github-actions[bot] commented on code in PR #69723:
URL: https://github.com/apache/doris/pull/69723#discussion_r4398580838
##########
be/src/exec/operator/file_scan_operator.cpp:
##########
@@ -247,41 +140,12 @@ Status
FileScanLocalState::_init_scanners(std::list<ScannerSPtr>* scanners) {
scan_params = _split_source->get_params();
}
RETURN_IF_ERROR(validate_file_scan_support(*scan_params));
- const bool is_load =
-
state()->desc_tbl().get_tuple_descriptor(scan_params->src_tuple_id) != nullptr;
- // TODO: Use scanner v2 for all queries.
- const bool use_file_scanner_v2 =
- should_use_file_scanner_v2(state()->query_options(), is_load,
*scan_params);
- _operator_profile->add_info_string("UseScannerV2", use_file_scanner_v2 ?
"true" : "false");
- const auto* output_tuple_desc =
state()->desc_tbl().get_tuple_descriptor(_output_tuple_id);
- DORIS_CHECK(output_tuple_desc != nullptr);
- const bool metadata_only_count =
- is_count_star_pushdown() &&
_split_source->all_ranges_have_table_level_row_count();
- if (!is_load && !use_file_scanner_v2 && !metadata_only_count &&
- std::ranges::any_of(output_tuple_desc->slots(), [](const
SlotDescriptor* slot) {
- return contains_variant_type(slot->type());
- })) {
- // A syntactic COUNT(*) alone is insufficient: every assigned range
must prove that the
- // legacy scanner will emit metadata counts without decoding a Variant
carrier.
- return Status::NotSupported(
- "External VARIANT columns require FileScannerV2; the legacy
file scanner does "
- "not support VARIANT");
- }
- if (use_file_scanner_v2) {
- _parquet_profile_cache =
-
std::make_unique<format::parquet::ParquetProfileCache>(_scanner_profile.get());
- }
+ _parquet_profile_cache =
+
std::make_unique<format::parquet::ParquetProfileCache>(_scanner_profile.get());
for (int i = 0; i < _max_scanners; ++i) {
- ScannerSPtr scanner;
- if (use_file_scanner_v2) {
- scanner = FileScannerV2::create_shared(state(), this, p._limit,
_split_source,
- _scanner_profile.get(),
p._kv_cache.get(),
- &p._colname_to_slot_id);
- } else {
- scanner = FileScanner::create_shared(state(), this, p._limit,
_split_source,
- _scanner_profile.get(),
p._kv_cache.get(),
- &p._colname_to_slot_id);
- }
+ ScannerSPtr scanner = FileScannerV2::create_shared(
Review Comment:
[P2] Preserve ORC batch limits before forcing V2. Ordinary Hive ORC scans
with enable_file_scanner_v2=false previously used V1, whose
OrcReader::set_batch_size rebuilt the ORC batch at the scanner's requested cap.
V2's OrcReader inherits the no-op FileReader::set_batch_size and always calls
createRowBatch(4096), so this unconditional switch removes both the session
batch limit and adaptive memory control for those scans. For example,
batch_size=32 over 64-KiB strings can now decode about 256 MiB per read instead
of 2 MiB, causing memory-limit failures that lowering the batch size can no
longer avoid. Make V2 ORC apply the supplied batch cap before removing the
bounded path.
--
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]
