update to testthat 3e + consolidate translate() tests (#699)

* bump testthat version

* drop calls to `context()`

if `context()` wasn't redundant with file name, convert to comment

* remove usage of `expect_equivalent()`

mostly just switching to `expect_equal` and updating attributes as needed. it also seems that `expect_warning()` (and other prompt friends) return the prompt rather than output--changed one of these that was related to `expect_equivalent, but will need to do a more thorough sweep when switching to `expect_snapshot`

* favor `expect_equal` args over ad-hoc deattr

* drop calls to `source`, standardize header

* drop `library` calls, namespace w `::` when needed

* drop `new_empty_quosure()`

`quo` does the trick in most of these situations, and any exceptions ought to be transitioned to `expect_snapshot`. still a good few errors related to function environment.

* add `test_translate` skeleton, spec out with `linear_reg`

* add helper for `expect_snapshot`ting translate `args`

I think that it helps to standardize indentation + simplify figuring out what a snapshot is capturing, and it cuts down on lines quite a bit, but I could also see the argument that it obscures a bit too much. Implemented primary argument checks for three models, dropping the rest.

* adjust layout of `test_translate.R`

it looks like just about every model type has associated changes to translation of primary arguments, so pulling back from the "selected methods" approach to testing each.

* condense primary arguments testing

* refactor engine arguments tests

* `expect_warning` -> `expect_snapshot` when `regex` arg present

also transitioned on output of `translate`

* revert cc6bb34

given how large this PR will be already, and the brittleness of `expect_snapshot` as compared to `expect_warning` (and `expect_error`), reverting the change `expect_warning` <- `expect_snapshot`.

* `expect_output` -> `expect_snapshot`

* revert to simpler expectation

following Hannah's lead--we just want to make sure we get a model object back, and `expect_snapshot` will pick up differences in fit time etc

* add snaps

* misc bug fixes

a good bit of `expect_warning` output changes affecting assignments. some namespacing, some environment variable tomfoolery as well.

* adjust helper interface

ended up running into issues with `expect_snapshot` call environments—not worth the loss in interpretability of errors to wrap `expect_snapshot` itself.

still trying to figure out the best way to ignore function environments here.

* make envs reproducible, update snaps

wipe the environment from quosures in the translation helper so that `expect_snapshot` is reproducible!

* small `r cmd check` fixes

* typo fix

* invoke `lung` outside of test body

for whatever reason, `oldrel-2` can't find `lung` inside of the tests here. thinking this might be some kind of scoping issue—from what i can tell on the `survival` repo, this shouldn't be due to changes in that package, though it's hard to trace source history from there.

* load `lung` data conditionally

for `oldrel-2` and older, pak uses survival 3.1-8 via R 3.6.3. in these versions of survival, loading the lung data with `data(lung` seems to have worked (though was not preferred? ¯\_(ツ)_/¯)

* transition `$fit` to `extract_fit_*()`

exceptions:
* when assigning (i.e. no `extract_fit_*<-()`)
* when subsetted into `$method`
* when testing `NULL`

* transition `update()` tests to snapshots

* transition last bits of `rlang` nastiness to `expect_snapshot()`

* revert to `quo` for ignoring env

* drop outdated snapshot

Author: simonpcouch

DESCRIPTION

@@ -60,7 +60,7 @@ Suggests:
     rpart,
     sparklyr (>= 1.0.0),
     survival,
-    testthat,
+    testthat (>= 3.0.0),
     xgboost (>= 1.5.0.1)
 VignetteBuilder: 
     knitr
@@ -89,4 +89,5 @@ Config/rcmdcheck/ignore-inconsequential-notes: true
 Encoding: UTF-8
 LazyData: true
 Roxygen: list(markdown = TRUE)
+Config/testthat/edition: 3
 RoxygenNote: 7.1.2.9000

tests/testthat/_snaps/boost_tree.md

@@ -0,0 +1,92 @@
+# updating
+
+    Code
+      expr1 %>% update(trees = 10)
+    Output
+      Boosted Tree Model Specification (unknown)
+      
+      Main Arguments:
+        trees = 10
+      
+      Engine-Specific Arguments:
+        verbose = 0
+      
+      Computational engine: xgboost 
+      
+
+---
+
+    Code
+      expr1 %>% update(param_tibb)
+    Output
+      Boosted Tree Model Specification (unknown)
+      
+      Main Arguments:
+        mtry = 1
+        trees = 7
+      
+      Engine-Specific Arguments:
+        verbose = 0
+      
+      Computational engine: xgboost 
+      
+
+---
+
+    Code
+      expr1 %>% update(param_list)
+    Output
+      Boosted Tree Model Specification (unknown)
+      
+      Main Arguments:
+        mtry = 1
+        trees = 7
+      
+      Engine-Specific Arguments:
+        verbose = 0
+      
+      Computational engine: xgboost 
+      
+
+---
+
+    Code
+      expr2 %>% update(bands = 10)
+    Output
+      Boosted Tree Model Specification (unknown)
+      
+      Main Arguments:
+        trees = tune()
+      
+      Engine-Specific Arguments:
+        bands = 10
+      
+      Computational engine: C5.0 
+      
+
+---
+
+    Code
+      expr3 %>% update(trees = 1, fresh = TRUE)
+    Output
+      Boosted Tree Model Specification (unknown)
+      
+      Main Arguments:
+        trees = 1
+      
+      Computational engine: xgboost 
+      
+
+---
+
+    Code
+      expr4 %>% update(noGlobalPruning = TRUE)
+    Output
+      Boosted Tree Model Specification (unknown)
+      
+      Engine-Specific Arguments:
+        noGlobalPruning = TRUE
+      
+      Computational engine: C5.0 
+      
+

tests/testthat/_snaps/decision_tree.md

@@ -0,0 +1,79 @@
+# updating
+
+    Code
+      expr1 %>% update(cost_complexity = 0.1)
+    Output
+      Decision Tree Model Specification (unknown)
+      
+      Main Arguments:
+        cost_complexity = 0.1
+      
+      Engine-Specific Arguments:
+        model = FALSE
+      
+      Computational engine: rpart 
+      
+
+---
+
+    Code
+      expr1 %>% update(param_tibb)
+    Output
+      Decision Tree Model Specification (unknown)
+      
+      Main Arguments:
+        cost_complexity = 0.1
+        min_n = 1
+      
+      Engine-Specific Arguments:
+        model = FALSE
+      
+      Computational engine: rpart 
+      
+
+---
+
+    Code
+      expr1 %>% update(param_list)
+    Output
+      Decision Tree Model Specification (unknown)
+      
+      Main Arguments:
+        cost_complexity = 0.1
+        min_n = 1
+      
+      Engine-Specific Arguments:
+        model = FALSE
+      
+      Computational engine: rpart 
+      
+
+---
+
+    Code
+      expr2 %>% update(model = FALSE)
+    Output
+      Decision Tree Model Specification (unknown)
+      
+      Main Arguments:
+        cost_complexity = tune()
+      
+      Engine-Specific Arguments:
+        model = FALSE
+      
+      Computational engine: rpart 
+      
+
+---
+
+    Code
+      expr3 %>% update(cost_complexity = 1, fresh = TRUE)
+    Output
+      Decision Tree Model Specification (unknown)
+      
+      Main Arguments:
+        cost_complexity = 1
+      
+      Computational engine: rpart 
+      
+

tests/testthat/_snaps/linear_reg.md

@@ -0,0 +1,109 @@
+# updating
+
+    Code
+      expr1 %>% update(mixture = 0)
+    Output
+      Linear Regression Model Specification (regression)
+      
+      Main Arguments:
+        mixture = 0
+      
+      Engine-Specific Arguments:
+        model = FALSE
+      
+      Computational engine: lm 
+      
+
+---
+
+    Code
+      expr2 %>% update(nlambda = 10)
+    Output
+      Linear Regression Model Specification (regression)
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+
+---
+
+    Code
+      expr3 %>% update(mixture = 1, fresh = TRUE, nlambda = 10)
+    Output
+      Linear Regression Model Specification (regression)
+      
+      Main Arguments:
+        mixture = 1
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+
+---
+
+    Code
+      expr3 %>% update(nlambda = 10)
+    Output
+      Linear Regression Model Specification (regression)
+      
+      Main Arguments:
+        penalty = tune()
+        mixture = 0
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+
+---
+
+    Code
+      expr4 %>% update(param_tibb)
+    Output
+      Linear Regression Model Specification (regression)
+      
+      Main Arguments:
+        penalty = 1
+        mixture = 0.333333333333333
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+
+---
+
+    Code
+      expr4 %>% update(param_list)
+    Output
+      Linear Regression Model Specification (regression)
+      
+      Main Arguments:
+        penalty = 1
+        mixture = 0.333333333333333
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+
+---
+
+    Code
+      expr5 %>% update(family = "poisson")
+    Output
+      Linear Regression Model Specification (regression)
+      
+      Engine-Specific Arguments:
+        family = poisson
+      
+      Computational engine: glm 
+      
+

tests/testthat/_snaps/logistic_reg.md

@@ -0,0 +1,82 @@
+# updating
+
+    Code
+      expr1 %>% update(mixture = 0)
+    Output
+      Logistic Regression Model Specification (classification)
+      
+      Main Arguments:
+        mixture = 0
+      
+      Engine-Specific Arguments:
+        family = expr(binomial(link = "probit"))
+      
+      Computational engine: glm 
+      
+
+---
+
+    Code
+      expr2 %>% update(nlambda = 10)
+    Output
+      Logistic Regression Model Specification (classification)
+      
+      Main Arguments:
+        mixture = tune()
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+
+---
+
+    Code
+      expr3 %>% update(mixture = 1, fresh = TRUE, nlambda = 10)
+    Output
+      Logistic Regression Model Specification (classification)
+      
+      Main Arguments:
+        mixture = 1
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+
+---
+
+    Code
+      expr4 %>% update(param_tibb)
+    Output
+      Logistic Regression Model Specification (classification)
+      
+      Main Arguments:
+        penalty = 1
+        mixture = 0.333333333333333
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+
+---
+
+    Code
+      expr4 %>% update(param_list)
+    Output
+      Logistic Regression Model Specification (classification)
+      
+      Main Arguments:
+        penalty = 1
+        mixture = 0.333333333333333
+      
+      Engine-Specific Arguments:
+        nlambda = 10
+      
+      Computational engine: glmnet 
+      
+