[
https://issues.apache.org/jira/browse/FLINK-10769?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Bowen Li updated FLINK-10769:
-----------------------------
Description:
In the Flink-Hive integration design, we propose a new FlinkInMemoryCatalog
(FLINK-10697) for production use. FlinkInMemoryCatalog will share some part
with the existing InMemoryExternalCatalog, thus we need to make changes to
InMemoryExternalCatalog.
Background: there are two parallel efforts going on right now - FLINK-10686,
driven by Timo, includes moving external catalogs APIs from flink-table to
flink-table-common, also from Scala to Java; FLINK-10744 I'm working on right
now to integrate Flink with Hive and enhance external catalog functionality.
As discussed with @twalthr in FLINK-10689, we'd better parallelize these
efforts while introducing minimal overhead for integrating them later. An
agreed way is to writing new code/feature related to external catalogs/hive in
Java in flink-table. This way, it will minimize migration efforts later to move
these new code into flink-table-common. If existing classes are modified for a
feature we can start migrating it to Java in a separate commit first and then
perform the actual feature changes, and migrated classes can be placed in
flink-table/src/main/java until we find a better module structure.
Therefore, we will port InMemoryExternalCatalog to java first. This PR only
port scala to java with NO feature or behavior change. This is a pre-requisite
for FLINK-10697
was:
In the Flink-Hive integration design, we propose a new FlinkInMemoryCatalog
(FLINK-10697) for production use. FlinkInMemoryCatalog will share some part
with the existing InMemoryExternalCatalog, thus we need to make changes to
InMemoryExternalCatalog.
As we are moving away from Scala to Java, we should write all new code/feature
in Java. Therefore, we will port InMemoryExternalCatalog to java first without
any feature or behavior change.
This is a pre-requisite for FLINK-10697
> port InMemoryExternalCatalog to java
> ------------------------------------
>
> Key: FLINK-10769
> URL: https://issues.apache.org/jira/browse/FLINK-10769
> Project: Flink
> Issue Type: Sub-task
> Components: Table API & SQL
> Reporter: Bowen Li
> Assignee: Bowen Li
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.8.0
>
>
> In the Flink-Hive integration design, we propose a new FlinkInMemoryCatalog
> (FLINK-10697) for production use. FlinkInMemoryCatalog will share some part
> with the existing InMemoryExternalCatalog, thus we need to make changes to
> InMemoryExternalCatalog.
> Background: there are two parallel efforts going on right now - FLINK-10686,
> driven by Timo, includes moving external catalogs APIs from flink-table to
> flink-table-common, also from Scala to Java; FLINK-10744 I'm working on right
> now to integrate Flink with Hive and enhance external catalog functionality.
> As discussed with @twalthr in FLINK-10689, we'd better parallelize these
> efforts while introducing minimal overhead for integrating them later. An
> agreed way is to writing new code/feature related to external catalogs/hive
> in Java in flink-table. This way, it will minimize migration efforts later to
> move these new code into flink-table-common. If existing classes are modified
> for a feature we can start migrating it to Java in a separate commit first
> and then perform the actual feature changes, and migrated classes can be
> placed in flink-table/src/main/java until we find a better module structure.
> Therefore, we will port InMemoryExternalCatalog to java first. This PR only
> port scala to java with NO feature or behavior change. This is a
> pre-requisite for FLINK-10697
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
