ttOptSetColStats

Description

Modifies the statistics for the specified columns. This procedure allows an application to set statistics manually rather than have TimesTen automatically compute them. This feature is useful for preparing commands before the data has been inserted or for seeing how table characteristics can affect the choice of execution plan. This procedure modifies the relevant row(s) in the COL_STATS system table.

Because this procedure can be used before any data are in the table, the values specified do not need to bear any relation to the actual values, although some basic validity checking is performed.

Required privilege

This procedure requires no privilege (if owner) or ALTER ANY TABLE privilege (if not owner).

Syntax

ttOptSetColStats('tblName', 'colName', numUniq, minVal,maxVal, invalidate, numNull)

Parameters

ttOptSetColStats has these parameters:

Parameter	Type	Description
`tblName`	TT_CHAR(61) NOT NULL	Name of an application table. Can include table owner.
`colName`	TT_CHAR(30) NOT NULL	Name of a column in that table.
`numUniq`	TT_INTEGER NOT NULL	Number of unique values in the column.
`minVal`	VARBINARY(1024) NOT NULL	Minimum value in the column (possibly truncated).
`maxVal`	VARBINARY(1024) NOT NULL	Maximum value in the column (possibly truncated).
`invalidate`	TT_INTEGER	0 (no) or 1 (yes). If `invalidate` is 1, all commands that reference the affected tables are automatically prepared again when re-executed. This includes commands prepared by other users. If `invalidate` is 0, the statistics are not considered to have been modified and existing commands are not reprepared.
`numNull`	TT_INTEGER	Indicates the total number of NULLs in the column.

Result set

ttOptSetColStats returns no results.

Example

CALL ttOptSetColStats ( 'SALLY.ACCTS', 'BALANCE', 400, 0x00001388, 0x000186A0, 1, 0);

Notes

The minimum and maximum values need to be given as VARBINARY. NULL values are not permitted as minimum or maximum values. The value is stored in the platform-specific endian format.

The statistics are treated as a single interval of column values that are uniformly distributed between the minimum value and the maximum value.