloListener: {}
    bindings:
      $ref: "#/components/channelBindings/connectionQuery"
      
operations:
  helloListenerOperation:
    action: receive
    channel:
      $ref: "#/channels/root"
    messages: 
      - $ref: "#/channels/root/messages/helloListener"
  reactionListenerOperation:
    action: receive
    channel: 
      $ref: "#/channels/root"
    messages: 
      - $ref: "#/channels/root/messages/reactionListener"

components:
  channelBindings:
    connectionQuery:
      ws:
        query: {}

but this says is not allowed

bindings:
      $ref: "#/components/channelBindings/connectionQuery"

After so many years, I still do not feel like JSON Schema expert so can you confirm, but I think in here https://github.com/asyncapi/spec-json-schemas/blob/next-major-spec/definitions/3.0.0/channel.json#L65 we should instead have something like

"bindings": {
      "oneOf": [
        {
          "$ref": "http://asyncapi.com/definitions/3.0.0/Reference.json"
        },
        {
          "$ref": "http://asyncapi.com/definitions/3.0.0/channelBindingsObject.json"
        }
      ]

thoughts?

[smoya]

I'm from my phone so checking code is a pain, but i noticed the same happens with others like operation bindings.

Regarding your suggested solution, seems ok to me.

[fmvilas]

Yeah, I think it does make sense indeed.

[derberg]

@fmvilas @smoya PR opened: #442

[derberg]

solved in next-major-spec