Copilot commented on code in PR #66416: URL: https://github.com/apache/doris/pull/66416#discussion_r3717610703
########## regression-test/suites/point_query_p0/test_point_query_list_partition.groovy: ########## @@ -0,0 +1,114 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +// Regression test for issue #66030: a short-circuit point query on a table +// partitioned by LIST used to throw IllegalStateException because the tablet +// list was not pruned to a single tablet when the matched partition contained +// more than one tablet. The legacy distribution-prune path taken for point +// queries could not prune (no column filters for the nereids-planned conjuncts), +// so PointQueryExecutor.setScanRangeLocations hit its single-tablet checkState +// with an un-pruned list. With multiple buckets per partition the bug surfaces. +// +// This suite also covers the prepared-statement point query on the same table +// to guard against regressing that path: a prepared point query (parameter +// values unknown at planning time) must keep taking the legacy runtime-prune +// path and must not throw. + +suite("test_point_query_list_partition") { + def user = context.config.jdbcUser + def password = context.config.jdbcPassword + def realDb = "regression_test_serving_p0" + def tableName = realDb + ".tbl_point_query_list_partition" + sql "CREATE DATABASE IF NOT EXISTS ${realDb}" + + // Parse the JDBC url so we can build a server-side-prepared-statement url. + String jdbcUrl = context.config.jdbcUrl + String urlWithoutSchema = jdbcUrl.substring(jdbcUrl.indexOf("://") + 3) + def sql_ip = urlWithoutSchema.substring(0, urlWithoutSchema.indexOf(":")) + def sql_port + if (urlWithoutSchema.indexOf("/") >= 0) { + sql_port = urlWithoutSchema.substring(urlWithoutSchema.indexOf(":") + 1, urlWithoutSchema.indexOf("/")) + } else { + sql_port = urlWithoutSchema.substring(urlWithoutSchema.indexOf(":") + 1) + } + def prepare_url = "jdbc:mysql://" + sql_ip + ":" + sql_port + "/" + realDb + "?&useServerPrepStmts=true" + + sql """DROP TABLE IF EXISTS ${tableName}""" + sql """ + CREATE TABLE IF NOT EXISTS ${tableName} ( + `pk` varchar(64) NOT NULL, + `_id` bigint NOT NULL + ) ENGINE=OLAP + UNIQUE KEY(`pk`, `_id`) + PARTITION BY LIST(`pk`) ( + PARTITION `p_abcd` VALUES IN ("abcd"), + PARTITION `p_efgh` VALUES IN ("efgh") + ) + DISTRIBUTED BY HASH(`pk`, `_id`) BUCKETS 3 + PROPERTIES ( + "replication_allocation" = "tag.location.default: 1", + "store_row_column" = "true", + "enable_unique_key_merge_on_write" = "true", + "light_schema_change" = "true", + "storage_format" = "V2") + """ + + sql """INSERT INTO ${tableName} VALUES ('abcd', 1)""" + sql """INSERT INTO ${tableName} VALUES ('abcd', 2)""" + sql """INSERT INTO ${tableName} VALUES ('efgh', 1)""" + + // --- Direct point query (the path fixed by #66030) --- + // The point query covers the full unique key, so it must be planned as a + // short-circuit point query. Confirm the short-circuit path is taken so + // this test actually exercises the buggy code path. + def explain = sql """EXPLAIN SELECT * FROM ${tableName} WHERE pk = 'abcd' AND _id = 1""" + assertTrue(explain.toString().contains("SHORT-CIRCUIT")) + + // Before the fix this threw: (1105, 'IllegalStateException, msg: null'). + // The point query must succeed and return exactly the matched row. + def result1 = sql """SELECT * FROM ${tableName} WHERE pk = 'abcd' AND _id = 1""" + assertEquals(1, result1.size()) + + // Point query hitting a different LIST partition must also succeed. + def result2 = sql """SELECT * FROM ${tableName} WHERE pk = 'efgh' AND _id = 1""" + assertEquals(1, result2.size()) + + // A point query that matches no row should return zero rows, not throw. + def result3 = sql """SELECT * FROM ${tableName} WHERE pk = 'abcd' AND _id = 999""" + assertEquals(0, result3.size()) + + // --- Prepared-statement point query (the path that must NOT regress) --- + // A prepared point query cannot be pruned to a single tablet at planning + // time (parameter value unknown), so it keeps taking the legacy runtime + // distribution-prune path. It must still resolve to the right tablet at + // execution time and succeed. + connect(user, password, prepare_url) { + def stmt = prepareStatement "select * from ${tableName} where pk = ? and _id = ?" + stmt.setString(1, "abcd") + stmt.setLong(2, 1) + def rs = stmt.executeQuery() + int rowCount = 0 + while (rs.next()) { + assertEquals("abcd", rs.getString(1)) + assertEquals(1L, rs.getLong(2)) + rowCount++ + } + assertEquals(1, rowCount) + } + + sql """DROP TABLE IF EXISTS ${tableName}""" Review Comment: This suite drops the table at the end, but other point_query_p0 suites typically keep the table around and rely on `DROP TABLE IF EXISTS` at the start (to preserve state for debugging when a case fails). Since this suite already drops before creating, consider removing the final drop. ########## regression-test/suites/point_query_p0/test_point_query_list_partition.groovy: ########## @@ -0,0 +1,114 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +// Regression test for issue #66030: a short-circuit point query on a table +// partitioned by LIST used to throw IllegalStateException because the tablet +// list was not pruned to a single tablet when the matched partition contained +// more than one tablet. The legacy distribution-prune path taken for point +// queries could not prune (no column filters for the nereids-planned conjuncts), +// so PointQueryExecutor.setScanRangeLocations hit its single-tablet checkState +// with an un-pruned list. With multiple buckets per partition the bug surfaces. +// +// This suite also covers the prepared-statement point query on the same table +// to guard against regressing that path: a prepared point query (parameter +// values unknown at planning time) must keep taking the legacy runtime-prune +// path and must not throw. + +suite("test_point_query_list_partition") { + def user = context.config.jdbcUser + def password = context.config.jdbcPassword + def realDb = "regression_test_serving_p0" + def tableName = realDb + ".tbl_point_query_list_partition" + sql "CREATE DATABASE IF NOT EXISTS ${realDb}" + + // Parse the JDBC url so we can build a server-side-prepared-statement url. + String jdbcUrl = context.config.jdbcUrl + String urlWithoutSchema = jdbcUrl.substring(jdbcUrl.indexOf("://") + 3) + def sql_ip = urlWithoutSchema.substring(0, urlWithoutSchema.indexOf(":")) + def sql_port + if (urlWithoutSchema.indexOf("/") >= 0) { + sql_port = urlWithoutSchema.substring(urlWithoutSchema.indexOf(":") + 1, urlWithoutSchema.indexOf("/")) + } else { + sql_port = urlWithoutSchema.substring(urlWithoutSchema.indexOf(":") + 1) + } + def prepare_url = "jdbc:mysql://" + sql_ip + ":" + sql_port + "/" + realDb + "?&useServerPrepStmts=true" + + sql """DROP TABLE IF EXISTS ${tableName}""" + sql """ + CREATE TABLE IF NOT EXISTS ${tableName} ( + `pk` varchar(64) NOT NULL, + `_id` bigint NOT NULL + ) ENGINE=OLAP + UNIQUE KEY(`pk`, `_id`) + PARTITION BY LIST(`pk`) ( + PARTITION `p_abcd` VALUES IN ("abcd"), + PARTITION `p_efgh` VALUES IN ("efgh") + ) + DISTRIBUTED BY HASH(`pk`, `_id`) BUCKETS 3 + PROPERTIES ( + "replication_allocation" = "tag.location.default: 1", + "store_row_column" = "true", + "enable_unique_key_merge_on_write" = "true", + "light_schema_change" = "true", + "storage_format" = "V2") + """ + + sql """INSERT INTO ${tableName} VALUES ('abcd', 1)""" + sql """INSERT INTO ${tableName} VALUES ('abcd', 2)""" + sql """INSERT INTO ${tableName} VALUES ('efgh', 1)""" + + // --- Direct point query (the path fixed by #66030) --- + // The point query covers the full unique key, so it must be planned as a + // short-circuit point query. Confirm the short-circuit path is taken so + // this test actually exercises the buggy code path. + def explain = sql """EXPLAIN SELECT * FROM ${tableName} WHERE pk = 'abcd' AND _id = 1""" + assertTrue(explain.toString().contains("SHORT-CIRCUIT")) Review Comment: The regression framework already provides an `explain { ... contains ... }` helper (used in other point_query_p0 suites). Using `sql "EXPLAIN ..."` and then `toString().contains(...)` is more brittle because it depends on the result object's string formatting rather than the explain output. -- 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]
