[R package] move error handling to C++ and minor changes in documenta…

…tion

R-package/R/gpb.Dataset.R

@@ -558,26 +558,27 @@ Dataset <- R6::R6Class(
       if (gpb.is.null.handle(x = private$handle)) {
         private$params <- modifyList(private$params, params)
       } else {
-        call_state <- 0L
-        call_state <- .Call(
-          "LGBM_DatasetUpdateParamChecking_R"
-          , gpb.params2str(params = private$params)
-          , gpb.params2str(params = params)
-          , call_state
-          , PACKAGE = "lib_gpboost"
-        )
-        call_state <- as.integer(call_state)
-        if (call_state != 0L) {
-
-          # raise error if raw data is freed
+        tryCatch({
+          call_state <- 0L
+          .Call(
+            "LGBM_DatasetUpdateParamChecking_R"
+            , gpb.params2str(params = private$params)
+            , gpb.params2str(params = params)
+            , call_state
+            , PACKAGE = "lib_gpboost"
+          )
+        }, error = function(e) {
+          # If updating failed but raw data is not available, raise an error because
+          # achieving what the user asked for is not possible
           if (is.null(private$raw_data)) {
-            gpb.last_error()
+            stop(e)
           }
-
-          # Overwrite paramms
+
+          # If updating failed but raw data is available, modify the params
+          # on the R side and re-set ("deconstruct") the Dataset
           private$params <- modifyList(private$params, params)
           self$finalize()
-        }
+        })
       }
       return(invisible(self))
 
@@ -737,7 +738,7 @@ Dataset <- R6::R6Class(
 #' @param data a \code{matrix} object, a \code{dgCMatrix} object or a character representing a filename
 #' @param params a list of parameters. See
 #'               \href{https://github.com/fabsig/GPBoost/blob/master/docs/Parameters.rst#dataset-parameters}{
-#'               The "Dataset Parameters" section} for a list of parameters
+#'               the "Dataset Parameters" section of the parameter documentation} for a list of parameters
 #'               and valid values.
 #' @param reference reference dataset. When GPBoost creates a Dataset, it does some preprocessing like binning
 #'                  continuous features into histograms. If you want to apply the same bin boundaries from an existing

R-package/R/gpb.train.R

@@ -11,7 +11,7 @@
 #' @param reset_data Boolean, setting it to TRUE (not the default value) will transform the
 #'                   booster model into a predictor model which frees up memory and the
 #'                   original datasets
-#' @param ... other parameters, see \href{https://github.com/fabsig/GPBoost/blob/master/docs/Parameters.rst} for more information. 
+#' @param ... other parameters, see \href{https://github.com/fabsig/GPBoost/blob/master/docs/Parameters.rst}{the parameter documentation} for more information. 
 #' @inheritSection gpb_shared_params Early Stopping
 #' @return a trained booster model \code{gpb.Booster}.
 #'

R-package/R/gpboost.R

@@ -23,7 +23,7 @@
 #'                     If you provide a character vector to this argument, it should contain strings with valid
 #'                     evaluation metrics.
 #'                     See \href{https://github.com/fabsig/GPBoost/blob/master/docs/Parameters.rst#metric-parameters}{
-#'                     The "metric" section}
+#'                     the "metric" section of the parameter documentation}
 #'                     for a list of valid metrics.
 #'                 }
 #'                 \item{\bold{b. function}:
@@ -61,7 +61,7 @@
 #'            \code{regression}, \code{regression_l1}, \code{huber},
 #'            \code{binary}, \code{lambdarank}, \code{multiclass}, \code{multiclass}
 #' @param params list of ("tuning") parameters. 
-#' See \href{https://github.com/fabsig/GPBoost/blob/master/docs/Parameters.rst} for more information. 
+#' See \href{https://github.com/fabsig/GPBoost/blob/master/docs/Parameters.rst}{the parameter documentation} for more information. 
 #' A few key parameters:
 #'            \itemize{
 #'                \item{\code{learning_rate}}{ The learning rate, also called shrinkage or damping parameter 

R-package/R/utils.R

@@ -80,11 +80,6 @@ gpb.call <- function(fun_name, ret, ...) {
       , PACKAGE = "lib_gpboost"
     )
   }
-  call_state <- as.integer(call_state)
-  # Check for call state value post call
-  if (call_state != 0L) {
-    gpb.last_error()
-  }
 
   return(ret)
 

R-package/src/gpboost_R.cpp

@@ -13,6 +13,10 @@
 
 #include <R_ext/Rdynload.h>
 
+#define R_NO_REMAP
+#define R_USE_C99_IN_CXX
+#include <R_ext/Error.h>
+
 #include <string>
 #include <cstdio>
 #include <cstring>
@@ -32,7 +36,7 @@
 
 #define CHECK_CALL(x) \
   if ((x) != 0) { \
-    R_INT_PTR(call_state)[0] = -1;\
+    Rf_error(LGBM_GetLastError()); \
     return call_state;\
   }
 

R-package/tests/testthat/test_dataset.R

@@ -88,6 +88,19 @@ test_that("gpb.Dataset: Dataset should be able to construct from matrix and retu
   handle <- NULL
 })
 
+test_that("cpp errors should be raised as proper R errors", {
+  data(agaricus.train, package = "gpboost")
+  train <- agaricus.train
+  dtrain <- gpb.Dataset(
+    train$data
+    , label = train$label
+    , init_score = seq_len(10L)
+  )
+  expect_error({
+    dtrain$construct()
+  }, regexp = "Initial score size doesn't match data size")
+})
+
 test_that("gpb.Dataset$setinfo() should convert 'group' to integer", {
   ds <- gpb.Dataset(
     data = matrix(rnorm(100L), nrow = 50L, ncol = 2L)