Applies To: SQL Server 2014, SQL Server 2016 Preview
Returns a result set that contains metadata about undeclared parameters in a Transact-SQL batch. Considers each parameter that is used in the @tsql batch, but not declared in @params. A result set is returned that contains one row for each such parameter, with the deduced type information for that parameter. The procedure returns an empty result set if the @tsql input batch has no parameters except those declared in @params.
Applies to: SQL Server (SQL Server 2012 through current version), Azure SQL Database.
- [ @tsql = ] 'Transact-SQL_batch'
One or more Transact-SQL statements. Transact-SQL_batch may be nvarchar(n) or nvarchar(max).
- [ @params = ] N'parameters'
@params provides a declaration string for parameters for the Transact-SQL batch, similarly to the way sp_executesql works. Parameters may be nvarchar(n) or nvarchar(max).
Is one string that contains the definitions of all parameters that have been embedded in Transact-SQL_batch. The string must be either a Unicode constant or a Unicode variable. Each parameter definition consists of a parameter name and a data type. n is a placeholder that indicates additional parameter definitions. If the Transact-SQL statement or batch in the statement does not contain parameters, @params is not required. The default value for this parameter is NULL.
The data type of the parameter.
sp_describe_undeclared_parameters always returns return status of zero on success. If the procedure throws an error and the procedure is called as an RPC, the return status is populated by the type of error as described in the error_type column of sys.dm_exec_describe_first_result_set. If the procedure is called from Transact-SQL, the return value is always zero, even in error cases.
sp_describe_undeclared_parameters returns the following result set.
int NOT NULL
Contains the ordinal position of the parameter in the result set. Position of the first parameter will be specified as 1.
sysname NOT NULL
Contains the name of the parameter.
int NOT NULL
Contains the system_type_id of the data type of the parameter as specified in sys.types.
For CLR types, even though the system_type_name column will return NULL, this column will return the value 240.
nvarchar (256) NULL
Contains the data type name. Includes arguments (such as length, precision, scale) specified for the data type of the parameter. If the data type is a user-defined alias type, the underlying system type is specified here. If it is a CLR user-defined data type, NULL is returned in this column. If the type of the parameter cannot be deduced, NULL is returned.
smallint NOT NULL
See sys.columns. for max_length column description.
tinyint NOT NULL
See sys.columns. for precision column description.
tinyint NOT NULL
See sys.columns. for scale column description.
For CLR and alias types, contains the user_type_id of the data type of the column as specified in sys.types. Otherwise is NULL.
For CLR and alias types, contains the name of the database in which the type is defined. Otherwise is NULL.
For CLR and alias types, contains the name of the schema in which the type is defined. Otherwise is NULL.
For CLR and alias types, contains the name of the type. Otherwise is NULL.
nvarchar (4000) NULL
For CLR types, returns the name of the assembly and class that defines the type. Otherwise is NULL.
Contains the xml_collection_id of the data type of the parameter as specified in sys.columns. This column will return NULL if the type returned is not associated with an XML schema collection.
Contains the database in which the XML schema collection associated with this type is defined. This column will return NULL if the type returned is not associated with an XML schema collection.
Contains the schema in which the XML schema collection associated with this type is defined. This column will return NULL if the type returned is not associated with an XML schema collection.
Contains the name of the XML schema collection associated with this type. This column will return NULL if the type returned is not associated with an XML schema collection.
bit NOT NULL
Returns 1 if the type being returned is XML and that type is guaranteed to be an XML document. Otherwise returns 0.
bit NOT NULL
Returns 1 if the column is of a case-sensitive string type and 0 if it is not.
bit NOT NULL
Returns 1 if the column is of a fixed-length CLR type and 0 if it is not.
bit NOT NULL
Returns 1 if the parameter is used anywhere other than left side of an assignment. Otherwise returns 0.
bit NOT NULL
Returns 1 if the parameter is used on the left side of an assignment or is passed to an output parameter of a stored procedure. Otherwise returns 0.
If the parameter is an argument to a stored procedure or a user-defined function, returns the name of the corresponding formal parameter. Otherwise returns NULL.
int NOT NULL
For internal use.
int NOT NULL
For internal use.
sp_describe_undeclared_parameters always returns return status of zero.
The most common use is when an application is given a Transact-SQL statement that might contain parameters and must process them in some way. An example is a user interface (such as ODBCTest or RowsetViewer) where the user provides a query with ODBC parameter syntax. The application must dynamically discover the number of parameters and prompt the user for each one.
Another example is when without user input, an application must loop over the parameters and obtain the data for them from some other location (such as a table). In this case, the application does not have to pass all the parameter information at once. Instead, the application can get all the parameters information from the provider and obtain the data itself from the table. Code using sp_describe_undeclared_parameters is more generic and is less likely to require modification if the data structure changes later.
sp_describe_undeclared_parameters returns an error in any of the following cases.
If the input @tsql is not a valid Transact-SQL batch. Validity is determined by parsing and analyzing the Transact-SQL batch. Any errors caused by the batch during query optimization or during execution are not considered when determining whether the Transact-SQL batch is valid.
If @params is not NULL and contains a string that is not a syntactically valid declaration string for parameters, or if it contains a string that declares any parameter more than one time.
If the input Transact-SQL batch declares a local variable of the same name as a parameter declared in @params.
If the statement creates any temporary tables.
If @tsql has no parameters, other than those declared in @params, the procedure returns an empty result set.
For a query with undeclared parameters, data type deduction for undeclared parameters proceeds in three steps.
The first step in data type deduction for a query with undeclared parameters is to find the data types of all the sub-expressions whose data types do not depend on the undeclared parameters. The type can be determined for the following expressions:
Columns, constants, variables, and declared parameters.
Results of a call to a user-defined function (UDF).
An expression with data types that do not depend on the undeclared parameters for all inputs.
For example, consider the query SELECT dbo.tbl(@p1) + c1 FROM t1 WHERE c2 = @p2 + 2. The expressions dbo.tbl(@p1) + c1 and c2 have data types, and expression @p1 and @p2 + 2 do not.
After this step, if any expression (other than a call to a UDF) has two arguments without data types, type deduction fails with an error. For example, the following all produce errors:
SELECT * FROM t1 WHERE @p1 = @p2 SELECT * FROM t1 WHERE c1 = @p1 + @p2 SELECT * FROM t1 WHERE @p1 = SUBSTRING(@p2, 2, 3)
The following example does not produce an error:
SELECT * FROM t1 WHERE @p1 = dbo.tbl(c1, @p2, @p3)
For a given undeclared parameter @p, the type deduction algorithm finds the innermost expression E(@p) that contains @p and is one of the following:
An argument to a comparison or assignment operator.
An argument to a user-defined function (including table-valued UDF), procedure, or method.
An argument to a VALUES clause of an INSERT statement.
An argument to a CAST or CONVERT.
The type deduction algorithm finds a target data type TT(@p) for E(@p). Target data types for the previous examples are as follows:
The data type of the other side of the comparison or assignment.
The declared data type of the parameter to which this argument is passed.
The data type of the column into which this value is inserted.
The data type to which the statement is casting or converting.
For example, consider the query SELECT * FROM t WHERE @p1 = dbo.tbl(@p2 + c1). Then E(@p1) = @p1, E(@p2) = @p2 + c1, TT(@p1) is the declared return data type of dbo.tbl, and TT(@p2) is the declared parameter data type for dbo.tbl.
If @p is not contained in any expression listed at the beginning of step 2, the type deduction algorithm determines that E(@p) is the largest scalar expression that contains @p, and the type deduction algorithm does not compute a target data type TT(@p) for E(@p). For example, if the query is SELECT @p + 2 then E(@p) = @p + 2, and there is no TT(@p).
Now that E(@p) and TT(@p) are identified, the type deduction algorithm deduces a data type for @p in one of the following two ways:
If E(@p) = @p and TT(@p) exists, i.e., if @p is directly an argument to one of the expressions listed at the beginning of step 2, the type deduction algorithm deduces the data type of @p to be TT(@p). For example:
SELECT * FROM t WHERE c1 = @p1 AND @p2 = dbo.tbl(@p3)
The data type for @p1, @p2, and @p3 will be the data type of c1, the return data type of dbo.tbl, and the parameter data type for dbo.tbl respectively.
As a special case, if @p is an argument to a <, >, <=, or >= operator, simple deduction rules do not apply. The type deduction algorithm will use the general deduction rules explained in the next section. For example, if c1 is a column of data type char(30), consider the following two queries:
SELECT * FROM t WHERE c1 = @p SELECT * FROM t WHERE c1 > @p
In the first case, the type deduction algorithm deduces char(30) as the data type for @p as per rules earlier in this topic. In the second case, the type deduction algorithm deduces varchar(8000) according to the general deduction rules in the next section.
If simple deduction does not apply, the following data types are considered for undeclared parameters:
Integer data types (bit, tinyint, smallint, int, bigint)
Money data types (smallmoney, money)
Floating-point data types (float, real)
numeric(38, 19) - Other numeric or decimal data types are not considered.
varchar(8000), varchar(max), nvarchar(4000), and nvarchar(max) - Other string data types (such as text, char(8000), nvarchar(30), etc.) are not considered.
varbinary(8000) and varbinary(max) - Other binary data types are not considered (such as image, binary(8000), varbinary(30), etc.).
date, time(7), smalldatetime, datetime, datetime2(7), datetimeoffset(7) - Other date and time types, such as time(4), are not considered.
CLR system-defined types (hierarchyid, geometry, geography)
CLR user-defined types
Of the candidate data types, any data type that would invalidate the query is rejected. Of the remaining candidate data types, the type deduction algorithm selects one according to the following rules.
The data type that produces the smallest number of implicit conversions in E(@p) is selected. If a particular data type produces a data type for E(@p) that is different from TT(@p), the type deduction algorithm considers this to be an extra implicit conversion from the data type of E(@p) to TT(@p).
SELECT * FROM t WHERE Col_Int = Col_Int + @p
In this case, E(@p) is Col_Int + @p and TT(@p) is int. int is chosen for @p because it produces no implicit conversions. Any other choice of data type produces at least one implicit conversion.
If multiple data types tie for the smallest number of conversions, the data type with greater precedence is used. For example
SELECT * FROM t WHERE Col_Int = Col_smallint + @p
In this case, int and smallint produce one conversion. Every other data type produces more than one conversion. Because int takes precedence over smallint, int is used for @p. For more information about data type precedence, see Data Type Precedence (Transact-SQL).
This rule only applies if there is an implicit conversion between every data type that ties according to rule 1 and the data type with the greatest precedence. If there is no implicit conversion, then data type deduction fails with an error. For example in the query SELECT @p FROM t, data type deduction fails because any data type for @p would be equally good. For example, there is no implicit conversion from int to xml.
If two similar data types tie under rule 1, for example varchar(8000) and varchar(max), the smaller data type (varchar(8000)) is chosen. The same principle applies to nvarchar and varbinary data types.
For purposes of rule 1, the type deduction algorithm prefers certain conversions as better than others. Conversions in order from best to worst are:
Conversion between same basic data type of different length.
Conversion between fixed-length and variable-length version of same data types (e.g., char to varchar).
Conversion between NULL and int.
Any other conversion.
For example, for the query SELECT * FROM t WHERE [Col_varchar(30)] > @p, varchar(8000) is chosen because conversion (a) is best. For the query SELECT * FROM t WHERE [Col_char(30)] > @p, varchar(8000) is still chosen because it causes a type (b) conversion, and because another choice (such as varchar(4000)) would cause a type (d) conversion.
As a final example, given a query SELECT NULL + @p, int is chosen for @p because it results in a type (c) conversion.
Requires permission to execute the @tsql argument.
The following example returns information such as the expected data type for the undeclared @id and @name parameters.
sp_describe_undeclared_parameters @tsql = N'SELECT object_id, name, type_desc FROM sys.indexes WHERE object_id = @id OR name = @name'
When the @id parameter is provided as a @params reference, the @id parameter is omitted from the result set and only the @name parameter is described.
sp_describe_undeclared_parameters @tsql = N'SELECT object_id, name, type_desc FROM sys.indexes WHERE object_id = @id OR NAME = @name', @params = N'@id int'