dataroaring commented on code in PR #37960:
URL: https://github.com/apache/doris/pull/37960#discussion_r1680772973
##########
be/src/olap/rowset/vertical_beta_rowset_writer.cpp:
##########
@@ -82,37 +82,39 @@ Status VerticalBetaRowsetWriter<T>::add_columns(const
vectorized::Block* block,
}
RETURN_IF_ERROR(_segment_writers[_cur_writer_idx]->append_block(block,
0, num_rows));
} else {
- // value columns
- uint32_t num_rows_written =
_segment_writers[_cur_writer_idx]->num_rows_written();
- VLOG_NOTICE << "num_rows_written: " << num_rows_written
- << ", _cur_writer_idx: " << _cur_writer_idx;
- uint32_t num_rows_key_group =
_segment_writers[_cur_writer_idx]->row_count();
- // init if it's first value column write in current segment
- if (_cur_writer_idx == 0 && num_rows_written == 0) {
- VLOG_NOTICE << "init first value column segment writer";
- RETURN_IF_ERROR(_segment_writers[_cur_writer_idx]->init(col_ids,
is_key));
- }
- // when splitting segment, need to make rows align between key columns
and value columns
size_t start_offset = 0;
size_t limit = num_rows;
- if (num_rows_written + num_rows >= num_rows_key_group &&
- _cur_writer_idx < _segment_writers.size() - 1) {
- RETURN_IF_ERROR(_segment_writers[_cur_writer_idx]->append_block(
- block, 0, num_rows_key_group - num_rows_written));
-
RETURN_IF_ERROR(_flush_columns(_segment_writers[_cur_writer_idx].get()));
- start_offset = num_rows_key_group - num_rows_written;
- limit = num_rows - start_offset;
- ++_cur_writer_idx;
- // switch to next writer
- RETURN_IF_ERROR(_segment_writers[_cur_writer_idx]->init(col_ids,
is_key));
- num_rows_written = 0;
- num_rows_key_group =
_segment_writers[_cur_writer_idx]->row_count();
- }
- if (limit > 0) {
- RETURN_IF_ERROR(
- _segment_writers[_cur_writer_idx]->append_block(block,
start_offset, limit));
- DCHECK(_segment_writers[_cur_writer_idx]->num_rows_written() <=
- _segment_writers[_cur_writer_idx]->row_count());
+ while (limit > 0) {
+ // value columns
+ uint32_t num_rows_written =
_segment_writers[_cur_writer_idx]->num_rows_written();
+ VLOG_NOTICE << "num_rows_written: " << num_rows_written
+ << ", _cur_writer_idx: " << _cur_writer_idx;
+ uint32_t num_rows_key_group =
_segment_writers[_cur_writer_idx]->row_count();
+ // init if it's first value column write in current segment
+ if (_cur_writer_idx == 0 && num_rows_written == 0) {
+ VLOG_NOTICE << "init first value column segment writer";
+
RETURN_IF_ERROR(_segment_writers[_cur_writer_idx]->init(col_ids, is_key));
+ }
+ // when splitting segment, need to make rows align between key
columns and value columns
+ if (num_rows_written + limit >= num_rows_key_group &&
+ _cur_writer_idx < _segment_writers.size() - 1) {
Review Comment:
to_written = num_rows_written + limit >= num_rows_key_group ?
num_rows_key_group - num_rows_written : limit;
_segment_writers[_cur_writer_idx]->append_block(block, start_offset,
to_written);
start_offset += to_written;
limit -= to_written;
if (limit > 0) {
++_cur_writer_idx;
// switch to next writer
RETURN_IF_ERROR(_segment_writers[_cur_writer_idx]->init(col_ids, is_key));
num_rows_written = 0;
num_rows_key_group =
_segment_writers[_cur_writer_idx]->row_count();
}
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
