[jira] [Created] (FLINK-22404) Document inconsistencies around +I and +U generation in retract streams

[Shang Yuanchun (Jira)]

Nico Kruber created FLINK-22404:
-----------------------------------

             Summary: Document inconsistencies around +I and +U generation in retract streams
                 Key: FLINK-22404
                 URL: https://issues.apache.org/jira/browse/FLINK-22404
             Project: Flink
          Issue Type: Bug
          Components: Documentation, Table SQL / API
    Affects Versions: 1.12.2
            Reporter: Nico Kruber


Filters can remove +I messages and let a new row start with +U; similarly, they can remove the +U after -U which should instead have been reported as -D. This makes the semantics described on [https://ci.apache.org/projects/flink/flink-docs-release-1.12/dev/table/streaming/dynamic_tables.html#table-to-stream-conversion] wrong which makes users believe they can rely on +U vs. +I and -U vs. -D.
Either this needs to be fixed, or a disclaimer should be added to the respective paragraphs that +U/+I and -U/-D should be seem as the same.

 

Example:
{code:sql}
CREATE TEMPORARY VIEW test_cardinality AS
SELECT * FROM ( VALUES
  ('Alice', '[hidden email]', ARRAY [ '[hidden email]' ], 'Test Ltd'),
  ('Alice', '[hidden email]', ARRAY [ '[hidden email]' ], 'Test Ltd'),
  ('Alice', '[hidden email]', ARRAY [ '[hidden email]', '[hidden email]' ], 'Test Ltd'))
AS t ( name, email, aliases, company );
{code}

{code:sql}
SELECT
  name,
  LEAD(CAST(company AS VARCHAR), 0) AS company
FROM test_cardinality
WHERE CARDINALITY(aliases) >= 2
GROUP BY name;
{code}

vs.

{code:sql}
SELECT
  name,
  LEAD(CAST(company AS VARCHAR), 0) AS company
FROM test_cardinality
GROUP BY name
HAVING COUNT(DISTINCT email) >= 2;
{code}



--
This message was sent by Atlassian Jira
(v8.3.4#803005)