Release notes for 362

[martint]

Ashhar Hasan

• all checked
• 08-28 22:50:33 89e61ec Fix failing test due to missing comma
• 08-31 04:50:54 60e0337 Remove obsolete test
• 08-31 04:50:54 ab46771 Fix comment to mention memory connector instead of Cassandra
• 08-31 04:50:54 b7f4c81 Change assertions to explicitly check for existence of table
• 09-01 12:17:45 01d38b2 Apply overrides from provided config when running SuiteDescribe
• 09-03 11:52:02 00281a3 Add SQL support section to TPC-DS connector
• 09-03 11:54:28 90f6286 Move SQL support section
• 09-03 11:54:28 d254b7f Correct scope of memory release
• 09-07 23:03:37 41e596a Add SQL support section to Kudu connector
• 09-07 23:06:57 5d43d97 Document setdigest type
• 09-08 23:24:08 856331f Revise async and read-through caching
• 09-09 09:44:51 b065b01 Fix in clause handling in filter pushdown
• 09-09 11:42:35 bd4ca21 Minor cleanup to MemSQL query runner
• 09-09 11:42:35 c492daa Add main method to MemSQL query runner
• 09-09 21:36:11 11627c8 Fix connection name example and link to Oracle docs
• 09-09 22:39:58 108e733 Fix typo in PinotColumn error message
• 09-09 22:39:58 5d2631a Remove unused variables and functions from DynamicTableBuilder
• 09-09 22:39:58 ad7d903 Fixed mismatched schema names
• 09-09 22:39:58 c1cfbcb Fix realtime spec for alltypes table
• 09-09 23:17:33 1676545 Fix decoding infinite double values in Pinot
• 09-09 23:17:33 4ca1112 Add support for pushdown of varbinary filters to Pinot
• 09-09 23:17:33 f124f10 Fix decoding infinite real values in Pinot
• 09-10 09:31:54 2baea31 Remove ineffective cleanup step
• 09-10 09:31:54 fd7fa04 Upload test results as artifacts
• 09-10 09:57:37 29ac28c Set updated_at to a recent value to prevent expiration by Pinot

Dain Sundstrom

• all checked
• 09-01 19:46:48 0c43e24 Use Optional instead of nullable in SessionContext
• 09-01 19:46:48 13b8f60 Remove redundant "in this catalog" from ConnectorAccessControl docs
• 09-01 19:46:48 16c5281 Add MetadataUtil.getRequiredCatalogHandle
• 09-01 19:46:48 2d3251d Split HttpRequestSessionContext into factory and instance
• 09-01 19:46:48 3641ca1 Automatically add admin role to file based metastore
• 09-01 19:46:48 3cb2db9 Cleanup CreateSchemaTask
• 09-01 19:46:48 3e2bd90 Add metadata role exists method
• 09-01 19:46:48 411ad80 Remove catalog argument to ConnectorAccessControl methods
• 09-01 19:46:48 5ce1ae7 Convert SessionContext to a class
• 09-01 19:46:48 8c91ae8 Fix synchronization in TestWebUi
• 09-01 19:46:48 ce1eb14 Fix SetSchemaAuthorizationTask role validation
• 09-01 19:46:48 f386df0 Add checkRoleExists helper method
• 09-01 20:08:58 48ec9f3 Add Hive target max file size
• 09-02 13:36:33 e0a1f32 Document hive.target-max-file-size configuration
• 09-03 18:26:36 230f1c6 Extract abstract super class for TestHiveRoles
• 09-03 18:26:36 4b2565a Only send security metadata calls to connectors with an access control
• 09-03 18:26:36 56abe82 Add optional catalog declaration to role management statements
• 09-03 18:26:36 754b6b0 Update documentation for role management commands
• 09-03 18:26:36 7da3c1a Add SystemSecurityMetadata for management of system grants
• 09-03 18:26:36 868fed9 Add role to file based system access control
• 09-03 18:26:36 9b65a9e Add system access control checks for global role management
• 09-03 18:26:36 b7c34f4 Add support for global role management
• 09-03 18:26:36 c4782ed Add enabled system roles to Identity
• 09-03 18:26:36 cf3b571 Fix output of revoke role statement
• 09-03 18:26:36 fdeec92 Rename roles to connectorRoles in Identity
• 09-06 12:37:18 68a01e3 Fix warnings in TpchMetadata
• 09-06 12:37:18 9b6cf0c Add config option to map TPCH doubles as SQL decimal

David Phillips

• all checked
• 08-30 14:32:00 b03065e Remove bogus INVALID_COLUMN_NAME error code
• 08-31 17:59:06 627029c Add documentation for additional Hive Views
• 09-03 08:50:36 3192f36 Add query error info to cluster overview page in web UI
• 09-08 11:13:50 253a6b0 Minor formatting improvements for grammar
• 09-08 23:19:53 710b7e0 Add toString method to PropertyMetadata
• 09-08 23:24:19 bb06a1a Update Maven wrapper to 3.8.2
• 09-09 17:47:39 14a20c4 Move top level markdown files
• 09-09 17:55:16 945d4c3 Update to Avro 1.10.2
• 09-09 17:56:44 0b97643 Correct connection string examples
• 09-09 17:59:34 854c722 Update to ANTLR 4.9.2 and Jdbi 3.21.0
• 09-10 10:38:23 e0f3998 Add SQL support section to Thrift connector
• 09-10 10:43:17 1fe0f3e Add SQL support section to Phoenix connector
• 09-10 10:56:39 f687bac Add Predicate and Column Projection Pushdown Section
• 09-10 11:08:51 0b74ace Add note about no Java 17 support
• 09-13 11:58:29 2e0f3b8 Add SQL support section
• 09-14 13:43:39 d3a6c2a Move bin/ptl to testing directory
• 09-16 10:15:44 53d1cdf Add table of contents for DEVELOPMENT
• 09-16 10:15:44 5f501c1 Clarify multiple commits in a PR
• 09-16 10:15:44 d0df475 Create commits and pull requests section
• 09-16 10:47:10 905f033 Only support single statement writes by default

GitHub

• all checked
• 09-09 17:58:58 1cd649a Change Trino RPM architecture to noarch

Grzegorz Kokosiński

• all checked
• 08-30 01:03:59 0e8a992 Mock more functionalities in MockConnector
• 08-31 04:38:42 3900829 Update airbase to 112
• 08-31 07:42:07 3532ff9 Allow to override default system access control
• 09-01 22:27:57 f6b1e7e Accept that transaction is cancelled early in TestJdbcConnection
• 09-02 04:02:38 705023c Use configureTempto instead CONTAINER_TEMPTO_PROFILE_CONFIG
• 09-02 04:02:38 7c410a5 Extract configureTempto utility method
• 09-02 04:02:38 be32a75 Deprecate CONTAINER_TEMPTO_PROFILE_CONFIG
• 09-02 04:27:09 6cd9332 Make IdentityCacheMapping accept ConnectorSession
• 09-02 04:37:41 18814a3 Restore default system access control
• 09-03 05:30:16 3774afb Use URLS_SELF_ISSUER value for oauth2.issuer

Karol Sobczak

• all checked
• 08-31 02:01:00 a79829f Fix incorrect currentProbePositionProducedRow value for inner join
• 09-01 00:53:40 a762fdd Remove transitionToRunning on Stage schedulingComplete()
• 09-01 01:33:40 b838333 Improve JoinProbe
• 09-01 01:33:40 bdec925 Use blockBuilders.length in PageBuilder#reset()
• 09-01 01:33:40 cc622f9 Improve LookupJoinPageBuilder
• 09-01 12:09:46 ef10932 Doc remove keyword sorted of window function
• 09-01 14:03:06 8d8466b Introduce RowPagesBuilder#getHashChannels method and use it in tests
• 09-02 03:56:34 1fffa57 Do not use LocalPartitionGenerator
• 09-02 03:56:34 5c65f48 Add createRandomLongsBlock using limited number of unique values
• 09-02 03:56:34 9e4728d Make HashBucketFunction public and extract as top level class
• 09-02 03:56:34 a43fdbe Fix hash computation bug
• 09-02 03:56:34 be88e21 Add more cases to BenchmarkPartitionedOutputOperator
• 09-06 02:24:51 1786442 Fix formatting
• 09-06 02:24:51 2b8387d Do not set stats that are not propagated for nested operators
• 09-06 02:24:51 3982257 Mark OperatorContext#getOperatorStats as private
• 09-06 02:24:51 5a9617b Move private method below public
• 09-06 03:25:38 15c9c53 Optimize MarkDistinctHash for faster distinct mask creation
• 09-06 03:25:38 a673343 Add direct block construction helpers to BooleanType
• 09-06 13:17:29 6bd6172 Improve HashGenerator.getPartition performance
• 09-07 02:50:05 826558a Prevent NPE when getting freshness of non-existent materialized view
• 09-07 02:50:05 a14b9ac Remove unthrown exception declaration from Javadoc
• 09-07 02:50:05 fff485a Skip displaying materialized views with missing freshness
• 09-07 03:43:44 83a62fd Produce better hash for short decimal type
• 09-07 09:27:04 c059cda Allow for operators to define custom metrics
• 09-07 09:27:04 d3a3054 Rename connector metrics to metrics
• 09-10 02:51:26 ca8e71e Extract OperatorStats#connectorMetrics

Martin Traverso

• all checked
• 08-31 14:30:00 c4f08c7 Update to aircompressor 0.21
• 09-07 10:08:10 c3ff961 Improve Block#getLoadedBlock() and Page#getLoadedPage()
• 09-08 11:16:56 4c321a9 Support string joining in SQL via LISTAGG clause
• 09-08 11:59:43 639020c Add benchmarkHashPosition to BenchmarkGroupByHash
• 09-08 11:59:43 8de7380 Add positional only utility for InterpretedHashGenerator creation
• 09-09 10:16:05 abc2d24 Remove totalPartitions from TaskContext
• 09-10 13:35:40 fce0521 Speed up construction of parquet predicate from dictionary

Piotr Findeisen

• all checked
• 08-30 08:57:21 8ab3573 Report actual rows when returnsEmptyResult assertion fails
• 08-30 08:57:21 94217db Rename base test class to match current convention
• 08-31 09:25:18 031fbbe Fix testMaterializedViewsMetadata flakiness due to lack of pruning
• 08-31 09:25:18 40a4937 Fix test table name
• 08-31 09:25:18 8a27c4c Fix type of expected values
• 09-01 02:24:10 e7181f4 Remove useless todo list for residual expression
• 09-02 02:14:51 15a0ff3 Fix indentation
• 09-02 02:14:51 329c48d Convert deprecation comment into a javadoc
• 09-02 02:14:51 57594b8 Remove redundant SinglenodeSparkIceberg configuration
• 09-02 02:14:51 b9df5b6 Test Iceberg schema top-level evolution
• 09-02 03:24:17 0ee1ad4 Enforce naming convention for PTL configs
• 09-02 03:24:17 1b6bac7 Apply static imports consistently
• 09-02 03:24:17 2acfcc6 Enforce naming convention for PTL suites
• 09-02 03:24:17 86bc379 Test suite name generation
• 09-02 03:24:17 90797ae Rename utility methods
• 09-02 03:24:17 ac7629d Split Environments utility class
• 09-03 01:03:51 00626a4 Split environment and config name canonicalization
• 09-03 01:03:51 823540b Update environment names to naming convention
• 09-03 01:03:51 b42a30f Introduce naming convention for PTL environments
• 09-03 01:03:51 eb57307 Update environment name canonicalization tests
• 09-07 02:14:59 d47be23 Retain commit's failure when rollback fails
• 09-07 02:15:23 92c6cfe Move Hive tests's configs out of product tests
• 09-07 08:04:43 09dc0bf Add dereference pushdown smoke test coverage
• 09-07 08:04:43 0d0fd3b Use ID-based mapping for struct fields in Iceberg Parquet reader
• 09-07 08:04:43 a74fd53 Split ParquetColumnIOConverter for Hive and Iceberg
• 09-07 08:04:43 d3e22e0 Code cleanup in ParquetColumnIOConverter
• 09-07 08:04:43 ef5d9d1 Fix missing defensive copy in ColumnIdentity constructor
• 09-07 08:12:57 1a71ec5 Test Snappy-compressed Parquet table created in Trino
• 09-07 08:12:57 3cd36e3 Move Parquet Snappy test to different class
• 09-07 08:12:57 beef438 Add Hive compression_codec integration test for all formats
• 09-08 06:34:04 0bd01d8 Include actually set properties in exception message
• 09-08 06:34:04 8f89f04 Exercise optimized Parquet writer in Hive connector test
• 09-08 06:34:04 b3cfc0d Exercise optimized Parquet writer in Hive compression product test
• 09-08 06:34:04 b3fcd96 Reject unexpected type in some date/time column readers
• 09-08 06:34:04 c84a413 Read INT64 Parquet timestamp into timestamp(9) column
• 09-08 06:48:45 6349e6a Mark hive views testNestedGroupBy test as @flaky
• 09-08 06:49:09 ad1a432 Bump snappy-java version
• 09-09 23:55:54 b5ef0c9 Reorder parquet_use_column_index in HiveSessionProperties
• 09-10 02:11:54 31dbbdb Improve explanatory comment in test
• 09-10 02:11:54 3611b0c Test Hive connector with Spark-created Parquet table
• 09-10 02:11:54 66b7658 Rename hive_spark_bucketing test group to hive_spark

Pratham

• all checked
• 08-30 18:01:03 6f1297f Fix issue with querying Iceberg using a structural type predicate

Praveen Krishna

• all checked
• 09-02 22:47:04 659bf66 Use static import for BucketingType
• 09-02 22:47:04 8f5aa1c Treat table as unbucketed if bucketed on unsupported data type

Yuya Ebihara

• all checked
• 08-28 18:46:42 6883ebd Add SQL support section to Redis connector
• 08-28 18:48:43 778eb84 Add SQL support section to Prometheus connector
• 08-29 18:33:58 95d5f46 Add SQL support section to Cassandra connector
• 08-30 04:16:18 9024438 Fix query failure of predicate pushdown to BigQuery DATETIME type
• 08-30 08:47:54 1fc3ddc Append semicolons in DESCRIBE OUTPUT example
• 08-31 18:05:45 46b37aa Add SQL support section to Elasticsearch
• 08-31 21:26:35 14c4eb0 Add SQL support section to Pinot connector
• 08-31 23:37:51 1c3807d Fix wrong result of BigQuery parameterized NUMERIC type
• 08-31 23:37:51 4a02a18 Upgrade errorprone to version 2.9.0
• 09-01 21:55:48 3acb446 Skip unsupported type in BigQuery connector
• 09-02 23:37:37 193db65 Allow using 1 as cassandra.speculative-execution.limit
• 09-05 23:48:09 647b4d4 Cast float64 type correctly when predicate pushdown to BigQuery
• 09-06 21:14:05 ad40bd7 Make table wider in BigQuery documentation
• 09-06 21:14:05 f72a590 Change date/time precision in BigQuery from 3 to 6
• 09-08 04:13:37 4b45f1a Document ARRAY type mapping in BigQuery connector
• 09-08 06:25:00 d3d2ba7 Delete even non-empty dataset when preparing BigQuery tests
• 09-08 16:11:39 515d41d Recreate Oracle container when failed to setup
• 09-09 00:28:15 6d22525 Support case insensitive name matching to BigQuery views
• 09-09 00:28:15 7de96aa Allow setting name prefix in TestView
• 09-09 00:28:15 a5264f7 Prepend 'remote' to method and variables in BigQuery
• 09-09 00:28:15 b92df3c Allow deleting non-empty dataset in BigQuery test

kasiafi

• all checked
• 08-30 08:02:22 e961325 Fix row field names in JSON documentation
• 09-09 23:43:21 10a85b1 Fix planning of recursive queries with duplicate outputs of partial plan

Łukasz Osipiuk

• all checked
• 08-31 02:34:57 f947aed Fix failure query casts varchar to shorter char
• 09-02 13:51:00 25965e9 Rename variable
• 09-02 13:51:00 96e7323 Drop unnecessary braces
• 09-02 13:51:00 b87165d Include grouping sets in JdbcTableHandle column list
• 09-02 13:51:00 d55bb1c Implement count(distinct) pushdown for PostgreSQL
• 09-06 09:20:13 8216444 Fix locking of non-partitioned transactional tables
• 09-09 02:56:47 973e714 Record JMX metrics for calls made to the Glue statistics API

[ebyhr]

BigQuery
* Fix query failure for predicate pushdown to BigQuery `datetime` type. ({issue}`9005`)

#9005

[findepi]

General
* Fix query failure when query contains a cast from `varchar` to a shorter `char`. ({issue}`9036`)

#9036 #9038

[ebyhr]

BigQuery
* Fix incorrect results when using parameterized `numeric` type with non-default precision and scale. ({issue}`9060`)

#9060

[sopel39]

General
* Improve join performance. ({issue}`8974`)

#8974

[dain]

Hive
* Limit maximum file size to 1 GB by default.  The limit is not exact and is performed best effort.  The limit can be set with the `target_max_file_size` session property, or the `hive.target-max-file-size` configuration property. ({issue}`7991`)

#7991

[dain]

General
* Fix `ALTER SCHEMA ... SET AUTHORIZATION` to resolve roles using the catalog of the schema instead of the session catalog. ({issue}`9066`)

Hive
* Fix file-based metastore to always add the admin role by default. ({issue}`9066`)

#9066

[ebyhr]

BigQuery
* Fix query failure when accessing tables having unsupported type columns. ({issue}`9086`)

#9086

[ebyhr]

Cassandra
* Change minimum number of speculative executions from 2 to 1. ({issue}`9096`)

#9096

[dain]

Security
* Add `IN catalog` clause to `CREATE ROLE`, `DROP ROLE`, `GRANT ROLE`, `REVOKE ROLE`, and `SET ROLE` to specify the target catalog of the statement instead of using the current session catalog.  This change is necessary to support system roles in the future.  The old behavior of these statements can be restored by setting the `deprecated.legacy-catalog-roles` config property. ({issue}`9087`)

#9087

[ebyhr]

BigQuery
* Fix query failure for predicate pushdown to BigQuery `float64` type with `+infinity` or `-infinity` value. ({issue}`9122`)

#9122

[sopel39]

General:
* Improve performance of queries that contain `DISTINCT` clause. ({issue}`8967`)

#8967

[dain]

TPCH
* Add support for switch mapping for floating point values between SQL `DOUBLE` and `DECIMAL` types.  The mapping can be set  the `tpch.double-type-mapping` configuration property. ({issue}`7990`)

#7990

[sopel39]

General
* Improve query performance by reducing CPU overhead of repartitioning data across worker nodes. ({issue}`9102`)

#9102

[ebyhr]

BigQuery
* Chage type mapping of BigQuery `datetime` from `timestamp(3)` to `timestamp(6)` in Trino. ({issue}`9052`)
* Chage type mapping of BigQuery `time` from `time with time zone` to `time(6)` in Trino. ({issue}`9052`)
* Chage type mapping of BigQuery `timestamp` from `timestamp(3) with time zone` to `timestamp(6) with time zone` in Trino. ({issue}`9052`)

#9052

[sopel39]

General
* Prevent failure when querying `system.materialized_views` and materialized view is dropped concurrently. ({issue}`9050`)

#9050

[findepi]

Iceberg
* Fix the mapping of nested fields between table metadata and Parquet file metadata. This 
  enables evolution of `row` typed columns for Iceberg tables stored in Parquet. ({issue}`9124`)

(wording borrowed from 352 notes for similar change for ORC)
#8750
#9124

[findepi]

Hive
* Fix failure when writing Parquet files with Snappy compression on ARM64. ({issue}`9148`)

#9148

[martint]

* Add support for {func}`LISTAGG`. ({issue}`4835`)

#4835

[ebyhr]

BigQuery
* Support case insensitive name matching for BigQuery views. ({issue}`9164`)

#9164

[hashhar]

Pinot
* Fix handling of `IN` and `NOT IN` clauses for `VARCHAR` columns in presence of aggregations. ({issue}`9133`)

#9133, #9181

[hashhar]

Pinot
* Support pushdown of filters on `VARBINARY` columns to Pinot. ({issue}`9180`)

#9180

[kasiafi]

General
* Fix planning of recursive queries when the recursion the base plan or the recursion step plan produce duplicate outputs.

#9153

[sopel39]

General
* Split metrics that are reported in operator stats into `connectorMetrics` and `metrics`. ({issue}`9156`)

#9156

[martint]

Hive
* Fix performance regression where reading Parquet files that contain dictionaries. ({issue}`9161`)

#9161

[Praveen2112]

Hive
* Allow reads on table if they are bucketed on unsupported data type. 

#8986

[Praveen2112]

Hive
* Optionally specify the schema which enforces that a filter on a partition key be present in the query for its table. This can be specified by setting the 
  ``hive.query-partition-filter-required-schemas`` config property or the ``query_partition_filter_required_schemas`` session property

#9106

[kasiafi]

General
* Fix planning failure of INSERT statement when source table has hidden columns. ({issue}`9150`)

#9150, #9201

[phd3]

* Iceberg
- Fix failure for queries filtering on columns with array, map, or row type. ({issue}`8822`)

#8822

[hashhar]

Pinot
* Fix failure for queries on `REAL` or `DOUBLE` columns having `+Infinity` or `-Infinity` values. ({issue}`9180`)

#9180 (2nd and 3rd commits)

[findepi]

Hive
* Support reading Parquet timestamp stored with millisecond or microsecond precision as INT64 with logical type annotations when Hive timestamp precision is `NANOSECONDS`. ({issue}`9139`)
* Support reading Parquet timestamp stored as INT64 with nanosecond precision. ({issue}`9188`)
* Support writing Parquet timestamp stored as INT64 with nanosecond precision when experimental Parquet writer is enabled. 
  To use this, the Hive timestamp precision should be set to `NANOSECONDS`. ({issue}`9188`)

#9139 #9188

[kokosing]

Hive
 * Support loading of s3 mappings via http/s url. The `hive.s3.security-mapping.config-file property` can now either point to a local file or a URL. ({issue}`6210`)

#6210

[phd3]

* General
- Fix failure when querying `optimizer_rule_stats` system table. ({issue}`8700`)

#8700, #9237

[losipiuk]

Hive
 * Fix incorrect table locking in Hive metastore when querying transactional and not partitioned table. ({issue}`9070`)

#9070

[losipiuk]

Hive
 * Extend set of statistics reported by JMX to include metrics for calls made to the Glue statistics API. ({issue}`9100`)

#9100

[losipiuk]

General
 * Limit number of metadata calls done for particular queries reading from `system.jdbc.columns` where filter on schema name is present. ({issue}`9214`)

#9214

(should we mention Metadata API exposed by JDBC driver instead?)

[findepi]

Hive
* Fix `ArrayIndexOutOfBoundsException` when using the experimental Parquet writer. ({issue}`5518`)

Iceberg
* Fix `ArrayIndexOutOfBoundsException` when writing Parquet data. ({issue}`5518`)

#5518 #9245

[findepi]

(should we mention Metadata API exposed by JDBC driver instead?)

@losipiuk yes

[martint]

* Fix incorrect results for aggregations that contain `DISTINCT` and `FILTER` clauses. ({issue}`9265`)

#9265

[martint]

* Improve performance for aggregations that contain a `DISTINCT` clause. ({issue}`9194`)

#9194

[kasiafi]

General
* Fix incorrect query results when row pattern involves back references. ({issue}`9109`)

#9109

[phd3]

General
* Fix failure for queries that push projections into the connectors. ({issue}`6200`)

#6200, #9185, #9222

[losipiuk]

(should we mention Metadata API exposed by JDBC driver instead?)

@losipiuk yes

https://github.com/trinodb/trino/pull/9216/files#r711073226

[electrum]

## Server RPM

* Change RPM architecture to `noarch` to allow installing on any machine type. ({issue}`9187`)

#9187

[electrum]

## Web UI

* Add query error info to cluster overview page. ({issue}`8762`)

#8762

[electrum]

## Hive Connector

* Fix reading Avro tables written with older Avro versions that had certain semi-invalid schemas. ({issue}`9114`)

## Iceberg Connector

* Fix query failures caused by errors reading certain Avro metadata files. ({issue}`9114`)

#9114

[electrum]

## SPI

* Change `Connector.isSingleStatementWritesOnly()` to return `true` by default. ({issue}`8872`)

#8872

[findepi]

Iceberg
* Add support for Iceberg `uuid` type. ({issue}`6663`)

#6663 #8747

[losipiuk]

## Hive Connector

* Fix possible `INSERT`/`UPDATE`/`ANALYZE` query failure when Glue metastore is in use and table statistics collection is enabled. ({issue}`9297`)

#9297

[martint]

* Fix planning timeout for queries containing `IS NULL`, `AND` and `OR` predicates in the `WHERE` clause. ({issue}`9250`)
* Fix failure for queries containing `ORDER BY ... LIMIT` when columns in the subquery are known to be constant. ({issue}`9171`)

#9250, #9171