Improve on cols_hide() messages on edge cases and allow empty input

[olivroy]

Summary

Add cols_hide() and put better user-facing messages when something goes wrong. Follow-up to #1342

Outstanding

exibble |> gt() |> cols_hide(everything()) still gives an ugly error, but I will leave that to you!

Edit: I just discovered that it worked as expected in gt 0.10.0, but stopped working in gt 0.10.1.

It now gives Error in has_rtl[i, ] : subscript out of bounds

Not affected by this PR.

Question

Is there a way to bench mark gt?

# Rough testing proved that my change improved speed, but not convinced..
gt(exibble) |> cols_hide(columns =  fctr)

Improved messages

# before
gt(exibble) |> cols_hide(columns = fcd) 
#> Error in resolve_cols_i(expr = { : Can't select columns that don't exist.
#> ✖ Column `fcd` doesn't exist.

# this PR
Error in `cols_hide()`:
! Can't select columns that don't exist.
✖ Column `fcd` doesn't exist.

Related GitHub Issues and PRs

• fixes cols_hide() has a documentation issue and error msg #1631

Checklist

• I understand and agree to the Code of Conduct.
• I have listed any major changes in the NEWS.
• I have added testthat unit tests to tests/testthat for any new functionality.

[rich-iannone]

I found that the cols_move*() fns had similar issues with incorrect documentation. I pushed a commit that applies the same behavior as cols_hide() but I'm thinking now that those functions should instead error if there is nothing supplied to columns (the last commit here returns data unchanged).

Do you think there should be shared behavior between the column moving and hiding functions?

[olivroy]

I like having cols_hide() working (it worked in 0.10.0)

because

exibble |>
  gt()
  cols_hide(
    fctr
  )

But then if you change your mind, you can just comment it out.

That's actually how I came across this bug.

exibble |>
  gt()
  cols_hide(
    # fctr
  )

I have never used cols_move(). I tend to use relocate() in dplyr steps instead.

However, as a reference, data |> relocate() does nothing.

So, I like how you handled it in the last commit.

[rich-iannone]

LGTM!

[olivroy]

I know it is difficult without a reprex, but do you have any clue on where to start for #1635. It is actually the reason why I made these contributions today.

It has given me headaches for almost a week now. I am still pretty much clueless (apart from using 0.10.0, which I don't really like)

If anything of what I mentioned there rings a bell, please let me know!