Question

我正在使用PipelineDB 0.9.7u3

我使用连续视图输出流进行了一些操作，以确定是否可以通过一些更新获得新的连续视图。

这是我的测试用例。

CREATE STREAM stream_test(ticketid text, val int, status text);

-- simple continuous view on stream_test
CREATE CONTINUOUS VIEW cv_test AS 
  SELECT 
    ticketid, 
    min(val) as v0, 
    keyed_min(val, status) as v0_status
  FROM stream_test 
  GROUP BY ticketid;

-- continuous view to keep cv_test's updates and insertions
CREATE CONTINUOUS VIEW cv_test_upin AS 
  SELECT 
    (new).ticketid, 
    (old).v0 as oldV0, 
    (old).v0_status as oldV0Status,
    (new).v0 as newV0, 
    (new).v0_status as newV0Status    
  FROM output_of('cv_test')

-- continuous view to keep just some cv_test's updates
CREATE CONTINUOUS VIEW cv_test_up AS 
  SELECT 
    (new).ticketid, 
    (old).v0 as oldV0, 
    (old).v0_status as oldV0Status,
    (new).v0 as newV0, 
    (new).v0_status as newV0Status    
  FROM output_of('cv_test')
  WHERE (old).v0 != (new).v0;

让我们提供一些数据。

 INSERT INTO stream_test VALUES 
 ('t1', 124, 'open'),
 ('t2', 190, 'pending')

正如所料：

select * from cv_test;

＆＃34; T2＆＃34 ;; 190;＆＃34;未决＆＃34;

＆＃34; T1＆＃34 ;; 124;＆＃34;开＆＃34;

select * from cv_test_upin;

＆＃34; T2＆＃34 ;;;＆＃34;＆＃34 ;; 190;＆＃34;未决＆＃34;

＆＃34; T1＆＃34 ;;;＆＃34;＆＃34 ;; 124;＆＃34;开＆＃34;

select * from cv_test_up;

然后，一些更新。

 INSERT INTO stream_test VALUES 
 ('t2', 160, 'waiting'),
 ('t1', 100, 'pending')

正如所料：

select * from cv_test;

＆＃34; T2＆＃34 ;; 160;＆＃34;等待＆＃34;

＆＃34; T1＆＃34; 100;＆＃34;未决＆＃34;

select * from cv_test_upin;

＆＃34; T2＆＃34 ;;;＆＃34;＆＃34 ;; 190;＆＃34;未决＆＃34;

＆＃34; T1＆＃34 ;;;＆＃34;＆＃34 ;; 124;＆＃34;开＆＃34;

＆＃34; T2＆＃34 ;; 190;＆＃34;未决＆＃34 ;; 160;＆＃34;等待＆＃34;

＆＃34; T1＆＃34 ;; 124;＆＃34;开＆＃34; 100;＆＃34;未决＆＃34;

select * from cv_test_up;

＆＃34; T2＆＃34 ;; 190;＆＃34;未决＆＃34 ;; 160;＆＃34;等待＆＃34;

＆＃34; T1＆＃34 ;; 124;＆＃34;开＆＃34; 100;＆＃34;未决＆＃34;

现在，一些新数据和一些更新。

 INSERT INTO stream_test VALUES 
 ('t2', 90, 'spam'),
 ('t3', 140, 'open'),
 ('t1', 80, 'closed')

select * from cv_test;按预期返回，但select * from cv_test_upin;没有。

...

＆＃34; T2＆＃34 ;; 160;＆＃34;等待＆＃34 ;; 90;＆＃34;垃圾邮件＆＃34;

＆＃34; T3＆＃34 ;;;＆＃34;＆＃34 ;; 140;＆＃34;开＆＃34;

＆＃34; T1＆＃34 ;; 80;＆＃34;关闭＆＃34 ;; 80;＆＃34;关闭＆＃34;

我期待最后＆＃34; t1＆＃34;是"t1";100;"pending";80;"closed"

错误或预期的行为？

感谢。

Answer 1

在深入研究之后，您确实发现了一些意想不到的行为，而且很可能是一个错误。我们将很快解决它，这是问题所在：

https://github.com/pipelinedb/pipelinedb/issues/1797

解决后，我们将发布更新版本。

PipelineDB：连续视图输出流意外地显示相同（旧）和（新）值

1 个答案: