[jira] [Created] (FLINK-17158) Watermark strategy property cannot be expressed in YAML

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[jira] [Created] (FLINK-17158) Watermark strategy property cannot be expressed in YAML

Shang Yuanchun (Jira)
Timo Walther created FLINK-17158:
------------------------------------

             Summary: Watermark strategy property cannot be expressed in YAML
                 Key: FLINK-17158
                 URL: https://issues.apache.org/jira/browse/FLINK-17158
             Project: Flink
          Issue Type: Bug
          Components: Table SQL / API, Table SQL / Ecosystem
            Reporter: Timo Walther


If the watermark strategy needs to be serialized into string-based properties for catalog storage, it is serialized into the following string-based properties:
{code}
DescriptorProperties properties = new DescriptorProperties();
                properties.putTableSchema("schema", schema);
                Map<String, String> actual = properties.asMap();
                Map<String, String> expected = new HashMap<>();
                expected.put("schema.0.name", "f0");
                expected.put("schema.0.data-type", "BIGINT");
                expected.put("schema.1.name", "f1");
                expected.put("schema.1.data-type", "ROW<`q1` VARCHAR(2147483647), `q2` TIMESTAMP(9)>");
                expected.put("schema.2.name", "f2");
                expected.put("schema.2.data-type", "VARCHAR(2147483647)");
                expected.put("schema.3.name", "f3");
                expected.put("schema.3.data-type", "BIGINT");
                expected.put("schema.3.expr", "f0 + 1");
                expected.put("schema.4.name", "f4");
                expected.put("schema.4.data-type", "DECIMAL(10, 3)");
                expected.put("schema.watermark.0.rowtime", "f1.q2");
                expected.put("schema.watermark.0.strategy.expr", "`f1`.`q2` - INTERVAL '5' SECOND");
                expected.put("schema.watermark.0.strategy.data-type", "TIMESTAMP(3)");
{code}

However, this design does not follow a proper hierarchy as both an array and watermarks are stored under a common `schema.` key. Thus, it is impossible to express those properties in JSON or YAML file:

{code}
tables:
  - name: PROD_RATES_HISTORY
    type: source
    update-mode: append
    schema:
    - name: RS_TIMESTAMP
      data-type: TIMESTAMP(3)
    - name: RS_SYMBOL
      data-type: STRING
    - name: DOUBLE
      data-type: RS_RATE
      ???watermark:
{code}



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