[Bug]: Grand child doesn't have subject counts

[gogonzo]

What happened?

data <- teal_data()
data <- within(data, {
  parent <- data.frame(id = 1:10, a = runif(10), b = runif(10))
  child <- data.frame(id = 1:100, parent_id = rep(1:10, each = 10), c = runif(100), d = runif(100))
  grand_child <- data.frame(id = 1:1000, child_id = rep(1:100, each = 10), e = runif(1000), f = runif(1000))
})
join_keys(data) <- join_keys(
  join_key("child", "parent", keys = c("parent_id" = "id")),
  join_key("grand_child", "child", keys = c("child_id" = "id"))
)


app <- init(
  data = data,
  modules = example_module()
)
shinyApp(app$ui, app$server)

[gogonzo]

I'm not sure if this could be done before. If you have a grand-child dataset then which information should be taken to determine subjects? Join-keys are determining id of it's parent which is still a child. Information about the subjects in the parent are not obtainable at this moment, this should be done recursively. So I'm questioning "subjects" in general - what does it mean in general. If we have a structure with

• orders
• order_items
• items

Then items have no such thing like "subjects". In ADaM three levels could look like this

• parent (ADSL)
• analysis (ADTTE)
• col_details (PARAM)

We need to find the way to properly (universaly) describe parent-child relationship there. Theoretically it is rather:

• observations
• parent observations