[SPARK-20543][SPARKR] skip tests when running on CRAN

## What changes were proposed in this pull request?

General rule on skip or not:
skip if
- RDD tests
- tests could run long or complicated (streaming, hivecontext)
- tests on error conditions
- tests won't likely change/break

## How was this patch tested?

unit tests, `R CMD check --as-cran`, `R CMD check`

Author: Felix Cheung <[email protected]>

Closes apache#17817 from felixcheung/rskiptest.

R/pkg/inst/tests/testthat/test_Serde.R

@@ -20,6 +20,8 @@ context("SerDe functionality")
 sparkSession <- sparkR.session(enableHiveSupport = FALSE)
 
 test_that("SerDe of primitive types", {
+  skip_on_cran()
+
   x <- callJStatic("SparkRHandler", "echo", 1L)
   expect_equal(x, 1L)
   expect_equal(class(x), "integer")
@@ -38,6 +40,8 @@ test_that("SerDe of primitive types", {
 })
 
 test_that("SerDe of list of primitive types", {
+  skip_on_cran()
+
   x <- list(1L, 2L, 3L)
   y <- callJStatic("SparkRHandler", "echo", x)
   expect_equal(x, y)
@@ -65,6 +69,8 @@ test_that("SerDe of list of primitive types", {
 })
 
 test_that("SerDe of list of lists", {
+  skip_on_cran()
+
   x <- list(list(1L, 2L, 3L), list(1, 2, 3),
             list(TRUE, FALSE), list("a", "b", "c"))
   y <- callJStatic("SparkRHandler", "echo", x)

R/pkg/inst/tests/testthat/test_Windows.R

@@ -17,6 +17,8 @@
 context("Windows-specific tests")
 
 test_that("sparkJars tag in SparkContext", {
+  skip_on_cran()
+
   if (.Platform$OS.type != "windows") {
     skip("This test is only for Windows, skipped")
   }

R/pkg/inst/tests/testthat/test_binaryFile.R

@@ -24,6 +24,8 @@ sc <- callJStatic("org.apache.spark.sql.api.r.SQLUtils", "getJavaSparkContext",
 mockFile <- c("Spark is pretty.", "Spark is awesome.")
 
 test_that("saveAsObjectFile()/objectFile() following textFile() works", {
+  skip_on_cran()
+
   fileName1 <- tempfile(pattern = "spark-test", fileext = ".tmp")
   fileName2 <- tempfile(pattern = "spark-test", fileext = ".tmp")
   writeLines(mockFile, fileName1)
@@ -38,6 +40,8 @@ test_that("saveAsObjectFile()/objectFile() following textFile() works", {
 })
 
 test_that("saveAsObjectFile()/objectFile() works on a parallelized list", {
+  skip_on_cran()
+
   fileName <- tempfile(pattern = "spark-test", fileext = ".tmp")
 
   l <- list(1, 2, 3)
@@ -50,6 +54,8 @@ test_that("saveAsObjectFile()/objectFile() works on a parallelized list", {
 })
 
 test_that("saveAsObjectFile()/objectFile() following RDD transformations works", {
+  skip_on_cran()
+
   fileName1 <- tempfile(pattern = "spark-test", fileext = ".tmp")
   fileName2 <- tempfile(pattern = "spark-test", fileext = ".tmp")
   writeLines(mockFile, fileName1)
@@ -74,6 +80,8 @@ test_that("saveAsObjectFile()/objectFile() following RDD transformations works",
 })
 
 test_that("saveAsObjectFile()/objectFile() works with multiple paths", {
+  skip_on_cran()
+
   fileName1 <- tempfile(pattern = "spark-test", fileext = ".tmp")
   fileName2 <- tempfile(pattern = "spark-test", fileext = ".tmp")
 

R/pkg/inst/tests/testthat/test_binary_function.R

@@ -29,6 +29,8 @@ rdd <- parallelize(sc, nums, 2L)
 mockFile <- c("Spark is pretty.", "Spark is awesome.")
 
 test_that("union on two RDDs", {
+  skip_on_cran()
+
   actual <- collectRDD(unionRDD(rdd, rdd))
   expect_equal(actual, as.list(rep(nums, 2)))
 
@@ -51,6 +53,8 @@ test_that("union on two RDDs", {
 })
 
 test_that("cogroup on two RDDs", {
+  skip_on_cran()
+
   rdd1 <- parallelize(sc, list(list(1, 1), list(2, 4)))
   rdd2 <- parallelize(sc, list(list(1, 2), list(1, 3)))
   cogroup.rdd <- cogroup(rdd1, rdd2, numPartitions = 2L)
@@ -69,6 +73,8 @@ test_that("cogroup on two RDDs", {
 })
 
 test_that("zipPartitions() on RDDs", {
+  skip_on_cran()
+
   rdd1 <- parallelize(sc, 1:2, 2L)  # 1, 2
   rdd2 <- parallelize(sc, 1:4, 2L)  # 1:2, 3:4
   rdd3 <- parallelize(sc, 1:6, 2L)  # 1:3, 4:6

R/pkg/inst/tests/testthat/test_broadcast.R

@@ -26,6 +26,8 @@ nums <- 1:2
 rrdd <- parallelize(sc, nums, 2L)
 
 test_that("using broadcast variable", {
+  skip_on_cran()
+
   randomMat <- matrix(nrow = 10, ncol = 10, data = rnorm(100))
   randomMatBr <- broadcast(sc, randomMat)
 
@@ -38,6 +40,8 @@ test_that("using broadcast variable", {
 })
 
 test_that("without using broadcast variable", {
+  skip_on_cran()
+
   randomMat <- matrix(nrow = 10, ncol = 10, data = rnorm(100))
 
   useBroadcast <- function(x) {

R/pkg/inst/tests/testthat/test_client.R

@@ -18,6 +18,8 @@
 context("functions in client.R")
 
 test_that("adding spark-testing-base as a package works", {
+  skip_on_cran()
+
   args <- generateSparkSubmitArgs("", "", "", "",
                                   "holdenk:spark-testing-base:1.3.0_0.0.5")
   expect_equal(gsub("[[:space:]]", "", args),
@@ -26,16 +28,22 @@ test_that("adding spark-testing-base as a package works", {
 })
 
 test_that("no package specified doesn't add packages flag", {
+  skip_on_cran()
+
   args <- generateSparkSubmitArgs("", "", "", "", "")
   expect_equal(gsub("[[:space:]]", "", args),
                "")
 })
 
 test_that("multiple packages don't produce a warning", {
+  skip_on_cran()
+
   expect_warning(generateSparkSubmitArgs("", "", "", "", c("A", "B")), NA)
 })
 
 test_that("sparkJars sparkPackages as character vectors", {
+  skip_on_cran()
+
   args <- generateSparkSubmitArgs("", "", c("one.jar", "two.jar", "three.jar"), "",
                                   c("com.databricks:spark-avro_2.10:2.0.1"))
   expect_match(args, "--jars one.jar,two.jar,three.jar")

R/pkg/inst/tests/testthat/test_context.R

@@ -18,6 +18,8 @@
 context("test functions in sparkR.R")
 
 test_that("Check masked functions", {
+  skip_on_cran()
+
   # Check that we are not masking any new function from base, stats, testthat unexpectedly
   # NOTE: We should avoid adding entries to *namesOfMaskedCompletely* as masked functions make it
   # hard for users to use base R functions. Please check when in doubt.
@@ -55,6 +57,8 @@ test_that("Check masked functions", {
 })
 
 test_that("repeatedly starting and stopping SparkR", {
+  skip_on_cran()
+
   for (i in 1:4) {
     sc <- suppressWarnings(sparkR.init())
     rdd <- parallelize(sc, 1:20, 2L)
@@ -73,6 +77,8 @@ test_that("repeatedly starting and stopping SparkSession", {
 })
 
 test_that("rdd GC across sparkR.stop", {
+  skip_on_cran()
+
   sc <- sparkR.sparkContext() # sc should get id 0
   rdd1 <- parallelize(sc, 1:20, 2L) # rdd1 should get id 1
   rdd2 <- parallelize(sc, 1:10, 2L) # rdd2 should get id 2
@@ -96,6 +102,8 @@ test_that("rdd GC across sparkR.stop", {
 })
 
 test_that("job group functions can be called", {
+  skip_on_cran()
+
   sc <- sparkR.sparkContext()
   setJobGroup("groupId", "job description", TRUE)
   cancelJobGroup("groupId")
@@ -108,12 +116,16 @@ test_that("job group functions can be called", {
 })
 
 test_that("utility function can be called", {
+  skip_on_cran()
+
   sparkR.sparkContext()
   setLogLevel("ERROR")
   sparkR.session.stop()
 })
 
 test_that("getClientModeSparkSubmitOpts() returns spark-submit args from whitelist", {
+  skip_on_cran()
+
   e <- new.env()
   e[["spark.driver.memory"]] <- "512m"
   ops <- getClientModeSparkSubmitOpts("sparkrmain", e)
@@ -141,6 +153,8 @@ test_that("getClientModeSparkSubmitOpts() returns spark-submit args from whiteli
 })
 
 test_that("sparkJars sparkPackages as comma-separated strings", {
+  skip_on_cran()
+
   expect_warning(processSparkJars(" a, b "))
   jars <- suppressWarnings(processSparkJars(" a, b "))
   expect_equal(lapply(jars, basename), list("a", "b"))
@@ -168,6 +182,8 @@ test_that("spark.lapply should perform simple transforms", {
 })
 
 test_that("add and get file to be downloaded with Spark job on every node", {
+  skip_on_cran()
+
   sparkR.sparkContext()
   # Test add file.
   path <- tempfile(pattern = "hello", fileext = ".txt")

R/pkg/inst/tests/testthat/test_includePackage.R

@@ -26,6 +26,8 @@ nums <- 1:2
 rdd <- parallelize(sc, nums, 2L)
 
 test_that("include inside function", {
+  skip_on_cran()
+
   # Only run the test if plyr is installed.
   if ("plyr" %in% rownames(installed.packages())) {
     suppressPackageStartupMessages(library(plyr))
@@ -42,6 +44,8 @@ test_that("include inside function", {
 })
 
 test_that("use include package", {
+  skip_on_cran()
+
   # Only run the test if plyr is installed.
   if ("plyr" %in% rownames(installed.packages())) {
     suppressPackageStartupMessages(library(plyr))

R/pkg/inst/tests/testthat/test_mllib_clustering.R

@@ -255,6 +255,8 @@ test_that("spark.lda with libsvm", {
 })
 
 test_that("spark.lda with text input", {
+  skip_on_cran()
+
   text <- read.text(absoluteSparkPath("data/mllib/sample_lda_data.txt"))
   model <- spark.lda(text, optimizer = "online", features = "value")
 
@@ -297,6 +299,8 @@ test_that("spark.lda with text input", {
 })
 
 test_that("spark.posterior and spark.perplexity", {
+  skip_on_cran()
+
   text <- read.text(absoluteSparkPath("data/mllib/sample_lda_data.txt"))
   model <- spark.lda(text, features = "value", k = 3)
 

R/pkg/inst/tests/testthat/test_mllib_regression.R

@@ -23,6 +23,8 @@ context("MLlib regression algorithms, except for tree-based algorithms")
 sparkSession <- sparkR.session(enableHiveSupport = FALSE)
 
 test_that("formula of spark.glm", {
+  skip_on_cran()
+
   training <- suppressWarnings(createDataFrame(iris))
   # directly calling the spark API
   # dot minus and intercept vs native glm
@@ -195,6 +197,8 @@ test_that("spark.glm summary", {
 })
 
 test_that("spark.glm save/load", {
+  skip_on_cran()
+
   training <- suppressWarnings(createDataFrame(iris))
   m <- spark.glm(training, Sepal_Width ~ Sepal_Length + Species)
   s <- summary(m)
@@ -222,6 +226,8 @@ test_that("spark.glm save/load", {
 })
 
 test_that("formula of glm", {
+  skip_on_cran()
+
   training <- suppressWarnings(createDataFrame(iris))
   # dot minus and intercept vs native glm
   model <- glm(Sepal_Width ~ . - Species + 0, data = training)
@@ -248,6 +254,8 @@ test_that("formula of glm", {
 })
 
 test_that("glm and predict", {
+  skip_on_cran()
+
   training <- suppressWarnings(createDataFrame(iris))
   # gaussian family
   model <- glm(Sepal_Width ~ Sepal_Length + Species, data = training)
@@ -292,6 +300,8 @@ test_that("glm and predict", {
 })
 
 test_that("glm summary", {
+  skip_on_cran()
+
   # gaussian family
   training <- suppressWarnings(createDataFrame(iris))
   stats <- summary(glm(Sepal_Width ~ Sepal_Length + Species, data = training))
@@ -341,6 +351,8 @@ test_that("glm summary", {
 })
 
 test_that("glm save/load", {
+  skip_on_cran()
+
   training <- suppressWarnings(createDataFrame(iris))
   m <- glm(Sepal_Width ~ Sepal_Length + Species, data = training)
   s <- summary(m)

R/pkg/inst/tests/testthat/test_parallelize_collect.R

@@ -39,6 +39,8 @@ jsc <- callJStatic("org.apache.spark.sql.api.r.SQLUtils", "getJavaSparkContext",
 # Tests
 
 test_that("parallelize() on simple vectors and lists returns an RDD", {
+  skip_on_cran()
+
   numVectorRDD <- parallelize(jsc, numVector, 1)
   numVectorRDD2 <- parallelize(jsc, numVector, 10)
   numListRDD <- parallelize(jsc, numList, 1)
@@ -66,6 +68,8 @@ test_that("parallelize() on simple vectors and lists returns an RDD", {
 })
 
 test_that("collect(), following a parallelize(), gives back the original collections", {
+  skip_on_cran()
+
   numVectorRDD <- parallelize(jsc, numVector, 10)
   expect_equal(collectRDD(numVectorRDD), as.list(numVector))
 
@@ -86,6 +90,8 @@ test_that("collect(), following a parallelize(), gives back the original collect
 })
 
 test_that("regression: collect() following a parallelize() does not drop elements", {
+  skip_on_cran()
+
   # 10 %/% 6 = 1, ceiling(10 / 6) = 2
   collLen <- 10
   numPart <- 6
@@ -95,6 +101,8 @@ test_that("regression: collect() following a parallelize() does not drop element
 })
 
 test_that("parallelize() and collect() work for lists of pairs (pairwise data)", {
+  skip_on_cran()
+
   # use the pairwise logical to indicate pairwise data
   numPairsRDDD1 <- parallelize(jsc, numPairs, 1)
   numPairsRDDD2 <- parallelize(jsc, numPairs, 2)