Implement distinct()

[msberends]

Suggestion:

distinct <- function(.data, ..., .keep_all = FALSE) {
  check_is_dataframe(.data)
  UseMethod("distinct")
}

distinct.default <- function(.data, ..., .keep_all = FALSE) {
  names <- rownames(.data)
  rownames(.data) <- NULL
  if (length(deparse_dots(...)) == 0) {
    selected <- .data
  } else {
    selected <- select(.data, ...)
  }
  rows <- as.integer(rownames(unique(selected)))
  if (isTRUE(.keep_all)) {
    res <- .data[rows, , drop = FALSE]
  } else {
    res <- selected[rows, , drop = FALSE]
  }
  rownames(res) <- names[rows]
  res
}

distinct.grouped_data <- function(.data, ..., .keep_all = FALSE) {
  apply_grouped_function(.data, "distinct", ..., .keep_all = .keep_all)
}

I cannot get the grouped version to work to also include the group variables. It now only returns the distinct variable if set...

---

Another idea - You should mention on your README that this package is a great, great idea for package developers that do not want to be dependent on dplyr (as it changes too often for sustainable pkg development), but do want to code using dplyr methods. For those users you could also create an extra raw syntax file with all your functions without roxygen parts (remove all lines starting with #') and your name on it, so they can copy it to their package.

[nathaneastwood]

Thanks for the code 🙂 I’ll take a look when I get a chance.

Another idea - You should mention on your README that this package is a great, great idea for package developers that do not want to be dependent on dplyr (as it changes too often for sustainable pkg development), but do want to code using dplyr methods.

Yes and no. It’s good because there are much fewer dependencies. But the aim of poorman is to replicate dplyr as closely as possible. So any changes seen in dplyr are likely to make their way into poorman at some point.

For those users you could also create an extra raw syntax file with all your functions without roxygen parts (remove all lines starting with #') and your name on it, so they can copy it to their package.

Hmm, I think it might just be easier to depend on poorman to be honest. As more features get added, the code is likely to get more complex. Given there will only ever be the one package to depend on with poorman, I don’t think it’s too much of an issue.

[msberends]

👍