You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Chesnay Schepler (Jira)" <ji...@apache.org> on 2022/11/03 12:05:00 UTC
[jira] [Commented] (FLINK-29866) Split flink-dist
[ https://issues.apache.org/jira/browse/FLINK-29866?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17628289#comment-17628289 ]
Chesnay Schepler commented on FLINK-29866:
------------------------------------------
Note that this would also solve the "flink-connector-base" problem, where we don't really want connectors to bundle it, but it's currently required for a good user-experience (as otherwise it will not work during testing without adding an additional dependency).
> Split flink-dist
> ----------------
>
> Key: FLINK-29866
> URL: https://issues.apache.org/jira/browse/FLINK-29866
> Project: Flink
> Issue Type: New Feature
> Components: Build System
> Reporter: Chesnay Schepler
> Assignee: Chesnay Schepler
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.17.0
>
>
> Users currently have to setup dependencies to various Flink dependencies (APIs like streaming-java, formats, connectors etc), and have to set the scope for each of these correctly w.r.t. what is actually provided by the distribution.
> This is unnecessarily tedious and error-prone, made worse by some special-cases like connector-files or the json/csv formats being bundled.
> I propose to create a new module {{flink-dist-pub}} (name pending!) that acts as a sort-of BOM for what is in the distribution (specifically the flink-dist jar) that is relevant for common DataStream application development.
> As follow-ups we could consider creating a table API extension of this module, and publish fat jars for the use-case of [~thw] in FLINK-29497.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)