search

tidymodels / extratests

Integration and other testing for tidymodels
Other
20 stars 1 forks source link

new failures with glmnet penalty checks #209

Closed simonpcouch closed 8 months ago

simonpcouch commented 8 months ago

Including on the CRAN R CMD CHECK workflow, we're seeing:

  old[2:6] vs new[2:6]
      multinom_reg(penalty = 0.01) %>% set_engine("glmnet") %>% fit(class ~ ., data = hpc_data) %>%
        predict(hpc_data, penalty = 0:1)
    Condition
  -   Error in `.check_glmnet_penalty_predict()`:
  +   Error:
  -   ! `penalty` should be a single numeric value. `multi_predict()` can be used to get multiple predictions per row of data.
  +   ! argument "..3" is missing, with no default

CRAN checks for parsnip and censored are fine.

simonpcouch commented 8 months ago

R 4.3.3 was just released—maybe that just kicked in as the (release) version on GHA and this functionality was only tested in extratests?