compatibility with new paradox (#341)

* check with new paradox

* paradox argument tag -> tags

* use new paradox syntax

* default xor required

* dev cmd check with paradox master

* news

* fix xgboost parameter test

* ...

---------

Co-authored-by: Sebastian Fischer <sebf.fischer@gmail.com>

.github/workflows/devcmdcheck.yml

@@ -0,0 +1,72 @@
+# r cmd check workflow of the mlr3 ecosystem v0.3.1
+# https://github.com/mlr-org/actions
+on:
+  workflow_dispatch:
+    inputs:
+      debug_enabled:
+        type: boolean
+        description: 'Run the build with tmate debugging enabled'
+        required: false
+        default: false
+  push:
+    branches:
+      - main
+  pull_request:
+    branches:
+      - main
+  schedule:
+    - cron:  '0 4 * * 1'
+
+name: Dev Check
+
+jobs:
+  check-package:
+    runs-on: ${{ matrix.config.os }}
+
+    name: ${{ matrix.config.dev-package }}
+
+    strategy:
+      fail-fast: false
+      matrix:
+        config:
+          - {os: ubuntu-20.04, r: 'release', dev-package: "mlr-org/paradox', 'mlr-org/mlr3learners', 'mlr-org/mlr3pipelines', 'mlr-org/mlr3cluster"}
+
+    env:
+      GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: r-lib/actions/setup-pandoc@v2
+
+      - uses: r-lib/actions/setup-r@v2
+        with:
+          r-version: ${{ matrix.config.r }}
+
+      - uses: r-lib/actions/setup-r-dependencies@v2
+        with:
+          extra-packages: any::rcmdcheck
+          needs: check
+
+      - uses: mxschmitt/action-tmate@v3
+        if: ${{ github.event_name == 'workflow_dispatch' && inputs.debug_enabled }}
+        with:
+          limit-access-to-actor: true
+
+      - name: Install Python packages
+        run: |
+          pak::pkg_install('reticulate')
+          reticulate::install_miniconda(force = TRUE)
+          reticulate::use_condaenv('r-reticulate')
+          install.packages('keras')
+          reticulate::py_install("pytorch-cpu")
+          reticulate::py_install("tensorflow-cpu", pip = TRUE)
+          survivalmodels::install_pycox(pip = TRUE, install_torch = FALSE)
+          survivalmodels::install_keras(pip = TRUE, install_tensorflow = FALSE)
+        shell: Rscript {0}
+
+      - name: Install dev versions
+        run: pak::pkg_install(c('${{ matrix.config.dev-package }}'))
+        shell: Rscript {0}
+
+      - uses: r-lib/actions/check-r-package@v2

NEWS.md

@@ -16,6 +16,7 @@ learners (thanks to @bblodfon)
 * Fix: `I` and `F` params from IBk learner are too interdependent (`I` can only be `TRUE` when `F` is `FALSE` and vice versa).
   Combined them into one factor param `weight` that has two levels -- `I` and `F`.
 * Fix: `U` must be `FALSE` for `S` to be tunable in J48 learner.
+* Compatibility with upcoming 'paradox' release.
 
 # mlr3extralearners 0.7.1
 

R/learner_Cubist_regr_cubist.R

@@ -24,18 +24,16 @@ LearnerRegrCubist = R6Class("LearnerRegrCubist",
     #' Creates a new instance of this [R6][R6::R6Class] class.
     initialize = function() {
       # FIXME:
-      param_set = ParamSet$new(
-        list(
-          ParamInt$new("committees", lower = 1L, upper = 100L, default = 1L, tags = c("train", "required")),
-          ParamLgl$new("unbiased", default = FALSE, tags = "train"),
-          ParamInt$new("rules", lower = 1L, default = 100L, tags = "train"),
-          ParamDbl$new("extrapolation", lower = 0, upper = 100, default = 100, tags = "train"),
-          ParamInt$new("sample", lower = 0L, default = 0L, tags = "train"),
-          ParamInt$new("seed", default = NO_DEF, tags = "train"),
-          ParamUty$new("label", default = "outcome", tags = "train"),
-          ParamInt$new("neighbors", lower = 0L, upper = 9L, default = 0L, tags = c("predict", "required"))
+      param_set = ps(
+          committees = p_int(lower = 1L, upper = 100L, tags = c("train", "required")),
+          unbiased = p_lgl(default = FALSE, tags = "train"),
+          rules = p_int(lower = 1L, default = 100L, tags = "train"),
+          extrapolation = p_dbl(lower = 0, upper = 100, default = 100, tags = "train"),
+          sample = p_int(lower = 0L, default = 0L, tags = "train"),
+          seed = p_int(default = NO_DEF, tags = "train"),
+          label = p_uty(default = "outcome", tags = "train"),
+          neighbors = p_int(lower = 0L, upper = 9L, tags = c("predict", "required"))
         )
-      )
       param_set$values$committees = 1L
       param_set$values$neighbors = 0L
 

R/learner_aorsf_surv_aorsf.R

@@ -62,7 +62,7 @@ LearnerSurvAorsf = R6Class("LearnerSurvAorsf",
         split_min_stat = p_dbl(default = NULL, special_vals = list(NULL), lower = 0, tags = "train"),
         oobag_pred_type = p_fct(levels = c("none", "surv", "risk", "chf"), default = "surv", tags = "train"),
         importance = p_fct(levels = c("none", "anova", "negate", "permute"), default = "anova", tags = "train"),
-        importance_max_pvalue = p_dbl(default = 0.01, lower = 0.0001, upper = .9999, tag = "train"),
+        importance_max_pvalue = p_dbl(default = 0.01, lower = 0.0001, upper = .9999, tags = "train"),
         oobag_pred_horizon = p_dbl(default = NULL, special_vals = list(NULL), tags = "train", lower = 0),
         oobag_eval_every = p_int(default = NULL, special_vals = list(NULL), lower = 1, tags = "train"),
         attach_data = p_lgl(default = TRUE, tags = "train"),

R/learner_lme4_classif_glmer.R

@@ -28,7 +28,7 @@ LearnerClassifGlmer = R6Class("LearnerClassifGlmer",
     initialize = function() {
       action_levels = c("ignore", "warning", "message", "stop")
       param_set = ps(
-        formula = p_uty(default = formula(), tags = c("train", "required"), custom_check = check_formula),
+        formula = p_uty(tags = c("train", "required"), custom_check = check_formula),
         start = p_uty(default = NULL, tags = "train"),
         verbose = p_int(default = 0, lower = 0, tags = "train"),
         offset = p_uty(default = NULL, tags = "train"),

R/learner_lme4_regr_lmer.R

@@ -26,120 +26,102 @@ LearnerRegrLmer = R6Class("LearnerRegrLmer",
     initialize = function() {
       action_levels = c("ignore", "warning", "message", "stop")
 
-      ps = ParamSet$new(list(
+      ps = ps(
         # lmer gives a lot of freedom in the formula spec, so we ask
         # for the formula directly. The formula *must* be set.
-        ParamUty$new(
-          id = "formula",
+        formula = p_uty(
           tags = c("required", "train"),
-          default = formula(),
           custom_check = check_formula
         ),
-        ParamLgl$new(id = "REML", tags = "train", default = TRUE),
-        ParamUty$new(id = "start", tags = "train", default = NULL),
-        ParamInt$new(id = "verbose", tags = "train", default = 0, lower = 0),
-        ParamUty$new(id = "offset", tags = "train", default = NULL),
-        ParamUty$new(id = "contrasts", tags = "train", default = NULL),
+        REML = p_lgl(tags = "train", default = TRUE),
+        start = p_uty(tags = "train", default = NULL),
+        verbose = p_int(tags = "train", default = 0, lower = 0),
+        offset = p_uty(tags = "train", default = NULL),
+        contrasts = p_uty(tags = "train", default = NULL),
 
         # Params passed to lmerControl()
-        ParamFct$new(
-          id = "optimizer",
+        optimizer = p_fct(
           levels = c("Nelder_Mead", "bobyqa", "nlminbwrap", "nloptwrap"),
           default = "nloptwrap",
           tags = "train"
         ),
-        ParamLgl$new(
-          id = "restart_edge",
+        restart_edge = p_lgl(
           default = FALSE,
           tags = "train"),
-        ParamDbl$new(
-          id = "boundary.tol",
+        boundary.tol = p_dbl(
           default = 1e-5,
           lower = 0,
           tags = "train"
         ),
-        ParamLgl$new(
-          id = "calc.derivs",
+        calc.derivs = p_lgl(
           default = TRUE,
           tags = "train"),
         # Input checks
-        ParamFct$new(
-          id = "check.nobs.vs.rankZ",
+        check.nobs.vs.rankZ = p_fct(
           levels = action_levels,
           default = "ignore",
           tags = "train"),
-        ParamFct$new(
-          id = "check.nobs.vs.nlev",
+        check.nobs.vs.nlev = p_fct(
           levels = action_levels,
           default = "stop",
           tags = "train"),
-        ParamFct$new(
-          id = "check.nlev.gtreq.5",
+        check.nlev.gtreq.5 = p_fct(
           levels = action_levels,
           default = "ignore",
           tags = "train"),
-        ParamFct$new(
-          id = "check.nlev.gtr.1",
+        check.nlev.gtr.1 = p_fct(
           levels = action_levels,
           default = "stop",
           tags = "train"),
-        ParamFct$new(
-          id = "check.nobs.vs.nRE",
+        check.nobs.vs.nRE = p_fct(
           levels = action_levels,
           default = "stop",
           tags = "train"),
-        ParamFct$new(
-          id = "check.rankX",
+        check.rankX = p_fct(
           levels = c(
             "message+drop.cols", "silent.drop.cols", "warn+drop.cols",
             "stop.deficient", "ignore"),
           default = "message+drop.cols",
           tags = "train"
         ),
-        ParamFct$new(
-          id = "check.scaleX",
+        check.scaleX = p_fct(
           levels = c(
             "warning", "stop", "silent.rescale",
             "message+rescale", "warn+rescale", "ignore"),
           default = "warning",
           tags = "train"
         ),
-        ParamFct$new(
-          id = "check.formula.LHS",
+        check.formula.LHS = p_fct(
           levels = action_levels,
           default = "stop",
           tags = "train"
         ),
         # Convergence checks
-        ParamUty$new(
-          id = "check.conv.grad",
+        check.conv.grad = p_uty(
           default = 'lme4::.makeCC("warning", tol = 2e-3, relTol = NULL)',
           tags = "train"
         ),
-        ParamUty$new(
-          id = "check.conv.singular",
+        check.conv.singular = p_uty(
           default = 'lme4::.makeCC(action = "message", tol = formals(lme4::isSingular)$tol)',
           tags = "train"
         ),
-        ParamUty$new(
-          id = "check.conv.hess",
+        check.conv.hess = p_uty(
           default = 'lme4::.makeCC(action = "warning", tol = 1e-6)',
           tags = "train"
         ),
         # Additional optimizer controls
-        ParamUty$new(
-          id = "optCtrl",
+        optCtrl = p_uty(
           default = list(),
           tags = "train"
         ),
 
         # Prediction params
-        ParamUty$new(id = "newparams", default = NULL, tags = "predict"),
-        ParamUty$new(id = "re.form", default = NULL, tags = "predict"),
-        ParamLgl$new(id = "random.only", default = FALSE, tags = "predict"),
-        ParamLgl$new(id = "allow.new.levels", default = FALSE, tags = "predict"),
-        ParamUty$new(id = "na.action", default = "stats::na.pass", tags = "predict")
-      ))
+        newparams = p_uty(default = NULL, tags = "predict"),
+        re.form = p_uty(default = NULL, tags = "predict"),
+        random.only = p_lgl(default = FALSE, tags = "predict"),
+        allow.new.levels = p_lgl(default = FALSE, tags = "predict"),
+        na.action = p_uty(default = "stats::na.pass", tags = "predict")
+      )
 
       super$initialize(
         id = "regr.lmer",

R/learner_prioritylasso_classif_priority_lasso.R

@@ -22,8 +22,8 @@ LearnerClassifPriorityLasso = R6Class("LearnerClassifPriorityLasso",
     #' Creates a new instance of this [R6][R6::R6Class] class.
     initialize = function() {
       param_set = ps(
-        blocks               = p_uty(default = NULL, tags = c("train", "required")),
-        type.measure         = p_fct(default = "class", levels = c("class", "auc"), tags = c("train", "required")),
+        blocks               = p_uty(tags = c("train", "required")),
+        type.measure         = p_fct(levels = c("class", "auc"), tags = c("train", "required")),
         max.coef             = p_uty(default = NULL, tags = "train"),
         block1.penalization  = p_lgl(default = TRUE, tags = "train"),
         lambda.type          = p_fct(default = "lambda.min", levels = c("lambda.min", "lambda.1se"), tags = c("train", "predict")), #nolint

R/learner_prioritylasso_regr_priority_lasso.R

@@ -23,7 +23,7 @@ LearnerRegrPriorityLasso = R6Class("LearnerRegrPriorityLasso",
     #' Creates a new instance of this [R6][R6::R6Class] class.
     initialize = function() {
       param_set = ps(
-        blocks               = p_uty(default = NULL, tags = c("train", "required")), 
+        blocks               = p_uty(tags = c("train", "required")),
         max.coef             = p_uty(default = NULL, tags = "train"),
         block1.penalization  = p_lgl(default = TRUE, tags = "train"),
         lambda.type          = p_fct(default = "lambda.min", levels = c("lambda.min", "lambda.1se"), tags = c("train", "predict")),

R/learner_prioritylasso_surv_priority_lasso.R

@@ -22,7 +22,7 @@ LearnerSurvPriorityLasso = R6Class("LearnerSurvPriorityLasso",
     #' Creates a new instance of this [R6][R6::R6Class] class.
     initialize = function() {
       param_set = ps(
-        blocks               = p_uty(default = NULL, tags = c("train", "required")),
+        blocks               = p_uty(tags = c("train", "required")),
         max.coef             = p_uty(default = NULL, tags = "train"),
         block1.penalization  = p_lgl(default = TRUE, tags = "train"),
         lambda.type          = p_fct(default = "lambda.min", levels = c("lambda.min", "lambda.1se"), tags = c("train", "predict")), #nolint

R/learner_rsm_regr_rsm.R

@@ -28,7 +28,7 @@ LearnerRegrRSM = R6Class("LearnerRegrRSM",
     #' Creates a new instance of this [R6][R6::R6Class] class.
     initialize = function() {
       param_set = ps(
-        modelfun = p_fct(default = "FO", levels = c("FO", "TWI", "SO"), tags = c("train", "required"))
+        modelfun = p_fct(levels = c("FO", "TWI", "SO"), tags = c("train", "required"))
       )
       param_set$values = list(modelfun = "FO")
 

R/learner_survival_surv_parametric.R

@@ -80,7 +80,7 @@ LearnerSurvParametric = R6Class("LearnerSurvParametric",
         robust = p_lgl(default = FALSE, tags = "train"),
         score = p_lgl(default = FALSE, tags = "train"),
         cluster = p_uty(tags = "train"),
-        discrete = p_lgl(default = FALSE, tags = c("required", "predict"))
+        discrete = p_lgl(tags = c("required", "predict"))
       )
 
       ps$values = list(discrete = FALSE, dist = "weibull", type = "aft")

tests/testthat/test_paramtest_xgboost_surv_xgboost.R

@@ -1,6 +1,7 @@
 x = rvest::read_html("https://xgboost.readthedocs.io/en/latest/parameter.html")
-x = rvest::html_elements(x, c("li", "p"))
-x = rvest::html_text2(x)
+xli = rvest::html_elements(x, "li")
+xp = rvest::html_elements(x, "p")
+x = c(rvest::html_text2(xli), rvest::html_text2(xp))
 x = grep("default=", x, value = TRUE)
 x = strsplit(x, split = " ")
 x = mlr3misc::map_chr(x, function(x) x[1])