Unique operator

[hadley]

Or similar - would translate to DISTINICT in SQL.

[romainfrancois]

Perhaps distinct so that we don't have more problems with the namespace police. Should not be too much trouble to use the visitors classes we already have to get unique rows more efficiently than what R's unique does.

[romainfrancois]

Alright, R's unique pastes all columns together and then uses duplicated on that silly character vector.

> unique.data.frame
function (x, incomparables = FALSE, fromLast = FALSE, ...)
{
    if (!identical(incomparables, FALSE))
        .NotYetUsed("incomparables != FALSE")
    x[!duplicated(x, fromLast = fromLast, ...), , drop = FALSE]
}
<bytecode: 0x105a2a188>
<environment: namespace:base>
> duplicated.data.frame
function (x, incomparables = FALSE, fromLast = FALSE, ...)
{
    if (!identical(incomparables, FALSE))
        .NotYetUsed("incomparables != FALSE")
    if (length(x) != 1L)
        duplicated(do.call("paste", c(x, sep = "\r")), fromLast = fromLast)
    else duplicated(x[[1L]], fromLast = fromLast, ...)
}

[hadley]

I like distinct! Another option would be to use union which I've already made generic. We also need intersect and setdiff for data frames to cover the remaining options.

[romainfrancois]

I pushed some initial code.

> distinct <- distinct_impl
> mtcars %.% select(cyl) %.% distinct()
  cyl
1   6
2   4
3   8

Should the data be ordered at the end ? At the moment, it is in the order of appearance in the original data.

[romainfrancois]

Turns out, we already have union, intersect and setdiff, match:
https://github.com/hadley/dplyr/blob/master/src/dplyr.cpp#L806

[romainfrancois]

Perhaps something to do better than anyDuplicated. http://stackoverflow.com/questions/22834725/r-checking-if-a-set-of-variables-forms-a-unique-index/22834888#22834888

[hadley]

Looks good. I'll add the generic and methods for data frames, data table and sql.

[hadley]

@romainfrancois could you add a second argument to distinct_impl, a vector of column names/indices to use? Convention is to take first row if multiple matches.