[Bug]: Forced merge precludes using modules without joinin keys

[chlebowa]

What happened?

Even in modules that treat datasets independently, teal.transform attempts to join them and fails if no keys are provided.

modified example app

library(teal)
library(teal.modules.general)

data <- teal_data()
data <- within(data, {
  require(nestcolor)
  USArrests <- USArrests
  iris <- iris
})

app <- init(
  data = data,
  modules = modules(
    tm_a_pca(
      "PCA",
      dat = list(
        data_extract_spec(
          dataname = "USArrests",
          select = select_spec(
            choices = variable_choices(
              data = data[["USArrests"]], c("Murder", "Assault", "UrbanPop", "Rape")
            ),
            selected = c("Murder", "Assault"),
            multiple = TRUE
          ),
          filter = NULL
        ),
        data_extract_spec(
          dataname = "iris",
          select = select_spec(
            choices = variable_choices(
              data = data[["iris"]], function(data) names(Filter(is.numeric, data))
            ),
            selected = NULL,
            multiple = TRUE
          ),
          filter = NULL
        )
      )
    )
  )
)
if (interactive()) {
  shinyApp(app$ui, app$server)
}

sessionInfo()

> packageVersion("teal.transform")
[1] ‘0.5.0’
> packageVersion("teal.modules.general")
[1] ‘0.3.0’

Relevant log output

Code of Conduct

• I agree to follow this project's Code of Conduct.

Contribution Guidelines

• I agree to follow this project's Contribution Guidelines.

Security Policy

• I agree to follow this project's Security Policy.

[llrs-roche]

This is confusing users as this related reported shows: insightsengineering/teal#1512

[chlebowa]

I don't suppose this can be fixed without an overhaul of teal.transform so I don't expect a fix but please keep it in mind during the ongoing refactor.

[llrs-roche]

Totally, that's why I was aware of it 👍 . So far on the refactor if there are no keys I let the merging function detect them. But I might miss some cases.

[chlebowa]

"Detect" implies that there are always some keys to be detected. What if there are two completely independent datasets but both are passed into the module and intended to be analyzed separately? I know the proper way would be to create two module instances with one dataset per, but...