Question

我有一个存储对象的数据库。我有以下（简化）架构

CREATE TABLE MyObjects
(
  UniqueIdentifier Id;
  BigInt           GenerationId;
  BigInt           Value;
  Bit              DeleteAction;
)

每个对象都有一个唯一标识符（“Id”）和一个（一组）属性（“Value”）。每次更改对象的属性值时，我都会在此表中输入一个新行，其中包含一个新生成ID（“GenerationId”，它会单调递增）。如果一个对象被删除，那么我通过将“DeleteAction”位设置为true来记录这个事实。

在任何时间点（代），我想检索所有活动对象的状态！

以下是一个例子：

Id    GenerationId Value  DeleteAction
1        1          99       false
2        1          88       false
1        2          77       false
2        3          88       true

几代人的对象是：

  1: 1 {99}, 2 {88}
  2: 1 {77}, 2 {88}
  3: 1 {77}

关键是：如何找出生成ID最接近（但不超过）给定代ID 的每个唯一对象的行？然后，我可以执行后过滤步骤以删除DeleteAction字段为true的所有行。

Answer 1

这适用于MS SQL

SELECT id,value
FROM Myobjects
INNER JOIN ( 
     SELECT id, max(GenerationID) as LastGen 
     FROM MyObjects
     WHERE GenerationID <= @Wantedgeneration
     Group by ID)
    On GenerationID = LastGen
WHERE DelectedAction = false

Answer 2

我的版本使用表MyObjects与a的联合自身的子集，由子查询创建，仅包含最后一个每个对象的生成：

SELECT O.id,generation,value FROM 
     MyObjects O, 
     (SELECT id,max(generation) AS max_generation FROM MyObjects 
     WHERE generation <= $GENERATION_ID GROUP BY id) AS TheMax WHERE 
            TheMax.max_generation = generation AND O.deleted is False
     ORDER BY generation DESC;

在上面的查询中，GENERATION_ID是硬连线的。一种方式参数化就是写一个函数：

CREATE OR REPLACE FUNCTION generation_objects(INTEGER) RETURNS SETOF MyObjects AS
  'SELECT O.id,generation,value,deleted FROM 
       MyObjects O, 
       (SELECT id,max(generation) AS max_generation FROM MyObjects 
       WHERE generation <= $1 GROUP BY id) AS TheMax WHERE 
              TheMax.max_generation = generation AND O.deleted is False;'
  LANGUAGE SQL;

现在，它有效。使用此表：

> SELECT * FROM MyObjects;          
 id | generation | value | deleted 
----+------------+-------+---------
  1 |          1 |    99 | f
  2 |          2 |    88 | f
  1 |          3 |    77 | f
  2 |          4 |    88 | t
  3 |          5 |    33 | f
  4 |          6 |    22 | f
  3 |          7 |    11 | f
  2 |          8 |    11 | f

我明白了：

> SELECT * FROM generation_objects(1) ORDER by generation DESC;
 id | generation | value | deleted 
----+------------+-------+---------
  1 |          1 |    99 | f

> SELECT * FROM generation_objects(2) ORDER by generation DESC;
 id | generation | value | deleted 
----+------------+-------+---------
  2 |          2 |    88 | f
  1 |          1 |    99 | f

> SELECT * FROM generation_objects(3) ORDER by generation DESC;
 id | generation | value | deleted 
----+------------+-------+---------
  1 |          3 |    77 | f
  2 |          2 |    88 | f

然后，在下一代中，对象2被删除：

> SELECT * FROM generation_objects(4) ORDER by generation DESC;
 id | generation | value | deleted 
----+------------+-------+---------
  1 |          3 |    77 | f

Answer 3

这是工作版本：

SELECT MyObjects.Id,Value
FROM Myobjects
INNER JOIN 
(      
  SELECT Id, max(GenerationId) as LastGen
  FROM MyObjects
  WHERE GenerationId <= @TargetGeneration
  Group by Id
) T1
ON MyObjects.Id = T1.Id AND MyObjects.GenerationId = LastGen
WHERE DeleteAction = 'False'

Answer 4

不确定这是否是标准SQL，但在Postgres中，您可以使用LIMIT标志：

 select GenerationId,Value,DeleteAction from MyObjects 
    where Id=1 and GenerationId < 3 
    order by GenerationId
    limit 1;

版本控制算法

4 个答案: