[
https://issues.apache.org/jira/browse/AVRO-314?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12799833#action_12799833
]
Doug Cutting commented on AVRO-314:
-----------------------------------
It seems circuitous to me to use Maven to publish things locally so that Ivy
can read them. Ivy should be able to publish locally to Ivy, no? I got this
to work by manually copying jars to my local ivy repo
(~/.ivy2/local/org.apache.hadoop/avro/1.2.0/jars/avro.jar) then modifying
Hadoop's "internal" resolvers in its ivysettings.xml files to include "local",
as in HADOOP-6486 and HDFS-892.
Have you tried using ivy's publish task to write Avro's jar to ivy's local
resolver and configuring Hadoop to use Ivy's local resolver to read it?
Other comments on the patch:
- you intermix two .gitignore styles: one at top-level or per-directory. pick
one. i prefer per-directory.
- downloaded jars should go in the lib/ directory, not in the ivy/ directory,
no?
- we use Ivy to generate Avro's pom file. if we must use Maven to publish,
can you please generate the pom this way? we should not maintain our
dependencies two places, ivy.xml and avro.pom.
> Allow Avro to publish to a local m2 repository
> ----------------------------------------------
>
> Key: AVRO-314
> URL: https://issues.apache.org/jira/browse/AVRO-314
> Project: Avro
> Issue Type: Improvement
> Components: build, java
> Reporter: Aaron Kimball
> Attachments: AVRO-314.patch
>
>
> To test Hadoop or other systems with a locally-built avro jar, it is
> considerably easier if you can use a local maven repository to resolve
> dependencies.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
