all-columns / columns

[RickMoynihan]

(columns (test-dataset 5 20) [:A :B :D :C])
; =>

| A | B | D |
|---+---+---|
| 0 | 0 | 0 |
| 1 | 1 | 1 |
| 2 | 2 | 2 |
| 3 | 3 | 3 |
| 4 | 4 | 4 |

The above should return a dataset containing the columns :A :B :C :D.

It doesn't go back and get :C. Because it makes a confused assumption that a dataset might be infinite in its columns.

However because of the row oriented nature of most file formats (e.g. csv) Datasets are almost always large in rows not in columns.

As columns is simply selecting a subset of columns from a finite list/vector of columns, it should (take (count (:column-names ds)) cols) to crop the potentially infinite sequence to the dataset and then find all of them.

Once we do this, there will be no need for remove all-columns, and we can just remove it.

[RickMoynihan]

@Robsteranium How does this relate to your recent work on make-dataset? I seem to remember us deciding that both of these fixes were desirable and similar in nature.