Question

我是Postgres的新手。这是我的表job_defn 的结构和一些示例数据：

CREATE TABLE job_defn (
        job_id INTEGER NOT NULL,
        job_name CHARACTER VARYING(255) NOT NULL,
        system CHARACTER VARYING(50) NOT NULL,
        frequency CHARACTER VARYING(10) NOT NULL,
        run_day_id INTEGER NOT NULL,
        run_day_of_month INTEGER NOT NULL,
        eff_start_date DATE NOT NULL,
        eff_end_date DATE NOT NULL
    );

样本数据：

---------------------------
job_id         frequency
---------------------------
1001           DAILY 
1002           WEEKLY
---------------------------

运行以下函数时，我期望l_freq变量包含表中的frequency列值。但是，即使表行包含一个值，它也始终显示NULL：

CREATE OR REPLACE FUNCTION testing()
   RETURNS void AS $$
   DECLARE

       l_job_id INTEGER;
       l_job_defn_record job_defn%ROWTYPE;
       l_freq CHARACTER VARYING(10);
   BEGIN
       FOR l_job_defn_record IN 
               SELECT job_id, frequency FROM job_defn ORDER BY job_id ASC
       LOOP
               l_job_id := l_job_defn_record.job_id;
               l_freq := l_job_defn_record.frequency;
               raise info'job id: %,  frq: %', l_job_id, l_freq;
               raise info'row==>%', l_job_defn_record;

       END LOOP;
   END;
   $$ LANGUAGE plpgsql VOLATILE;

日志：

17:50:52  [CREATE - 0 row(s), 0.036 secs]  Command processed. No rows were affected
Code: 0 SQL State: 00000 --- job id: 10001,  frq: <NULL>
Code: 0 SQL State: 00000 --- row==>(10001,DAILY,,,,,,,,,,)
Code: 0 SQL State: 00000 --- job id: 10001,  frq: <NULL>
Code: 0 SQL State: 00000 --- row==>(10001,DAILY,,,,,,,,,,)
Code: 0 SQL State: 00000 --- job id: 10001,  frq: <NULL>
Code: 0 SQL State: 00000 --- row==>(10001,DAILY,,,,,,,,,,)
Code: 0 SQL State: 00000 --- job id: 10001,  frq: <NULL>
Code: 0 SQL State: 00000 --- row==>(10001,DAILY,,,,,,,,,,)
Code: 0 SQL State: 00000 --- job id: 10001,  frq: <NULL>
Code: 0 SQL State: 00000 --- row==>(10001,DAILY,,,,,,,,,,)
Code: 0 SQL State: 00000 --- job id: 10001,  frq: <NULL>
Code: 0 SQL State: 00000 --- row==>(10001,DAILY,,,,,,,,,,)
Code: 0 SQL State: 00000 --- job id: 10001,  frq: <NULL>
Code: 0 SQL State: 00000 --- row==>(10001,DAILY,,,,,,,,,,)
Code: 0 SQL State: 00000 --- job id: 10002,  frq: <NULL>
Code: 0 SQL State: 00000 --- row==>(10002,WEEKLY,,,,,,,,,,)
... 1 statement(s) executed, 0 row(s) affected, exec/fetch time: 0.036/0.000 sec  [0 successful, 1 warnings, 0 errors]

Answer 1

您将记录变量l_job_defn_record声明为类型job_defn%ROWTYPE。正如您稍后所阐明的，表job_defn实际上不仅包含您最初公开的两列。 FOR循环中的分配仅分配记录变量的 前两个字段 ：

...
   FOR l_job_defn_record IN 
      SELECT job_id, frequency FROM ...  -- wrong!
...

通过这种方式，来自SELECT查询（job_defn.frequency）的第二列被分配给记录变量（l_job_defn_record.job_name）的第二个字段-显然不是预期的。由于frequency的类型也是varchar，因此它可以工作。记录的尾部字段未分配，默认为NULL。

这将解决该问题：

...
   FOR l_job_defn_record IN 
      SELECT * FROM job_defn ORDER BY job_id
...

SELECT *通常是可疑的编码，但是在这种情况下，这是正确的方法，因为根据定义，l_job_defn_record具有与表job_defn相同的行类型。

Answer 2

Erwin在回答中清楚地说明了错误的原因。我建议对变量使用the RECORD type而不是ROWTYPE。

    l_job_defn_record RECORD;

此类变量的实际结构是在为其分配值时动态定义的。因此，变量和赋值之间不存在不兼容的可能性。

Postgres在FOR循环中访问行值

2 个答案: