27 lines
848 B
SQL
27 lines
848 B
SQL
# name: test/parquet/parquet_long_string_stats.test
|
|
# description: Test internal issue #2289 - Performance of Parquet reader
|
|
# group: [parquet]
|
|
|
|
require httpfs
|
|
|
|
require parquet
|
|
|
|
# need to disable this otherwise we just cache everything
|
|
statement ok
|
|
set enable_external_file_cache=false;
|
|
|
|
statement ok
|
|
set parquet_metadata_cache=true;
|
|
|
|
# the constant comparison that is pushed down is longer than DuckDB's 8 bytes that are used in StringStatistics
|
|
# its prefix is equal to the max up to the last byte
|
|
# previously, we would read 5.4MB to figure out that we can prune the entire file
|
|
# now, we can prune it based on the metadata
|
|
query II
|
|
explain analyze
|
|
select count(*)
|
|
FROM 'https://github.com/duckdb/duckdb-data/releases/download/v1.0/event_baserunning_advance_attempt.parquet'
|
|
where game_id > 'WS2197109301';
|
|
----
|
|
analyzed_plan <REGEX>:.*GET: 1.*
|