Make message more informative

[rossellhayes]

• length problem message now includes the first missing or unexpected value if two conditions are met:
  • The difference between the expected and actual length ≤ 2.
  • There are ≤ 2 different values between the expected and actual vector.
  • In cases where the vector lengths differ by more than two or where there are more than two different values between the vectors, existing behavior is retained (i.e. only the difference in length is signalled).

library(tblcheck)

# Lengths differ <= 2 and <= 2 value differences
.result <- 1:3
.solution <- 1:5
vec_grade_length()
#> <gradethis_graded: [Incorrect]
#>   Your result should contain 5 values, but it has 3 values. I expected
#>   your result to include the value `4`.
#> >

# Lengths differ > 2
.result <- 1:3
.solution <- 1:6
vec_grade_length()
#> <gradethis_graded: [Incorrect]
#>   Your result should contain 6 values, but it has 3 values.
#> >

# > 2 value differences
.result <- 1:3
.solution <- 4:7
vec_grade_length()
#> <gradethis_graded: [Incorrect]
#>   Your result should contain 4 values, but it has 3 values.
#> >

• values problem messages now inform the user of problems after the first max_diffs values:

library(tblcheck)

# Difference within first three values
.result <- 4:13
.solution <- 1:10
vec_grade_values()
#> <gradethis_graded: [Incorrect]
#>   The first 3 values of your result should be `1`, `2`, and `3`, not
#>   `4`, `5`, and `6`.
#> >

# Unexpected values after first three
.result <- c(1:3, 7:13)
.solution <- 1:10
vec_grade_values()
#> <gradethis_graded: [Incorrect]
#>   I didn't expect your result to include the values `11`, `12`, and
#>   `13`.
#> >

# Missing values after first three
.result <- c(1:3, rep(4L, 7))
.solution <- 1:10
vec_grade_values()
#> <gradethis_graded: [Incorrect]
#>   I expected your result to include the values `5`, `6`, and `7`.
#> >

^{Created on 2021-10-21 by the reprex package (v2.0.1)}

^{Created on 2021-10-13 by the reprex package (v2.0.1)}

Miscellaneous

• Include vec_*_length() as alias for vec_*_dimensions()

For #65.

[gadenbuie]

It feels like we could be doing a little more here and showing both values if there are 2 or fewer mistakes.

# Lengths differ <= 2 and <= 2 value differences
.result <- 1:3
.solution <- 1:5
vec_grade_length()
#> <gradethis_graded: [Incorrect]
#>   Your result should contain 5 values, but it has 3 values. I expected
#>   your result to include the values `4` and `5`.
#> >

I'm sure we could get into complicated territory around giving better descriptors, e.g.

I expected your result to include the values 4 and 5 after 3.

or showing up to n values when they are contiguous. But that's also a lot more complicated and we can wait on that.

In general, I think we should try hard to make clear the total number of problematic values, even if we only show 1-3 of them explicitly. In the following example, only showing the first 3 values without any more information makes it seem like there the result is off by three when it's in fact off by 6 values.

# Missing values after first three
.result <- c(1:3, rep(4L, 7))
.solution <- 1:10
vec_grade_values()
#> <gradethis_graded: [Incorrect]
#>   I expected your result to include the values `5`, `6`, and `7`.
#> >

We could instead add a little more context

I expected your result to include the values 5, 6, and 7 and 3 other values.

But more broadly, I think we might need to think about an overall different approach for some length/values problems that handles both at the same time to try to give more helpful feedback.