Question

我有一个简单的存储过程，其中target table和where子句条件被指定为参数。 “将nvarchar转换为数据类型数字的算术溢出错误”。错误让我有点沮丧。

@TableName varchar(50), -- 'A_RECORD_ETH0' test value

@Time_ms decimal(18,4), -- '40388629085.6481' test value

@Records int out,       -- should result with '1' if test values are used

这有效：

SELECT COUNT(*) as Count FROM A_RECORD_ETH0 WHERE Time_ms = 40388629085.6481

这有效：

SET @sql = N'SET @Records = (SELECT COUNT(*) FROM A_RECORD_ETH0 WHERE Time_ms = 40388629085.6481 )'

EXEC sp_executesql @sql, N'@Records int output', @Records output

这有效：

SET @sql = N'SET @Records = (SELECT COUNT(*) FROM '+ @TableName + ' WHERE Time_ms = 40388629085.6481 )'

EXEC sp_executesql @sql, N'@Records int output', @Records output

这会导致算术溢出错误：

SET @sql = N'SET @Records = (SELECT COUNT(*) FROM '+ @TableName + ' WHERE Time_ms = ' + @Time_ms + ' )'

EXEC sp_executesql @sql, N'@Records int output', @Records output

即使使用强制转换，也会导致算术溢出错误：

SET @sql = N'SET @Records = (SELECT COUNT(*) FROM '+ @TableName + ' WHERE Time_ms = Cast ( ' + @Time_ms + ' as decimal(18,4)) )'

EXEC sp_executesql @sql, N'@Records int output', @Records output

Answer 1

直接传递@Time_ms可避免使用decimal-＆gt; string-＆gt;十进制转换。它也更有效率;当@Time_ms改变时，服务器仍然可以重用执行计划而不是生成新计划。

SET @sql = N'SET @Records = (SELECT COUNT(*) FROM '+ @TableName + ' WHERE Time_ms = @Time_ms'

EXEC sp_executesql @sql, N'@Time_ms decimal(18,4), @Records int output', @Time_ms=@Time_ms, @Records output

Answer 2

我根据您的代码完成了以下工作：

DECLARE @sqlSelect nvarchar(500);
DECLARE @ParmSelect nvarchar(500);
SET @sqlSelect = N'SET @RecordsOut = (SELECT COUNT(*) FROM ' + @TableName + ' WHERE Time_ms = @Time_msIn)'
SET @ParmSelect = N'@Time_msIn decimal(18,4), @RecordsOut int OUTPUT';
EXEC sp_executesql @sqlSelect, @ParmSelect, @Time_msIn = @Time_ms, @RecordsOut = @Records OUTPUT;

如何修复“将nvarchar转换为数据类型数字的算术溢出错误”。错误？

2 个答案: