[R-package] limit number of threads used in tests and examples (fixes #…

…5102, fixes #5987)

R-package/R/callback.R

@@ -24,7 +24,7 @@ CB_ENV <- R6::R6Class(
 )
 
 # Format the evaluation metric string
-format.eval.string <- function(eval_res, eval_err) {
+.format_eval_string <- function(eval_res, eval_err) {
 
   # Check for empty evaluation string
   if (is.null(eval_res) || length(eval_res) == 0L) {
@@ -40,7 +40,7 @@ format.eval.string <- function(eval_res, eval_err) {
 
 }
 
-merge.eval.string <- function(env) {
+.merge_eval_string <- function(env) {
 
   # Check length of evaluation list
   if (length(env$eval_list) <= 0L) {
@@ -63,7 +63,7 @@ merge.eval.string <- function(env) {
     }
 
     # Set error message
-    msg <- c(msg, format.eval.string(eval_res = env$eval_list[[j]], eval_err = eval_err))
+    msg <- c(msg, .format_eval_string(eval_res = env$eval_list[[j]], eval_err = eval_err))
 
   }
 
@@ -86,11 +86,11 @@ cb_print_evaluation <- function(period) {
       if ((i - 1L) %% period == 0L || is.element(i, c(env$begin_iteration, env$end_iteration))) {
 
         # Merge evaluation string
-        msg <- merge.eval.string(env = env)
+        msg <- .merge_eval_string(env = env)
 
         # Check if message is existing
         if (nchar(msg) > 0L) {
-          print(merge.eval.string(env = env))
+          print(.merge_eval_string(env = env))
         }
 
       }
@@ -270,7 +270,7 @@ cb_early_stop <- function(stopping_rounds, first_metric_only, verbose) {
 
           # Prepare to print if verbose
           if (verbose) {
-            best_msg[[i]] <<- as.character(merge.eval.string(env = env))
+            best_msg[[i]] <<- as.character(.merge_eval_string(env = env))
           }
 
         } else {

R-package/R/lgb.Booster.R

@@ -27,6 +27,23 @@ Booster <- R6::R6Class(
                           modelfile = NULL,
                           model_str = NULL) {
 
+      # LightGBM-internal fix to comply with CRAN policy of only using up to 2 threads in tests and example.
+      #
+      # per https://cran.r-project.org/web/packages/policies.html:
+      #
+      # "If running a package uses multiple threads/cores it must never use more than two simultaneously:
+      #  the check farm is a shared resource and will typically be running many checks simultaneously."
+      #
+      # This mechanism could be removed at any time, and isn't considered part of the public API.
+      #
+      threads_from_opts <- options("lightgbm.cran.testing.threads")[[1L]]
+      if (!is.null(threads_from_opts)) {
+          # put an upper limit on num_threads
+          params[["num_threads"]] <- min(params[["num_threads"]], as.integer(threads_from_opts))
+          # handle the case where 0 is passed to mean "use OpenMP default number of threads"
+          params[["num_threads"]] <- max(params[["num_threads"]], 1L)
+      }
+
       handle <- NULL
 
       if (!is.null(train_set)) {
@@ -928,6 +945,7 @@ NULL
 #'   , metric = "l2"
 #'   , min_data = 1L
 #'   , learning_rate = 1.0
+#'   , num_threads = 2L
 #' )
 #' valids <- list(test = dtest)
 #' model <- lgb.train(
@@ -1086,7 +1104,10 @@ predict.lgb.Booster <- function(object,
 #' X <- as.matrix(mtcars[, -1L])
 #' y <- mtcars[, 1L]
 #' dtrain <- lgb.Dataset(X, label = y, params = list(max_bin = 5L))
-#' params <- list(min_data_in_leaf = 2L)
+#' params <- list(
+#'   min_data_in_leaf = 2L
+#'   , num_threads = 2L
+#' )
 #' model <- lgb.train(
 #'   params = params
 #'  , data = dtrain
@@ -1231,6 +1252,7 @@ summary.lgb.Booster <- function(object, ...) {
 #'   , metric = "l2"
 #'   , min_data = 1L
 #'   , learning_rate = 1.0
+#'   , num_threads = 2L
 #' )
 #' valids <- list(test = dtest)
 #' model <- lgb.train(
@@ -1296,6 +1318,7 @@ lgb.load <- function(filename = NULL, model_str = NULL) {
 #'   , metric = "l2"
 #'   , min_data = 1L
 #'   , learning_rate = 1.0
+#'   , num_threads = 2L
 #' )
 #' valids <- list(test = dtest)
 #' model <- lgb.train(
@@ -1351,6 +1374,7 @@ lgb.save <- function(booster, filename, num_iteration = NULL) {
 #'   , metric = "l2"
 #'   , min_data = 1L
 #'   , learning_rate = 1.0
+#'   , num_threads = 2L
 #' )
 #' valids <- list(test = dtest)
 #' model <- lgb.train(
@@ -1401,6 +1425,7 @@ lgb.dump <- function(booster, num_iteration = NULL) {
 #'   , metric = "l2"
 #'   , min_data = 1L
 #'   , learning_rate = 1.0
+#'   , num_threads = 2L
 #' )
 #' valids <- list(test = dtest)
 #' model <- lgb.train(

R-package/R/lgb.cv.R

@@ -65,6 +65,7 @@ CVBooster <- R6::R6Class(
 #'   , data = dtrain
 #'   , nrounds = 5L
 #'   , nfold = 3L
+#'   , num_threads = 2L
 #' )
 #' }
 #' @importFrom data.table data.table setorderv

R-package/R/lgb.importance.R

@@ -24,6 +24,7 @@
 #'   , max_depth = -1L
 #'   , min_data_in_leaf = 1L
 #'   , min_sum_hessian_in_leaf = 1.0
+#'   , num_threads = 2L
 #' )
 #' model <- lgb.train(
 #'     params = params

R-package/R/lgb.interprete.R

@@ -35,6 +35,7 @@
 #'     , max_depth = -1L
 #'     , min_data_in_leaf = 1L
 #'     , min_sum_hessian_in_leaf = 1.0
+#'     , num_threads = 2L
 #' )
 #' model <- lgb.train(
 #'     params = params

R-package/R/lgb.model.dt.tree.R

@@ -40,6 +40,7 @@
 #'   , max_depth = -1L
 #'   , min_data_in_leaf = 1L
 #'   , min_sum_hessian_in_leaf = 1.0
+#'   , num_threads = 2L
 #' )
 #' model <- lgb.train(params, dtrain, 10L)
 #'

R-package/R/lgb.plot.importance.R

@@ -28,6 +28,7 @@
 #'     , learning_rate = 0.1
 #'     , min_data_in_leaf = 1L
 #'     , min_sum_hessian_in_leaf = 1.0
+#'     , num_threads = 2L
 #' )
 #'
 #' model <- lgb.train(

R-package/R/lgb.plot.interpretation.R

@@ -39,6 +39,7 @@
 #'   , max_depth = -1L
 #'   , min_data_in_leaf = 1L
 #'   , min_sum_hessian_in_leaf = 1.0
+#'   , num_threads = 2L
 #' )
 #' model <- lgb.train(
 #'   params = params

R-package/R/lgb.restore_handle.R

@@ -23,7 +23,9 @@
 #'   , agaricus.train$label
 #'   , params = list(objective = "binary")
 #'   , nrounds = 5L
-#'   , verbose = 0)
+#'   , verbose = 0
+#'   , num_threads = 2L
+#' )
 #' fname <- tempfile(fileext="rds")
 #' saveRDS(model, fname)
 #'

R-package/R/lgb.train.R

@@ -30,6 +30,7 @@
 #'   , metric = "l2"
 #'   , min_data = 1L
 #'   , learning_rate = 1.0
+#'   , num_threads = 2L
 #' )
 #' valids <- list(test = dtest)
 #' model <- lgb.train(

R-package/R/readRDS.lgb.Booster.R

@@ -23,6 +23,7 @@
 #'   , metric = "l2"
 #'   , min_data = 1L
 #'   , learning_rate = 1.0
+#'   , num_threads = 2L
 #' )
 #' valids <- list(test = dtest)
 #' model <- lgb.train(

R-package/R/saveRDS.lgb.Booster.R

@@ -33,6 +33,7 @@
 #'   , metric = "l2"
 #'   , min_data = 1L
 #'   , learning_rate = 1.0
+#'   , num_threads = 2L
 #' )
 #' valids <- list(test = dtest)
 #' model <- lgb.train(

R-package/tests/testthat.R

@@ -1,6 +1,10 @@
 library(testthat)
 library(lightgbm)
 
+options(
+    "lightgbm.cran.testing.threads" = 2.0
+)
+
 test_check(
     package = "lightgbm"
     , stop_on_failure = TRUE

R-package/tests/testthat/test_lgb.Booster.R

@@ -631,6 +631,7 @@ test_that("Booster should store parameters and Booster$reset_parameter() should
         , boosting = "gbdt"
         , num_class = 5L
         , verbose = VERBOSITY
+        , num_threads = 2L
     )
     bst <- Booster$new(
         params = params
@@ -658,6 +659,7 @@ test_that("Booster$params should include dataset params, before and after Booste
         , max_depth = 4L
         , bagging_fraction = 0.8
         , verbose = VERBOSITY
+        , num_threads = 2L
     )
     bst <- Booster$new(
         params = params
@@ -670,6 +672,7 @@ test_that("Booster$params should include dataset params, before and after Booste
             , max_depth = 4L
             , bagging_fraction = 0.8
             , verbose = VERBOSITY
+            , num_threads = 2L
             , max_bin = 17L
         )
     )
@@ -681,6 +684,7 @@ test_that("Booster$params should include dataset params, before and after Booste
         , max_depth = 4L
         , bagging_fraction = 0.9
         , verbose = VERBOSITY
+        , num_threads = 2L
         , max_bin = 17L
     )
     expect_identical(ret_bst$params, expected_params)

R-package/tests/testthat/test_weighted_loss.R

@@ -3,6 +3,8 @@ VERBOSITY <- as.integer(
 )
 
 test_that("Gamma regression reacts on 'weight'", {
+
+  stop(options("lightgbm.cran.testing.threads")[[1L]])
   n <- 100L
   set.seed(87L)
   X <- matrix(runif(2L * n), ncol = 2L)