HappenLee commented on a change in pull request #8051:
URL: https://github.com/apache/incubator-doris/pull/8051#discussion_r806556854
##########
File path: be/src/vec/exec/join/vhash_join_node.cpp
##########
@@ -168,89 +177,136 @@ struct ProcessHashTableProbe {
// the output block struct is same with mutable block. we can do more opt
on it and simplify
// the logic of probe
// TODO: opt the visited here to reduce the size of hash table
+ template <class JoinOpType>
Status do_process(HashTableContext& hash_table_ctx, ConstNullMapPtr
null_map,
MutableBlock& mutable_block, Block* output_block) {
using KeyGetter = typename HashTableContext::State;
using Mapped = typename HashTableContext::Mapped;
KeyGetter key_getter(_probe_raw_ptrs, _join_node->_probe_key_sz,
nullptr);
-
+
std::vector<uint32_t> items_counts(_probe_rows);
auto& mcol = mutable_block.mutable_columns();
-
- int right_col_idx = _join_node->_is_right_semi_anti ? 0 :
_left_table_data_types.size();
- int right_col_len = _right_table_data_types.size();
int current_offset = 0;
for (; _probe_index < _probe_rows;) {
- // ignore null rows
if constexpr (ignore_null) {
if ((*null_map)[_probe_index]) {
items_counts[_probe_index++] = 0;
continue;
}
}
-
int repeat_count = 0;
- auto find_result =
- (*null_map)[_probe_index]
+ if constexpr (JoinOpType::value == TJoinOp::INNER_JOIN) {
+ if (!(*null_map)[_probe_index]) {
+ auto find_result =
key_getter.find_key(hash_table_ctx.hash_table, _probe_index, _arena);
+
+ if (find_result.is_found()) {
+ auto& mapped = find_result.get_mapped();
+
+ // TODO: Iterators are currently considered to be a
heavy operation and have a certain impact on performance.
+ // We should rethink whether to use this iterator mode
in the future. Now just opt the one row case
+ if (mapped.get_row_count() == 1) {
+ mapped.visited = true;
Review comment:
no need set visited here
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
