You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "Jingsong Lee (JIRA)" <ji...@apache.org> on 2019/02/21 03:51:00 UTC

[jira] [Created] (FLINK-11701) Introduce an abstract set of data formats

Jingsong Lee created FLINK-11701:
------------------------------------

             Summary: Introduce an abstract set of data formats
                 Key: FLINK-11701
                 URL: https://issues.apache.org/jira/browse/FLINK-11701
             Project: Flink
          Issue Type: New Feature
          Components: Table API &amp; SQL
            Reporter: Jingsong Lee
            Assignee: Jingsong Lee


Blink uses an abstract set of data formats to make internal calculations use the binary format as much as possible. This minimizes the serialization overhead and java object overhead.

It includes:

BaseRow <=> Row

BaseMap <=> Java Map

BaseArray <=> Java array

BaseString  <=> Java String

Decimal <=> BigDecimal  //Scale of this object is specified by the user, not automatically determined(like BigDecimal).

int <=> Date //Flink used to use int in the calculation, but the remaining in Row is still Date, we will change it completely.

int <=> Time

long <=> Timestamp

byte[] <=> byte[]

BaseGeneric <=> T (GenericRelDataType, we don't know it, let user define serializer)

primitive type keep same, but use less boxed type.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)