Uninformative error if wrong control_*() is used

[EmilHvitfeldt]

See reprex below. This should be a fairly easy since since the control functions return specific classes we can check against.

library(tidymodels)

rec_no_tune_1 <-
  recipe(mpg ~ ., data = mtcars) %>%
  step_normalize(all_predictors())

set.seed(363)
mt_folds <- vfold_cv(mtcars, v = 5)

glmn_mod <- decision_tree(tree_depth = tune(), min_n = 2) %>% 
  set_mode("regression") %>%
  set_engine("rpart")

workflow() %>%
  add_recipe(rec_no_tune_1) %>%
  add_model(glmn_mod) %>%
  tune_bayes(mt_folds, control = control_grid())
#> Error in if (is.na(limit)) {: argument is of length zero

^{Created on 2022-02-18 by the reprex package (v2.0.1)}

[topepo]

In iterative tuning methods and racing, we pass a single control function for the function (e.g. tune_sim_anneal()) and that is used to work with another function, usually tune_grid().

I think that we need a set of as_control_grid() methods that will convert from one control type to another. We’d need these to go from more complete methods to less complex. For example, we would need as_control_resamples.control_grid() but not as_control_grid.control_resamples().

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.