testing updates after inactivity #121
Conversation
by skipping the snapshot test on multiple penalty values on the (current) CRAN - will run again in the CRAN GHA once poissonreg is released
|
Whoo, it skips the right tests now! Status on Matrix: the |
|
Looks like we need the following versions for those three packages (based on an issue comment on the lme4 repo): The session info shows that we are getting those versions for lme4 and RcppEigen but not yet for Matrix. That one gets installed in version 1.6-1.1 from CRAN. So I'm also going to change the skip for |
which got released a few days ago
| @@ -121,19 +121,25 @@ test_that("formula interface can deal with missing values", { | |||
| expect_true(is.na(f_pred$.pred[1])) | |||
| }) | |||
|
|
|||
| test_that('error traps', { | |||
There was a problem hiding this comment.
Appreciating how this was split up. :)
| Error <rlang_error> | ||
| `data` must only contain numeric columns. | ||
| Condition | ||
| Error in `recompose()`: |
There was a problem hiding this comment.
Is this a regression? Not sure if this is a change in rlang or our own, but if this seems like a step backward I say we file an issue.
There was a problem hiding this comment.
edit: referring to the reference to recompose() newly showing up.
| Error <rlang_error> | ||
| For the glmnet engine, `penalty` must be a single number (or a value of `tune()`). | ||
| Condition | ||
| Error in `.check_glmnet_penalty_fit()`: |
There was a problem hiding this comment.
Same thought re: new call output.
|
Re error call: |
|
That makes sense, thanks! :) |
|
I shall merge then! 🎉 |
No description provided.