You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flink.apache.org by "Ron Crocker (JIRA)" <ji...@apache.org> on 2017/11/14 19:10:00 UTC
[jira] [Created] (FLINK-8074) Launch Flink jobs using Maven
coordinates
Ron Crocker created FLINK-8074:
----------------------------------
Summary: Launch Flink jobs using Maven coordinates
Key: FLINK-8074
URL: https://issues.apache.org/jira/browse/FLINK-8074
Project: Flink
Issue Type: Improvement
Reporter: Ron Crocker
Priority: Minor
As a Flink user, I want to be able to submit my job using the Maven coordinates (see https://maven.apache.org/pom.html#Maven_Coordinates) of its jar instead of a path to a local copy of that jar.
For example, instead of submitting my job using:
{{bin/flink run /local/path/to/word-count-1.0.1.jar <options>}}
I would specify it's Maven coordinates:
{{bin/flink run com.newrelic:word-count:1.0.1 <options>}}
This latter form would contact known Maven repositories to acquire the jar at the specified coordinates and submit that to the cluster.
Considerations:
* No transitive dependencies should be included - the target, specified either as a jar file in the local file system or by its maven coordinates, should be a complete Flink job.
* Maven repositories need to be specified _somewhere_. It's reasonable to expect that these repositories are independent of the cluster configurations.
* Specified repositories must meet the Maven API, but don't need to be Maven - artifactory, for example, is a valid repository as long as it meets the Maven API.
* _minor point_: _Indeterminate versions_ should be prohibited - that is, consider {{com.newrelic:word-count:+}} an invalid coordinate specification.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)