summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorDouglas Rumbaugh <dbr4@psu.edu>2023-06-05 11:43:14 -0400
committerDouglas Rumbaugh <dbr4@psu.edu>2023-06-05 11:43:14 -0400
commit7dcec45b0339cd8b9b2bfab8ce1fc4c6080ea958 (patch)
treedb48d4068a29204f6f5725e43841c974cc4cc6f5
parentcd6231c89643d450f538c6063fb759b3bfcea924 (diff)
downloaddynamic-extension-7dcec45b0339cd8b9b2bfab8ce1fc4c6080ea958.tar.gz
WSS tests + bugfixes
-rw-r--r--CMakeLists.txt4
-rw-r--r--include/shard/WSS.h16
-rw-r--r--tests/wirs_tests.cpp26
-rw-r--r--tests/wss_tests.cpp346
4 files changed, 376 insertions, 16 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 1c32700..367f2e9 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -52,6 +52,10 @@ if (tests)
add_executable(memisam_tests ${CMAKE_CURRENT_SOURCE_DIR}/tests/memisam_tests.cpp)
target_link_libraries(memisam_tests PUBLIC gsl check subunit pthread)
target_include_directories(memisam_tests PRIVATE include)
+
+ add_executable(wss_tests ${CMAKE_CURRENT_SOURCE_DIR}/tests/wss_tests.cpp)
+ target_link_libraries(wss_tests PUBLIC gsl check subunit pthread)
+ target_include_directories(wss_tests PRIVATE include)
endif()
# Benchmark build instructions
diff --git a/include/shard/WSS.h b/include/shard/WSS.h
index 929f9df..bb7ee2a 100644
--- a/include/shard/WSS.h
+++ b/include/shard/WSS.h
@@ -74,7 +74,7 @@ public:
friend class WSSQuery<R, false>;
WSS(MutableBuffer<R>* buffer)
- : m_reccnt(0), m_tombstone_cnt(0), m_total_weight(0) {
+ : m_reccnt(0), m_tombstone_cnt(0), m_total_weight(0), m_alias(nullptr), m_bf(nullptr) {
size_t alloc_size = (buffer->get_record_count() * sizeof(Wrapped<R>)) + (CACHELINE_SIZE - (buffer->get_record_count() * sizeof(Wrapped<R>)) % CACHELINE_SIZE);
assert(alloc_size % CACHELINE_SIZE == 0);
@@ -117,12 +117,12 @@ public:
}
if (m_reccnt > 0) {
- build_alias_structure();
+ build_alias_structure(weights);
}
}
WSS(WSS** shards, size_t len)
- : m_reccnt(0), m_tombstone_cnt(0), m_total_weight(0) {
+ : m_reccnt(0), m_tombstone_cnt(0), m_total_weight(0), m_alias(nullptr), m_bf(nullptr) {
std::vector<Cursor<Wrapped<R>>> cursors;
cursors.reserve(len);
@@ -186,7 +186,7 @@ public:
~WSS() {
if (m_data) free(m_data);
- if (m_alias) free(m_alias);
+ if (m_alias) delete m_alias;
if (m_bf) delete m_bf;
}
@@ -253,13 +253,16 @@ private:
}
void build_alias_structure(std::vector<W> &weights) {
+
// normalize the weights vector
+ std::vector<double> norm_weights(weights.size());
+
for (size_t i=0; i<weights.size(); i++) {
- weights[i] = weights[i] / (double) m_total_weight;
+ norm_weights[i] = (double) weights[i] / (double) m_total_weight;
}
// build the alias structure
- m_alias = new Alias(weights);
+ m_alias = new Alias(norm_weights);
}
Wrapped<R>* m_data;
@@ -298,7 +301,6 @@ public:
for (size_t i = 0; i <= state->cutoff; i++) {
auto rec = buffer->get_data() + i;
weights.push_back(rec->rec.weight);
- state->records.push_back(*rec);
tot_weight += rec->rec.weight;
}
diff --git a/tests/wirs_tests.cpp b/tests/wirs_tests.cpp
index 32cb863..eab94de 100644
--- a/tests/wirs_tests.cpp
+++ b/tests/wirs_tests.cpp
@@ -142,10 +142,14 @@ START_TEST(t_wirs_query)
wirs_query_parms<WRec> parms = {lower_key, upper_key, k};
parms.rng = gsl_rng_alloc(gsl_rng_mt19937);
+ size_t total_samples = 0;
+
for (size_t i=0; i<1000; i++) {
auto state = WIRSQuery<WRec>::get_query_state(shard, &parms);
auto result = WIRSQuery<WRec>::query(shard, state, &parms);
+ total_samples += result.size();
+
for (size_t j=0; j<result.size(); j++) {
cnt[result[j].rec.key - 1]++;
}
@@ -153,9 +157,9 @@ START_TEST(t_wirs_query)
WIRSQuery<WRec>::delete_query_state(state);
}
- ck_assert(roughly_equal(cnt[0] / 1000, (double) k/4.0, k, .05));
- ck_assert(roughly_equal(cnt[1] / 1000, (double) k/4.0, k, .05));
- ck_assert(roughly_equal(cnt[2] / 1000, (double) k/2.0, k, .05));
+ ck_assert(roughly_equal(cnt[0], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[1], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[2], (double) total_samples/2.0, total_samples, .05));
gsl_rng_free(parms.rng);
delete shard;
@@ -233,10 +237,14 @@ START_TEST(t_wirs_buffer_query_scan)
wirs_query_parms<WRec> parms = {lower_key, upper_key, k};
parms.rng = gsl_rng_alloc(gsl_rng_mt19937);
+ size_t total_samples = 0;
+
for (size_t i=0; i<1000; i++) {
auto state = WIRSQuery<WRec, false>::get_buffer_query_state(buffer, &parms);
auto result = WIRSQuery<WRec, false>::buffer_query(buffer, state, &parms);
+ total_samples += result.size();
+
for (size_t j=0; j<result.size(); j++) {
cnt[result[j].rec.key - 1]++;
}
@@ -244,9 +252,9 @@ START_TEST(t_wirs_buffer_query_scan)
WIRSQuery<WRec, false>::delete_buffer_query_state(state);
}
- ck_assert(roughly_equal(cnt[0] / 1000, (double) k/4.0, k, .05));
- ck_assert(roughly_equal(cnt[1] / 1000, (double) k/4.0, k, .05));
- ck_assert(roughly_equal(cnt[2] / 1000, (double) k/2.0, k, .05));
+ ck_assert(roughly_equal(cnt[0], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[1], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[2], (double) total_samples/2.0, total_samples, .05));
gsl_rng_free(parms.rng);
delete buffer;
@@ -283,9 +291,9 @@ START_TEST(t_wirs_buffer_query_rejection)
WIRSQuery<WRec>::delete_buffer_query_state(state);
}
- ck_assert(roughly_equal(cnt[0] / total_samples, (double) k/4.0, k, .05));
- ck_assert(roughly_equal(cnt[1] / total_samples, (double) k/4.0, k, .05));
- ck_assert(roughly_equal(cnt[2] / total_samples, (double) k/2.0, k, .05));
+ ck_assert(roughly_equal(cnt[0], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[1], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[2], (double) total_samples/2.0, total_samples, .05));
gsl_rng_free(parms.rng);
delete buffer;
diff --git a/tests/wss_tests.cpp b/tests/wss_tests.cpp
new file mode 100644
index 0000000..235f664
--- /dev/null
+++ b/tests/wss_tests.cpp
@@ -0,0 +1,346 @@
+/*
+ * tests/wss_tests.cpp
+ *
+ * Unit tests for WSS (Augmented B+Tree) shard
+ *
+ * Copyright (C) 2023 Douglas Rumbaugh <drumbaugh@psu.edu>
+ * Dong Xie <dongx@psu.edu>
+ *
+ * All rights reserved. Published under the Modified BSD License.
+ *
+ */
+
+#include "shard/WSS.h"
+#include "testing.h"
+
+#include <check.h>
+
+using namespace de;
+
+typedef WSS<WRec> Shard;
+
+START_TEST(t_mbuffer_init)
+{
+ auto buffer = new MutableBuffer<WRec>(1024, true, 1024);
+ for (uint64_t i = 512; i > 0; i--) {
+ uint32_t v = i;
+ buffer->append({i,v, 1});
+ }
+
+ for (uint64_t i = 1; i <= 256; ++i) {
+ uint32_t v = i;
+ buffer->append({i, v, 1}, true);
+ }
+
+ for (uint64_t i = 257; i <= 512; ++i) {
+ uint32_t v = i + 1;
+ buffer->append({i, v, 1});
+ }
+
+ Shard* shard = new Shard(buffer);
+ ck_assert_uint_eq(shard->get_record_count(), 512);
+
+ delete buffer;
+ delete shard;
+}
+
+
+START_TEST(t_wss_init)
+{
+ size_t n = 512;
+ auto mbuffer1 = create_test_mbuffer<WRec>(n);
+ auto mbuffer2 = create_test_mbuffer<WRec>(n);
+ auto mbuffer3 = create_test_mbuffer<WRec>(n);
+
+ auto shard1 = new Shard(mbuffer1);
+ auto shard2 = new Shard(mbuffer2);
+ auto shard3 = new Shard(mbuffer3);
+
+ Shard* shards[3] = {shard1, shard2, shard3};
+ auto shard4 = new Shard(shards, 3);
+
+ ck_assert_int_eq(shard4->get_record_count(), n * 3);
+ ck_assert_int_eq(shard4->get_tombstone_count(), 0);
+
+ size_t total_cnt = 0;
+ size_t shard1_idx = 0;
+ size_t shard2_idx = 0;
+ size_t shard3_idx = 0;
+
+ for (size_t i = 0; i < shard4->get_record_count(); ++i) {
+ auto rec1 = shard1->get_record_at(shard1_idx);
+ auto rec2 = shard2->get_record_at(shard2_idx);
+ auto rec3 = shard3->get_record_at(shard3_idx);
+
+ auto cur_rec = shard4->get_record_at(i);
+
+ if (shard1_idx < n && cur_rec->rec == rec1->rec) {
+ ++shard1_idx;
+ } else if (shard2_idx < n && cur_rec->rec == rec2->rec) {
+ ++shard2_idx;
+ } else if (shard3_idx < n && cur_rec->rec == rec3->rec) {
+ ++shard3_idx;
+ } else {
+ assert(false);
+ }
+ }
+
+ delete mbuffer1;
+ delete mbuffer2;
+ delete mbuffer3;
+
+ delete shard1;
+ delete shard2;
+ delete shard3;
+ delete shard4;
+}
+
+
+START_TEST(t_full_cancelation)
+{
+ size_t n = 100;
+ auto buffer = create_double_seq_mbuffer<WRec>(n, false);
+ auto buffer_ts = create_double_seq_mbuffer<WRec>(n, true);
+
+ Shard* shard = new Shard(buffer);
+ Shard* shard_ts = new Shard(buffer_ts);
+
+ ck_assert_int_eq(shard->get_record_count(), n);
+ ck_assert_int_eq(shard->get_tombstone_count(), 0);
+ ck_assert_int_eq(shard_ts->get_record_count(), n);
+ ck_assert_int_eq(shard_ts->get_tombstone_count(), n);
+
+ Shard* shards[] = {shard, shard_ts};
+
+ Shard* merged = new Shard(shards, 2);
+
+ ck_assert_int_eq(merged->get_tombstone_count(), 0);
+ ck_assert_int_eq(merged->get_record_count(), 0);
+
+ delete buffer;
+ delete buffer_ts;
+ delete shard;
+ delete shard_ts;
+ delete merged;
+}
+END_TEST
+
+
+START_TEST(t_wss_query)
+{
+ size_t n=1000;
+ auto buffer = create_weighted_mbuffer<WRec>(n);
+
+ Shard* shard = new Shard(buffer);
+
+ size_t k = 1000;
+
+ size_t cnt[3] = {0};
+ wss_query_parms<WRec> parms = {k};
+ parms.rng = gsl_rng_alloc(gsl_rng_mt19937);
+
+ size_t total_samples = 0;
+
+ for (size_t i=0; i<1000; i++) {
+ auto state = WSSQuery<WRec>::get_query_state(shard, &parms);
+ auto result = WSSQuery<WRec>::query(shard, state, &parms);
+
+ total_samples += result.size();
+
+ for (size_t j=0; j<result.size(); j++) {
+ cnt[result[j].rec.key - 1]++;
+ }
+
+ WSSQuery<WRec>::delete_query_state(state);
+ }
+
+ ck_assert(roughly_equal(cnt[0], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[1], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[2], (double) total_samples/2.0, total_samples, .05));
+
+ gsl_rng_free(parms.rng);
+ delete shard;
+ delete buffer;
+}
+END_TEST
+
+
+template <RecordInterface R>
+std::vector<R> strip_wrapping(std::vector<Wrapped<R>> vec) {
+ std::vector<R> out(vec.size());
+ for (size_t i=0; i<vec.size(); i++) {
+ out[i] = vec[i].rec;
+ }
+
+ return out;
+}
+
+
+START_TEST(t_wss_query_merge)
+{
+ size_t n=1000;
+ auto buffer = create_weighted_mbuffer<WRec>(n);
+
+ Shard* shard = new Shard(buffer);
+
+ uint64_t lower_key = 0;
+ uint64_t upper_key = 5;
+
+ size_t k = 1000;
+
+ size_t cnt[3] = {0};
+ wss_query_parms<WRec> parms = {k};
+ parms.rng = gsl_rng_alloc(gsl_rng_mt19937);
+
+ std::vector<std::vector<WRec>> results(2);
+
+ for (size_t i=0; i<1000; i++) {
+ auto state1 = WSSQuery<WRec>::get_query_state(shard, &parms);
+ results[0] = strip_wrapping(WSSQuery<WRec>::query(shard, state1, &parms));
+
+ auto state2 = WSSQuery<WRec>::get_query_state(shard, &parms);
+ results[1] = strip_wrapping(WSSQuery<WRec>::query(shard, state2, &parms));
+
+ WSSQuery<WRec>::delete_query_state(state1);
+ WSSQuery<WRec>::delete_query_state(state2);
+ }
+
+ auto merged = WSSQuery<WRec>::merge(results);
+
+ ck_assert_int_eq(merged.size(), 2*k);
+ for (size_t i=0; i<merged.size(); i++) {
+ ck_assert_int_ge(merged[i].key, lower_key);
+ ck_assert_int_le(merged[i].key, upper_key);
+ }
+
+ gsl_rng_free(parms.rng);
+ delete shard;
+ delete buffer;
+}
+END_TEST
+
+
+START_TEST(t_wss_buffer_query_scan)
+{
+ size_t n=1000;
+ auto buffer = create_weighted_mbuffer<WRec>(n);
+
+ uint64_t lower_key = 0;
+ uint64_t upper_key = 5;
+
+ size_t k = 1000;
+
+ size_t cnt[3] = {0};
+ wss_query_parms<WRec> parms = {k};
+ parms.rng = gsl_rng_alloc(gsl_rng_mt19937);
+
+ size_t total_samples = 0;
+
+ for (size_t i=0; i<1000; i++) {
+ auto state = WSSQuery<WRec, false>::get_buffer_query_state(buffer, &parms);
+ auto result = WSSQuery<WRec, false>::buffer_query(buffer, state, &parms);
+ total_samples += result.size();
+
+ for (size_t j=0; j<result.size(); j++) {
+ cnt[result[j].rec.key - 1]++;
+ }
+
+ WSSQuery<WRec, false>::delete_buffer_query_state(state);
+ }
+
+ ck_assert(roughly_equal(cnt[0], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[1], (double) total_samples/4.0, total_samples, .05));
+ ck_assert(roughly_equal(cnt[2], (double) total_samples/2.0, total_samples, .05));
+
+ gsl_rng_free(parms.rng);
+ delete buffer;
+}
+END_TEST
+
+
+START_TEST(t_wss_buffer_query_rejection)
+{
+ size_t n=1000;
+ auto buffer = create_weighted_mbuffer<WRec>(n);
+
+ uint64_t lower_key = 0;
+ uint64_t upper_key = 5;
+
+ size_t k = 1000;
+
+ size_t cnt[3] = {0};
+ wss_query_parms<WRec> parms = {k};
+ parms.rng = gsl_rng_alloc(gsl_rng_mt19937);
+
+ size_t total_samples = 0;
+
+ for (size_t i=0; i<1000; i++) {
+ auto state = WSSQuery<WRec>::get_buffer_query_state(buffer, &parms);
+ auto result = WSSQuery<WRec>::buffer_query(buffer, state, &parms);
+
+ total_samples += result.size();
+
+ for (size_t j=0; j<result.size(); j++) {
+ cnt[result[j].rec.key - 1]++;
+ }
+
+ WSSQuery<WRec>::delete_buffer_query_state(state);
+ }
+
+ ck_assert(roughly_equal(cnt[0], (double) total_samples/4.0, total_samples, .1));
+ ck_assert(roughly_equal(cnt[1], (double) total_samples/4.0, total_samples, .1));
+ ck_assert(roughly_equal(cnt[2], (double) total_samples/2.0, total_samples, .1));
+
+ gsl_rng_free(parms.rng);
+ delete buffer;
+}
+END_TEST
+
+
+Suite *unit_testing()
+{
+ Suite *unit = suite_create("WSS Shard Unit Testing");
+
+ TCase *create = tcase_create("de::WSS constructor Testing");
+ tcase_add_test(create, t_mbuffer_init);
+ tcase_add_test(create, t_wss_init);
+ tcase_set_timeout(create, 100);
+ suite_add_tcase(unit, create);
+
+
+ TCase *tombstone = tcase_create("de:WSS::tombstone cancellation Testing");
+ tcase_add_test(tombstone, t_full_cancelation);
+ suite_add_tcase(unit, tombstone);
+
+
+ TCase *sampling = tcase_create("de:WSS::WSSQuery Testing");
+ tcase_add_test(sampling, t_wss_query);
+ tcase_add_test(sampling, t_wss_query_merge);
+ tcase_add_test(sampling, t_wss_buffer_query_rejection);
+ tcase_add_test(sampling, t_wss_buffer_query_scan);
+ suite_add_tcase(unit, sampling);
+
+ return unit;
+}
+
+
+int shard_unit_tests()
+{
+ int failed = 0;
+ Suite *unit = unit_testing();
+ SRunner *unit_shardner = srunner_create(unit);
+
+ srunner_run_all(unit_shardner, CK_NORMAL);
+ failed = srunner_ntests_failed(unit_shardner);
+ srunner_free(unit_shardner);
+
+ return failed;
+}
+
+
+int main()
+{
+ int unit_failed = shard_unit_tests();
+
+ return (unit_failed == 0) ? EXIT_SUCCESS : EXIT_FAILURE;
+}