From 0a5eba11839d129f2c0e0b4dbdfa9a656f8ca22b Mon Sep 17 00:00:00 2001
From: Charles Hastings <charlesh@nvidia.com>
Date: Thu, 29 Aug 2024 12:08:33 -0700
Subject: [PATCH] use new test tarball in s3 when running C++ unit tests

---
 ci/test_cpp.sh            | 7 +++----
 datasets/get_test_data.sh | 8 ++++++++
 2 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/ci/test_cpp.sh b/ci/test_cpp.sh
index 012ad8b741d..e2043408656 100755
--- a/ci/test_cpp.sh
+++ b/ci/test_cpp.sh
@@ -37,10 +37,9 @@ nvidia-smi
 
 # RAPIDS_DATASET_ROOT_DIR is used by test scripts
 export RAPIDS_DATASET_ROOT_DIR="$(realpath datasets)"
-
-echo "DEBUGGING, ${RAPIDS_DATASET_ROOT_DIR}"
-ls -l $RAPIDS_DATASET_ROOT_DIR
-echo "END DEBUGGING"
+pushd "${RAPIDS_DATASET_ROOT_DIR}"
+ ./get_test_data.sh --cpp_ci_subset
+popd
 
 export GTEST_OUTPUT=xml:${RAPIDS_TESTS_DIR}/
 
diff --git a/datasets/get_test_data.sh b/datasets/get_test_data.sh
index eea789ef3e3..6778166ab6e 100755
--- a/datasets/get_test_data.sh
+++ b/datasets/get_test_data.sh
@@ -27,6 +27,12 @@ cd "$( cd "$( dirname "$(realpath -m "${BASH_SOURCE[0]}")" )" && pwd )";
 #
 # FIXME: some test data needs to be extracted to "benchmarks", which is
 # confusing now that there's dedicated datasets for benchmarks.
+CPP_CI_DATASET_DATA="
+# ~10s download
+https://data.rapids.ai/cugraph/test/cpp_ci_datasets.tgz
+test
+"
+
 BASE_DATASET_DATA="
 # ~22s download
 https://data.rapids.ai/cugraph/test/datasets.tgz
@@ -89,6 +95,8 @@ if hasArg "--benchmark"; then
     DATASET_DATA="${BENCHMARK_DATASET_DATA}"
 elif hasArg "--subset"; then
     DATASET_DATA="${BASE_DATASET_DATA}"
+elif hasArg "--cpp_ci_subset"; then
+    DATASET_DATA="${CPP_CI_DATASET_DATA}"
 elif hasArg "--self_loops"; then
     DATASET_DATA="${SELF_LOOPS_DATASET_DATA}"
 # Do not include benchmark datasets by default - too big