...

...

Discussion thread: http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/DISCUSS-FLIP-57-Rework-FunctionCatalog-td32291.html#a32613

JIRA: FLINK-14090

...

Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).

...

This FLIP would add explicit temporary function support by renaming a few variable names, and potential deprecating renaming some APIs in favor of new APIs that reflects their nature of dealing with temporary functions.

...

However, they need to be done simultaneously to achieve the functionality in full stack.

Proposed Changes

We will first clarify the dimensions of functions.

1. Support Two Types of Temporary Functions

...

• temporary system function that has no namespace and overrides built-in functions
• temporary catalog function that has catalog and database namespaces and overrides catalog functions.

...

Their DDLs are “CREATE/DROP /ALTER TEMPORARY SYSTEM FUNCTION”.

They will be renamed from “registerScalar/Table/AggregateFunctions()”.

b) Temporary Catalog Functions

We will add a new member variable to FunctionCatalog as “Map<ObjectIdentifier, UserDefinedFunction>  tempFunctions“ tempCatalogFunctions“ to hold those temporary catalog functions in a central place, and new APIs “registerTemporaryScalar/Table/AggregateFunction(ObjectIdentifier, UserDefinedFunction)”. 

These functions have to reside in an available catalog and database - their lifespans are tied to catalog and database. 

• Upon registration of such functions, we need to first confirm that the catalog and database exist
• Upon lookup, we also need to first confirm the assigned catalog and database exist. If so, return the function; otherwise, we drop all the temporary functions in either the assigned catalog or the database. Such lazy dropping approach is necessary because these temporary functions don’t live in their assigned catalogs, and they are not aware of when databases or catalogs are dropped. (FYI, we had a discussion of whether catalogs should support temporary functions within themselves in which case such unawareness problem won’t occur, but decided to not do so at the moment) 

Lifespan of temp functions are not tied to those of catalogs and databases. Users can create temp catalog functions even though catalogs/dbs in their fully qualified names don't even exist.

Their DDLs are “CREATE/DROP TEMPORARY FUNCTION”.

Some other proposed SQL commands are:

"SHOW FUNCTIONS" - list names of temp and non-temp system/built-in functions, and names of temp and catalog functions in the current catalog and db

"SHOW ALL FUNCTIONS" - list names of temp and non-temp system/built functions, and fully qualified names of temp catalog functions and catalog functions in all catalogs and dbs

"SHOW ALL TEMPORARY FUNCTIONS" - list fully qualified names of temp catalog functions in all catalog and db 

"SHOW ALL TEMPORARY SYSTEM FUNCTIONS" - list names of all temp system functionsTheir DDLs are “CREATE/DROP/ALTER TEMPORARY SYSTEM FUNCTION”.

Lifespan of both types of temporary functions will be within a session, and will destroyed upon session end.

Note: corresponding DDL and SQL commands are not part of this FLIP

2. Support Precise Function Reference

Because built-in system functions don’t have namespaces, a precise function reference in Flink must be either temporary catalog functions with namespaces or catalog functions.

The resolution order will be 

1. Temporary catalog functions with no namespace
2. Catalog functions

3. Support Ambiguous Function Reference with a Redefined Resolution Order

For ambiguous function reference, there are 4 types of functions to consider: temporary functions with and without no namespaces, Flink built-in system functions, and catalog functions.

...

1. Temporary system functions
2. Flink Built-in System functions
3. Temporary catalog functions, in the current catalog and current database of the session
4. Catalog functions, in the current catalog and current database of the session

The new resolution order will be a breaking change, compared to existing resolution order.

Temp functions should rank above their corresponding persistent/built-in functions due to its temporary nature - users want to overwrite built-in or persistent functions with something temporary that is only visible to themselves and the session, and not impacting other users. In contrary, 1) if users don’t have the intention of overwriting other functions, they can just name the temporary functions to something else, considering the manipulation cost is so low for temporary objects, and 2) if built-in functions precede temporary functions, there’s no way to reference temp functions anymore

Flink built-in System functions should precede catalog functions, because 1)  it always give a deterministic resolution order on ambiguous reference by invoking the built-in functions 2) catalog functions can always be precisely referenced with fully/partially qualified names. In contrary, if catalog functions precede built-in functions, built-in functions can never be referenced.

4. Code Changes to ObjectIdentifier

FunctionIdentifier

Class ObjectIdentifier {
	Optional<String> catalogName;

	Optional<String> databaseName;

	String objectName;
}

4. Changes to FunctionLookup interface

In order to adapt to the two cases, we propose updating FunctionLookup existing API to lookupFunction(ObjectIdentifier oi).

Calls from planner have to pass an ObjectIdentifier in which 

1. Neither catalog nor database names exist. The call will be routed to ambiguous functions reference logic 
2. Both catalog and database names exist, or only database name exist but not catalog name. The call will be routed to precise function reference lo

FunctionIdentifier {
    // for temporary/non-temporary system function

    // for temporary/non-temporary catalog function
	ObjectIdentifier oi;

    Optional<ObjectIdentifier> getIdentifier() {}
    Optional<String> getName() {}

    Optional<FunctionIdentifier> of(ObjectIdentifier oi) {}
    Optional<FunctionIdentifier> of(String name) {}
    
}

Changes to CallExpression and UnresolvedCallExpression

We should replace ObjectIdentifier with FunctionIdentifier in CallExpression and UnresolvedCallExpression. This would be a breaking change for these public evolving classes.

public final class CallExpression implements Expression {
	private final @Nullable FunctionIdentifier functionIdentifier;

    ...

    public Optional<FunctionIdentifier> getFunctionIdentifier() {
         return Optional.ofNullable(functionIdentifier);
    }
}

public final class UnresolvedCallExpression implements Expression {
	private final @Nullable FunctionIdentifier functionIdentifier;

    ...

	public Optional<FunctionIdentifier> getFunctionIdentifier() {
         return Optional.ofNullable(functionIdentifier);
    }
}

Changes to FunctionLookup interfaceOtherwise, it will throw IllegalArgumentException

In case of a partially qualified name (<db_name>.<function_name>) in SQL, planner/parser should recognize it with context be padding name of the current catalog of the session first to build an ObjectIdentifier.

...

/**
 * Resolves a function. ObjectIdentifer should either contain both catalog and database names, or neither of them.
 *
 * @param oi objectfi identifier of function
 * @return An optional result of FunctionLookup
 */
public Optional<FunctionLookup.Result> lookupFunction(ObjectIdentifierFunctionIdentifier oifi);

5. Changes to FunctionCatalog class

...

private final Map<String, FunctionDefintion> tempSystemFunctions = new LinkedHashMap<>();
private final Map<ObjectIdentifierMap<FunctionIdentifier, FunctionDefintion>FunctionDefinition> tempFunctions = new LinkedHashMap<>();

public void registerTemporarySystemScalarFunctionregisterTempSystemScalarFunction(String name, ScalarFunction function) {
// put into tempSystemFunctions
}

public void registerTemporarySystemTableFunctionregisterTempSystemTableFunction(String name, TableFunction function) {
// put into tempSystemFunctions

}

public void registerTemporarySystemAggregateFunctionregisterTempSystemAggregateFunction(String name, AggregateFunction function) {
// put into tempSystemFunctions

}

public void registerTemporaryScalarFunctionregisterTempCatalogScalarFunction(ObjectIdentifier oifoi, ScalarFunction function) {
// put into tempFunctions
}

public void registerTemporarySystemTableFunctionregisterTempCatalogTableFunction(ObjectIdentifier oifi, TableFunction function) {
// put into tempFunctions
}

public void registerTemporarySystemAggregateFunctionregisterTempCatalogAggregateFunction(ObjectIdentifier oifi, AggregateFunction function) {
// put into tempFunctions
}

public Optional<FunctionLookup.Result>void lookupFunctiondropTemporarySystemFunction(ObjectIdentifierString oiname) {
	if ((oi.getCatalogName().isPresent() && oi.getDatabaseName().isPresent())
|| (!oi.getCatalogName().isPresent() && oi.getDatabaseName}

public void dropTemporaryCatalogFunction(FunctionIdentifier fi) {}

public Optional<FunctionLookup.Result> lookupFunction(FunctionIdentifier fi) {
	if (fi.getObjectIdentifier().isPresent())) {
		// resolvePreciseFunctionReference(fi.getObjectIdentifier(oi));
	}

if (!oi.getCatalogName().isPresent() && !oi.getDatabaseName().isPresent()) else {
		resolveAmbiguousFunctionReference(oifi.getObjectNamegetName());
}

throw new IllegalArgumentException()/TableException();	}
}

private Optional<FunctionLookup.Result> resolvePreciseFunctionReference(ObjectIdentifierFunctionIdentifier oifi) {
	// resolve order:
	// 1. Temporary functions
	// 2. Catalog functions
}


private Optional<FunctionLookup.Result> resolveAmbiguousFunctionReference(String name);
	// resolve order:
	// 1. Temporary system functions
	// 2. Builtin functions
	// 3. Temporary functions, in the current catalog/db
	// 2. Catalog functions, in the current catalog/db
}

...