Reproducible metadata attribute

[gogonzo]

Feature description

If TealDataset's metadata was an attribute of the data - then:

1. Would be reproducible - attr(<dataset>, "metadata") <- <metadata call>.
2. Would be easy to obtain attr(<raw data>, "metadata")
3. Wouldn't be needed to specify this as a argument in FilteredData as attr will follow the (raw) data. Only one operation which removes attribute from data.frame is removing a columns - dplyr calls preserve attributes

a <- data.frame(a = 1, b = 2)
attr(a, "metadata") <- "test"
a$c <- 3
attr(a, "metadata")
# "test"

a |>
  dplyr::filter(a == 0) |>
  dplyr::select(-c) |>
  dplyr::left_join(a, "a") |>
  attr("metadata")
# [1] "test"

a[1:2] |> attr("metadata")
# NULL

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.

[gogonzo]

@kumamiao @nikolas-burkoff What do you think? Currently metadata is put as a slot in TealDataset and FilteredDataset. Maybe it would be better if we keep it in raw data?

[nikolas-burkoff]

I'm happy with this, we'll also need to make changes in internal connector packages of course

[kumamiao]

Sounds good, no strong preferences from me as long as app developers can 1) grab the metadata within a module 2) use the front page module to check the metadata

[gogonzo]

sounds like plan - metadata could be obtained in the module by:

# in case of datasets <FilteredData>
attr(datasets$get_data(<dataname>), "metadata")

# in case of data <list of reactive data>
attr(data[[<dataname>]](), "metadata")

[gogonzo]

implemented with new teal_data