Better handling of infix operators #84
Comments
|
Another error with +:
|
* First pass at revised detect_mistakes - solves initial + inquiry. * Debugs detect_mistake to pass all tests, adds test for + case, edits detect_mistakes to flagged unmatched named arguments passed to ... as surplus * Adds tests for #84, #91, #94, #95. * Cleans up tests in test_detect_mistakes.R. Tests no longer use name of first argument. * Checks user code for bad syntax before standardising. Fixes #91, #94, #95. * Adds message generators for new cases: wrong_call(), bad_argument_name(), too_many_arguments(). * Adds new message generators to detect_mistakes. * Adds enclosing calls to student messages * Adds sanity-check.Rmd for checking that messages are sensible (which is not easy to automate). * Streamlines message generators. * Fixes bug in detect_mistakes that failed to catch unused unnamed arguments before passing student code to call_standardise_formals. * Improves the messages returned by the message generators. * Completes suite of sanity_checks for reading whether or not gradethis provides intelligible messages to students. * Corrects argument names in too_many_matches * Adds checks in sanity_checks.Rmd to tests as test_sanity_checks.R * Moves sanity-checks.Rmd to the tutorials as grade_code-messages * Changes content of old sanity-checks.R to be a tutorial. * Fixes awkward message returned by missing_argument when the missing argument does not have a name. * Fixes hard-coded tests that incorrectly fail in the presence of improvements. * Adds your_char variable to missing_argument() based on reviewer feedback * Improves handling of comparisons in build_intro() based on reviewer feedback. * Checks for case where user uses the same argument name twice. Adds duplicate_name() message generator. * Small improvements based on reviewer feedback * Fixes bug in how detect_mistakes handles partially matched formal names. * Further refactoring with real_names() inside of detect_mistakes() * Fixes bug that caused detect_mistakes to prematurely return NULL (correct) is one set of non-identical arguments were none-the-less correct.
|
Here's a reprex as this stands now. @garrettgman do you have any suggestions for the messages we wish to see? pkgload::load_all("~/work/gradethis")
#> Loading gradethis (a8f772102d5b51c2ff3810d45fb4f532f94d8a72)
code_feedback("sqrt(1)", "sqrt(1 + 2)")
#> In `sqrt(1)`, I expected `+` where you wrote `1`.
code_feedback("0 + sqrt(log(2))", "sqrt(log(2))")
#> I expected you to call `sqrt()` where you called `+`. |
|
For infix operators , I think we need to say "something
Suggests to a new user that they should do this, which leads them to a frustrating dead end that they may not be able to see their way out of:
A better message would be "In This replacement should generalize to all infix operators. Should I open this as a separate issue? Its not clear to me what the thrust of Daniel's issue was. |
|
Here's another example. ```{r starwars-setup}
library(dplyr)
```
```{r starwars, exercise=TRUE}
starwars %>%
transmute(
mass = as.integer(mass * 2.205),
height = as.integer(height / 2.54)
)
```
```{r starwars-solution}
starwars %>%
transmute(
height = height / 2.54,
mass = mass * 2.205
)
```
```{r starwars-check}
grade_this_table(check_column_order = TRUE)
```
|
given:
returns
The text was updated successfully, but these errors were encountered: