…ence issues (Steven Talbot)

Close apache#1239

…ence issues (Steven Talbot)

Fix by unparsing operands of SqlRollupOperator with higher precendence
than the current operator: requires making private helper unparseCube
no longer static to use inherited predence getters.

Close apache#1239

…ence issues (Steven Talbot)

RelToSqlConverter was converting each composite item in GROUPING SETS into
a SqlNodeList when it should be a call to the ROW operator.

Close apache#1239

* [CALCITE-3003] AssertionError when GROUP BY nested field (Will Yu)

Close apache#1186

* Oops, remove a println

* [CALCITE-3049] When simplifying "IS NULL" and "IS NOT NULL", simplify the operand first

When simplifying "(x IS NULL) IS NOT NULL", remove redundant casts
before simplifying the operand.

Julian Hyde started work on this case, simplifying before and after,
and Danny Chan improved it by only simplifying before.

Close apache#1198

* [CALCITE-3039] In Interpreter, min() incorrectly returns maximum double value (dijkspicy)

Close apache#1197

* [CALCITE-3054] Elasticsearch adapter. Avoid scripting for simple projections

Simple query like below is using `script` when `_source` is sufficient.
```sql
select  _MAP['a'], _MAP['b.a'] from elastic
```

* [CALCITE-2846] Document Oracle-specific functions, such as NVL and LTRIM, in the SQL reference

* [CALCITE-3053] Add a test to ensure that all functions are documented in the SQL reference

Following [CALCITE-2846], fix up the documentation of MySQL- and
Oracle-specific operators to make it better fit narrow web pages.

* [CALCITE-3046] CompileException when inserting casted value of composited user defined type into table

Also, add several test cases for EnumerableRelImplementor.TypeFinder.

* [CALCITE-3056] Elasticsearch adapter. Invalid result with cast function on raw queries

Fix queries of type (with elastic adapter):

```sql
select max(cast(_MAP['foo'] as integer)) from elastic
```

* Mongo adapter. Mongo checker validates only first line of the Bson query in tests

`BsonDocument.parse()` is more lenient with input and this omission was not noticed before

* Site: Improve contribution guidelines for JIRA

Summarize the respective discussion in the dev list(
https://lists.apache.org/thread.html/59d028ec300884e432f9be1abe894d076e94b9f1ac9427cb22fa745d@%3Cdev.calcite.apache.org%3E) and include the suggestions from all participants.

* [CALCITE-3017] Re-organize how we represent built-in operators that are not in the standard operator table

Add enum SqlLibrary, with values STANDARD, SPATIAL, ORACLE, MYSQL etc.,
and SqlLibraryOperatorTableFactory, that can create operator tables for
any combination of libraries.

Further changes by Julian Hyde:
 * Adopt the term "library" (earlier revisions had "dialect" and "flavor");
 * Change SqlDialectOperatorTableFactory into a cache (previously it
   pre-built one table per library);
 * Make it responsible for standard and spatial libraries in addition to
   database-specific ones;
 * Move operators from SqlLibraryOperatorTableFactory into a dedicated
   holder class, SqlLibraryOperators;
 * Deprecate OracleSqlOperatorTable, and move its operator definitions.

Close apache#1203

* [CALCITE-3061] Query with WITH clause fails when alias is the same as the table with rolled up column

Close apache#1206

* [CALCITE-2803] ProjectTransposeJoinRule messes INDF expressions

ProjectTransposeJoinRule does not identify expanded versions of IS NOT
DISTINCT FROM expressions in the join filter, and might push them below
the join operator in a way which makes impossible for
RelOptUtil#splitJoinCondition() to identify the INDF condition and mis-
categorize the join as not being an equi-join.

Fix the issue by collapsing INDF expressions before invoking PushProjector.
Also add support for expanded form CASE(WHEN A IS NULL THEN B IS NULL
WHEN B IS NULL THEN A IS NULL ELSE A = B).

* Site: Add Danny Chan as committer

* [CALCITE-3034] CSV test case description does not match it's code logic (FaxianZhao)

The o.a.c.a.c.CsvFilterableTable will ignore the whole filter condition
with AND as SqlKind before this patch.

Close apache#3034

* [CALCITE-2936] Simplify EXISTS or NOT EXISTS sub-query that has "GROUP BY ()"

* [CALCITE-3017] Improve null handling of JsonValueExpressionOperator

* Remove SqlJsonApiCommonSyntaxOperator;
* Move implementation methods of JSON functions from SqlFunctions to JsonFunctions;
* Add new null policy NullPolicy.ARG0 (currently used by some JSON functions only);
* Change return type to nullable for some operators of JSON functions;
* In code generation, perform regular null semantic check on operands of a call when NullAs.IS_NULL/IS_NOT_NULL is used.

* [CALCITE-3028] Support FULL OUTER JOIN with AggregateJoinTransposeRule (Vineet Garg)

* [CALCITE-3066] RelToSqlConverter may incorrectly throw an AssertionError for some decimal literals

Close apache#1213

* [CALCITE-3052] Error while applying rule MaterializedViewAggregateRule(Project-Aggregate): ArrayIndexOutOfBoundsException

Close apache#1212

* [CALCITE-2282] Remove sql operator table from parser

Differentiate for unresolved functions and non-parameters function identifier unparsing;
Merge logic in SqlIdeitifier and SqlUtil;
Fix Julian's comments address: Remove leftPrec and rightPrec from SqlCollation.

Close apache#1205

* [CALCITE-2933] Add timestamp extract for casts from timestamp type to other types

In Druid, timestamp is stored as a UTC long value, so the
Cast(timestamp as varchar) should transform to date format string.

Close apache#1120

* [CALCITE-2975] Implement JSON_REMOVE function (xuqianjin)

Close apache#1146

* [CALCITE-3062] Do not populate provenanceMap if not debug

VolcanoPlanner.provenanceMap captures provenance of all rel nodes added
to the planner but the information is only printed out if planner log
level is debug or finer.  As the map can get quite big with complex
queries, the memory usage can increase significantly.

Detect if planner log level is debug when creating a new volcano planner
and use a blackhole map so that the map is not populated by default.

Change-Id: If6393eb07438ea5f922c9f78f26089160926ef37

* [CALCITE-3074] Move MySQL's JSON operators to SqlLibraryOperators

* [CALCITE-2965] Implement string functions: REPEAT, SPACE, SOUNDEX, DIFFERENCE

* [CALCITE-2712] Add rule to remove null-generating side of a Join

* [CALCITE-3072] Generate right SQL for FLOOR&SUBSTRING functions in SparkSqlDialect (DonnyZone)

TimeUnit first in SparkSQL date_trunc;
Add test for floor with one operand.

Close apache#1216

* [CALCITE-3068] testSubprogram() does not test whether subprogram gets re-executed

* [CALCITE-3076] AggregateJoinTransposeRule throws error for unique under aggregate keys when generating merged calls

AggregateJoinTransposeRule generates wrong mapping for under agg
calls with unique aggregate keys.

Close apache#1223

* [CALCITE-2601] Add REVERSE function (pingle wang)

Close apache#1209

* [CALCITE-3067] Splunk adapter cannot parse right session keys from Splunk 7.2 (Shawn Chen)

Close apache#1222

* [CALCITE-3023] Upgrade elastic search to 7.x (Takako Shimamoto)

* [CALCITE-3050] Integrate SqlDialect and SqlParser.Config

Add a method SqlDialect.configureParser(SqlParser.ConfigBuilder),
which allows a dialect to pass on its settings to a SQL parser.

Also add SqlDialect methods getUnquotedCasing(), getQuotedCasing(),
getConformance(), isCaseSensitive().

Add Snowflake dialect.

* [CALCITE-2807] Fix `IS NOT DISTINCT FROM` expression identification in RelOptUtil#pushDownJoinConditions()

RelOptUtil#pushDownJoinConditions do not identify and preserve
expanded versions of `IS NOT DISTINCT FROM` expressions, causing
equi-joins to be miscategorized as inequality joins.

Modify the function to try to collapse the expression back to a
canonical `IS NOT DISTINCT FROM` expression if possible before
visiting the expression and pushing it below the join.

* Oops, broke a test on Windows

* [CALCITE-2985] Implement JSON_STORAGE_SIZE function (xuqianjin)

Also, correct content about error handling of MySQL JSON functions in documentation.

Close apache#1150

* [CALCITE-3084] Implement JDBC string functions: ASCII, REPEAT, SPACE, SOUNDEX, DIFFERENC (pingle wang)

* [CALCITE-3090] Use https for maven central

Closes apache#1233

Signed-off-by: Julian Hyde <[email protected]>
Signed-off-by: Kevin Risden <[email protected]>

* [CALCITE-2913] Adapter for Apache Kafka (Mingmin Xu)
Expose an Apache Kafka topic as a stream table.

* Revert "[CALCITE-3090] Use https for maven central"

This reverts commit 1310295.

* [CALCITE-3090] Remove Central configuration

We don't actually need the explicit configuration for Central. We're
just providing the same value that's already configured.

Closes apache#1234

Signed-off-by: Kevin Risden <[email protected]>

* Improve display of reproducer code in RexFuzzerTest for SqlStdOperatorTable.MULTIPLY

It was displayed as SqlStdOperatorTable.*, however mul(a, b) should be used

* Add system properties to RexProgramFuzzyTest to simplify git bisect

* [CALCITE-3077] Rewrite CUBE&ROLLUP queries in SparkSqlDialect (DonnyZone)

* [CALCITE-35] More test cases to guard against providing a broken fix for parenthesized join (Muhammad Gelbana)

Close apache#1228

* Make field JdbcTableScan.jdbcTable public

* Remove call to Class.newInstance(), which is deprecated as of JDK 9

* Cosmetic changes

* [CALCITE-3021] Equality of nested ROWs returns false for identical values

* [CALCITE-2952] Document JDK 12 support

* [CALCITE-3055] Use pair of relNode's rowType and digest as unique key for cache in RelOptPlanner (KazydubB)

Currently RelNode's digest can not uniquely identify itself,
two relational expressions are equivalent only if their digests and row
types are the same.

* Add unit test and update code for testing dynamic tables to reflect real use-cases;
* Add row type as part of cache key for HepPlanner.

close apache#1230

* [CALCITE-3096] In RelBuilder, make alias method idempotent

* Javadoc typos (Wenhui Tang, Muhammad Gelbana)

Close apache#1238
Close apache#1194

* [CALCITE-3022] Babel: Various SQL parsing issues

* Allow 'DATE(x)' function, per PostgreSQL and Redshift
* Allow "TIMESTAMP 'yyyy-mm-dd'" literal (missing time part), per PostgreSQL and Redshift

* [CALCITE-3048] Improve how JDBC adapter deduces current schema on Redshift

In JdbcSchema on PostgreSQL or Redshift, if schema is null, call
CURRENT_SCHEMA() to get it. Similarly catalog and CURRENT_DATABASE().

Without this fix, we sometimes call DatabaseMetaData.getTables with null
or empty schema, and get tables from other schemas, resulting in
a Guava "Multiple entries with same key" error.

* [CALCITE-3047] In JDBC adapter, expose multiple schemas of the back-end database

We do this by adding JdbcCatalogSchema, an implementation of Schema
that reads the back-end database catalog and for each schema, creates
a sub-schema that is a JdbcSchema.

Also add an experimental way to create a Calcite connection with a
given schema as its root schema (in this case, a JdbcCatalogSchema)
rather than the usual AbstractSchema.

Rather than calling JdbcSchema.getDataSource(), generated code now
calls schema.unwrap(DataSource.class). This works for JdbcCatalogSchema
as well as JdbcSchema.

* [CALCITE-3097] GROUPING SETS breaks on sets of size > 1 due to precedence issues (Steven Talbot)

RelToSqlConverter was converting each composite item in GROUPING SETS into
a SqlNodeList when it should be a call to the ROW operator.

Close apache#1239

* [CALCITE-2812] Add algebraic operators to allow expressing recursive queries

1. Add Spool, and RepeatUnion operators with their Logical and Enumerable counterparts as well as appropriate converter rules.
2. Add new interface allowing to express transient tables with a sample implementation.
3. Add new methods in RelBuilder useful for building recursive queries.
4. Update the website with examples for building recursive queries.
5. Add HierarchySchema allowing to express interesting hierarchies through recursive queries.
6. Add unit tests verifying the plan created by the builder and correctness of the new operators.

Close apache#1020

* RelOptUtilTest concurrency fixup

RelOptUtilTest was creating a static RelOptCluster which could be used
during tests running in parallel to acces the associated RelMetadataQuery
instance.  Alas the object is not safe to access from multiple threads
and in some occasions tests would fail with CyclicMetadataException.

Fixing the test class by creating a new RelOptCluster/RelBuilder
instance for each test, and removing the static instances.

* [CALCITE-2696] Improve design of join-like relational expressions

**Diff**
- Deprecate SemiJoin, EnumerableSemiJoin, SemiJoinType,
EnumerableSemiJoinRule, JoinToCorrelateRule#SEMI
- Add SEMI and ANTI join type to JoinRelType, add method
projectsRight() and isOuterJoin()
- Correlate use JoinRelType instead of SemiJoinType
- Rename EnumerableThetaJoin to EnumerableNestedLoopJoin
- Rename EnumerableJoin to EnumerableHashJoin
- EnumerableJoinRule converts semi join to EnumerableHashJoin
(EnumerableSemiJoin's functionality is merged into this rule)
- Add method isSemiJoin() in Join.java to decide if this join is a semi join
(comes from SemiJoinRule during decorrelateation),
the return value true means the join is a semi join equivalent to
SemiJoin before this patch.
- Cache the JoinInfo in Join and use it to get leftKeys and rightKeys,
merge SemiJoin#computeSelfCost logic into Join#computeSelfCost
- For RelBuilder, removes SemiJoinFactory, method #semiJoin
now returns a LogicalJoin with JoinRelType#SEMI
- Rename EnumerableDefaults#join to EnumerableDefaults#hashJoin
- Rename EnumerableDefaults#thetaJoin to EnumerableDefaults#nestedLoopJoin

**Rules tweak**
- JoinAddRedundantSemiJoinRule now creates LogicalJoin with JoinRelType#SEMI
instead of SemiJoin
- JoinToCorrelateRule deprecates SEMI instance and change the matches
condition to !join.getJoinType().generatesNullsOnLeft() which also
allowed ANTI compared before this patch.
- SemiJoinRule matches SEMI join specifically.

**Metadata tweak**
- RelMdAllPredicates, RelMdExpressionLineage: Add full rowType to
getAllPredicates(Join) cause semi join only outputs one side
- RelMdColumnUniqueness, RelMdSelectivity, RelMdDistinctRowCount,
RelMdSize, RelMdUniqueKeys: merge semi join logic to join

**Test cases change**
- MaterializationTest#testJoinMaterialization11 now can materialize
successfully, cause i allow logical SemiJoin node to match, the original
matches SemiJoin as SemiJoin.class.isAssignableFrom(), which i think is
wrong cause this will only matches subClasses of SemiJoin which is only
EnumerableSemiJoin before this patch.
- SortRemoveRuleTest#removeSortOverEnumerableCorrelate, because
CALCITE-2018, the final EnumerableSort's cost was cached by the previous
EnumerableSort with logical children, so i remove the EnumerableSortRule
and the best plan is correct

close apache#1157

* [CALCITE-3095] Add several system properties to control enabling/disabling of rules and traits

Also, add utility class RelOptRules to reorganize built-in rules from different places.

Close apache#1237

* [CALCITE-3093] Remove JDBC connection calls from PlannerImpl

Close apache#1241

* [CALCITE-3082] Fix NPE in SqlUtil#getSelectListItem

Queries similar to `SELECT 1 UNION SELECT 2, 3` might cause Calcite
validator to throw an NPE exception instead of a proper error message.

When validating operands of a set operation, if operands don't have
matching schema, and if one of the operand doesn't have a FROM clause,
SqlUtil#getSelectListItem throws NPE.

Fixing by adding a proper check.

* [CALCITE-2742] Read values of USER and SYSTEM_USER variables from DataContext (Siddharth Teotia, Jacques Nadeau)

Close apache#1031 (the PR used for this fix)
Close apache#973 (a previous PR, superceded by 1031)

* [CALCITE-3098] Upgrade SQLLine to 1.8.0

* [CALCITE-3102] Remove deprecation warnings following CALCITE-2969

* [CALCITE-2944] Deprecate Aggregate indicator and remove fields where possible

* Complete [CALCITE-2969] and [CALCITE-3102] by restoring APIs that use SemiJoin but mask deprecation warnings

Fix description of RelBuilder.semiJoin algebra.md.

* Following [CALCITE-2944] remove internal uses of Aggregate.indicator

These were generating deprecation warnings.

* Cosmetic changes

* [CALCITE-3106] Upgrade commons-pool2 from 2.6.0 to 2.6.2 (Fokko Driesprong)

Close apache#1248

* [CALCITE-3107] Upgrade commons-dbcp2 from 2.5.0 to 2.6.0 (Fokko Driesprong)

Close apache#1249

* [CALCITE-3005] Implement string functions: LEFT, RIGHT (xuqianjin)

Close apache#1168

* Following [CALCITE-2812] Disable parallel execution of parameterized test to avoid hanging

The patch of CALCITE-2812 introduced a parameterized test, but
parameterized JUnit test hangs when running in parallel mode due to
https://issues.apache.org/jira/browse/SUREFIRE-1430. We have to disable
parallel execution until the issue is fixed.

* [CALCITE-2804] Fix casting to timestamps in Druid

Close apache#1014

* [CALCITE-2744] RelDecorrelator use wrong output map for LogicalAggregate decorrelate (godfreyhe and Danny Chan)

godfreyhe started the work by apply a new map for LogicalAggregate
decorrelate register, and Danny Chan add shifts for constant keys
mapping.

Also fix the test case name and comments.

close apache#1254

* [CALCITE-3116] Upgrade to Avatica 1.15

Remove bug for CALCITE-2776 which was fixed in Avatica 1.15.
Remove bug for CALCITE-2993 which is fixed in Calcite but forgotten.

* [CALCITE-2721] Support parsing record-type [DOT] member-functions

Close apache#1011

* [CALCITE-2968] New AntiJoin relational expression

Close apache#1246

* [CALCITE-2822] Allow MultiJoin rules with any project/filter (Siddharth Teotia)

Close apache#1030

* Following [CALCITE-2822] simplify test and related classes/rules

I accidentally pushed CALCITE-2822 without these changes. Apologies for
the multiple commits.

* [CALCITE-3123] In RelBuilder, eliminate duplicate aggregate calls

* Following [CALCITE-3005], remove conflict warning in SQL parser

Alphabetize some lists in pom.xml.

Remove call to deprecated SqlAggFunction constructor.

* Following [CALCITE-2744] Remove usage of deprecated api in MockSqlOperatorTable

* Fix test exception caused by slightly different error message from regex in JDK 13

* [CALCITE-3087] AggregateOnProjectToAggregateUnifyRule ignores Project incorrectly when its Mapping breaks ordering (DonnyZone)

* [CALCITE-3119] Deprecate Linq4j CorrelateJoinType (in favor of JoinType)

* Bump version number for 1.20.0

* Release notes for 1.20.0

* Bump version for sanity check of properties

* [maven-release-plugin] prepare release calcite-1.20.0

* [maven-release-plugin] prepare for next development iteration

* Add 1.20.0 release announcement

* Add 1.20.0 release date

* Add committer names to 1.20.0 release notes

* Update example announcement

After sending the latest release announcement, the secretary indicated
that we should not link directly to closer.cgi, but instead to Calcite's
downloads page. Changing the sample announcement will save someone else
from future announcement rejection.

* [CALCITE-3148] Validator throws IndexOutOfBoundsException for SqlInsert when source and sink have non-equal fields number

* [CALCITE-3156] Mongo adapter. Replace fongo with Mongo Java Server for tests

mongo java server is a better alternative to fongo. All tests (including IT) are passing.

* [CALCITE-3157] Mongo java driver upgrade: 3.5.0 -> 3.10.2

* [CALCITE-3060] MutableProject should be generated based on INVERSE_SURJECTION mapping (DonnyZone)

* [CALCITE-3149] RelDataType CACHE in RelDataTypeFactoryImpl can't be garbage collected

* [CALCITE-2801] Check input type in AggregateUnionAggregateRule when remove the bottom Aggregate (Hequn Cheng)

Updated code to make sure to preserve the alias of Union, and add 1 more test
case for coverage by Haisheng Yuan.

Closes apache#1017

* [CALCITE-3151] RexCall's Monotonicity is not considered in determining a Calc's collation

* [CALCITE-3155] Empty LogicalValues can not be converted to UNION_ALL without operands which can not be unparsed(Musbah EL FIL)

Empty LogicalValues would be converted to
UNION_ALL Operator with no operands for rel to sql conversion,
which can not be unparsed to SQL string(invalid).

The fix converts the empty values to a "select NULL, NULL ...
from DUAL where 1 = 0"

Fix-ups(Danny Chan):
* Replace empty values check with Values.isEmpty(e)
* Always generate false condition as `1 = 0`

close apache#1286

* [CALCITE-3146] Support the detection of nested aggregations for JdbcAggregate in SqlImplementor.(Wenhui Tang)

* [CALCITE-3126] Remove deprecated SemiJoin usage completely

* [CALCITE-3133] Remove completely SemiJoinType

* [CALCITE-3125] Remove completely CorrelateJoinType

* [CALCITE-3152] Unify throws in sql parser

Replace generateParseException() and new ParseExecption to
SqlUtil.newContextException, this makes the error message more user
friendly.

* [CALCITE-3063] Parse and process PostgreSQL posix regular expressions

* [CALCITE-3168] Add test for invalid literal of sql parser

This patch also merge the redundant sql statements with
`SqlParserTest.Sql#sansCarets`.

* [CALCITE-3121] VolcanoPlanner hangs due to subquery with dynamic star

* [CALCITE-2460] [CALCITE-2459] Add TO_BASE64 / FROM_BASE64 SQL functions (Wenhui Tang)

* [CALCITE-3172] RelBuilder#empty does not keep aliases

* [CALCITE-3171] SemiJoin on conditions push down throws IndexOutOfBoundsException

RelOptUtil#pushDownJoinConditions should only project left side above
the join node when the join type only returns left side.

* [CALCITE-3169] decorrelateRel method should return when meeting SEMI/ANTI join in RelDecorrelator

* [CALCITE-3179] Bump Jackson from 2.9.8 to 2.9.9 (Fokko Driesprong)

* [CALCITE-3170] ANTI join on conditions push down generates wrong plan

Antijoin on conditions should not be pushed down to either left or right hand
side.

* [CALCITE-3160] Failed to materialize when the aggregate function uses group key (DonnyZone)

* [CALCITE-2995] Implement DAYNAME，MONTHNAME functions; add "locale" connection property (xuqianjin)

Add locale connection property (Julian Hyde).

Close apache#1163

* Fix complitation warnings after mongo java driver upgrade

Some methods were deprecated after client upgrade to 3.10.2 (see [CALCITE-3157])

* [CALCITE-3188] IndexOutOfBoundsException in ProjectFilterTransposeRule when executing SELECT COUNT(*)

* [CALCITE-3165] Project#accept(RexShuttle shuttle) does not update rowType

* [CALCITE-3189] Multiple fixes for Oracle SQL dialect

Close apache#1312

* [CALCITE-3183] During field trimming, Filter is copied with wrong traitSet (Juhwan Kim)

New Filter created after trimming uses its old trait set,
which could cause problem if there has been any update on collation.

Code changes:
- Add a FilterFactory method that includes correlating variables;
- Use RelBuilder when creating a new Filter in trimmer;
- Add test cases for the bug and new RelBuilder method.

Close apache#1309

* [CALCITE-3144] Add rule, AggregateCaseToFilterRule, that converts "SUM(CASE WHEN b THEN x END)" to "SUM(x) FILTER (WHERE b)"

Copied from Apache Druid's CaseFilteredAggregatorRule.

Some aggregate functions (e.g. SINGLE_VALUE, GROUPING, GROUP_ID) do
not allow filter, so we skip them in the rule.

* [CALCITE-3196] In Frameworks, add interface BasePrepareAction (a functional interface) and deprecate abstract class PrepareAction

* Change type of SqlStdOperatorTable.GROUPING field to public class

* [CALCITE-3145] RelBuilder.aggregate throws IndexOutOfBoundsException if groupKey is non-empty and there are duplicate aggregate functions

The cause is that [CALCITE-3123] did not handle the case of non-empty groupKey.

Enable RelBuilder.Config.dedupAggregateCalls by default.

* [CALCITE-3166] Make RelBuilder configurable

Add immutable class RelBuilder.Config, and a builder for it.

Disable elimination of duplicate aggregate calls (added in
[CALCITE-3123]), until [CALCITE-3145] is fixed.

* CALCITE-3187: Make decimal type inference overridable (Praveen Kumar)

Decimal return type inference for addition and modulus is hardcoded.

Add new methods to RelDataTypeSystem to allow implementers to override
behavior if needed. Deprecate existing methods for multiply and divide
in RelDataTypeFactory and move them over to RelDataTypeSystem for
consistency.

Close apache#1311

* Fix javadoc error

Fix a javadoc error in RelDataTypeFactory caused by a invalid signature
and introduced by the previous patch

* Merge branch 'master' of https://github.com/apache/calcite into Dm-calcite-Netezza

# Conflicts:
#	core/src/main/java/org/apache/calcite/rel/rel2sql/SqlImplementor.java
#	core/src/main/java/org/apache/calcite/sql/SqlDialect.java
#	core/src/main/java/org/apache/calcite/sql/SqlKind.java
#	core/src/main/java/org/apache/calcite/sql/dialect/SparkSqlDialect.java
#	core/src/test/java/org/apache/calcite/rel/rel2sql/RelToSqlConverterTest.java

* Dm-calcite-Netezza : calcite conformance changes

* Changes in relToSqlConverter

…ence issues (Steven Talbot)

RelToSqlConverter was converting each composite item in GROUPING SETS into
a SqlNodeList when it should be a call to the ROW operator.

Close apache#1239

…ence issues (Steven Talbot)

RelToSqlConverter was converting each composite item in GROUPING SETS into
a SqlNodeList when it should be a call to the ROW operator.

Close apache#1239