AppSync loading incorrect resolver for field - amazon-web-services

Question

AppSync appears to be loading the incorrect resolver template for some fields of nested objects. Also, it appears to only happen when the nested object has a field with the same name as a field on the parent object.
I've included an example below because I think that might be the best way to explain the issue. As you can see, the id fields for the nested objects are not being resolved correctly.
Each type, Task, User, List, and Tag, have a resolver for their id field because the data for each has a prefix on the id field. For example, Task.id has a resolver that returns $context.source.task_id and User.id has a resolver that returns $context.source.user_id. Same for List and Tag.
What appears to be happening is AppSync is loading the id resolver template for the parent type. You can see that this is the case for task.owner.id, where owner is a User but the id gets resolved as "$context.source.task_id". Same for task.list.id where list is a List. Again we can see this for task.tags[0].owner.id. owner is once again a User except this time the parent is a Tag so task.tags[0].owner.id is resolved as "$context.source.tag_id". These three example show that the problem is not with a particular type since User and List are behaving the same when they are nested in a Task. Also, we can see that the issue is not with Task since User is behaving similarly when nested in a Tag. Lastly, we can see that task.tags[1].owner.id actually behaves correctly. This indicates that the issue only presents itself on first execution.
At this point I strongly suspect this is a bug with AppSync however, I'm not 100% on that. Has anyone else experienced this issue? Am I doing something terribly wrong?
Example
Query
{
task(id: "task-123") {
id,
title,
owner {
id,
username,
},
list {
id,
name,
},
tags {
id,
name,
owner {
id,
username,
},
},
},
}
Result
{
"data": {
"task": {
"id": "task-123",
"title": "First Task",
"owner": {
"id": "$context.source.task_id",
"username": "tom"
},
"list": {
"id": "$context.source.task_id",
"name": "Inbox"
},
"tags": [
{
"id": "tag-123",
"name": "one",
"owner": {
"id": "$context.source.tag_id",
"username": "tom"
}
},
{
"id": "tag-234",
"name": "two",
"owner": {
"id": "user-123",
"username": "tom"
}
}
]
}
}
}
Task Schema
type Task {
id: ID!
title: String!
owner: User!
list: List
tags: [Tag]
}
User Schema
type User {
id: ID!
username: String!
}
List Schema
type List {
id: ID!
name: String!
}
Tag Schema
type Tag {
id: ID!
name: String!
owner: User!
}
Task Data
{
task_id: "task-123",
title: "First Task",
owner_id: "user-123",
list_id: "list-123",
tags: [
"tag-123",
"tag-234"
]
}
User Data
{
user_id: "user-123",
username: "tom"
}
List Data
{
list_id: "list-123",
name: "Inbox"
}
Tag Data
{
tag_id: "tag-123",
name: "one",
owner_id: "user-123"
}
{
tag_id: "tag-234",
name: "two",
owner_id: "user-123"
}
Example id resolver (User)
Request Mapping Template
{
"version": "2017-02-28",
"payload": "$context.source.user_id"
}
Response Mapping Template
$util.toJson($context.result)
The id resolvers for the other types are very similar
It may also be worth noting that I created different None Data Sources for each type, Task, User, List, and Tag. The id resolver for each type is using their respective None Data Source.