[
https://issues.apache.org/jira/browse/FLUME-2819?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14972127#comment-14972127
]
Mark Grover commented on FLUME-2819:
------------------------------------
I see what you mean, Roshan. I will start by saying that I have no horse in
this race, I don't work on Flume enough for me to have a strong opinion on this
particular case.
But in general, the flip argument to that (and a very strong one, in my
opinion) is that you test flume against a particular version of that artifact,
and that, of course, doesn't imply that it works with other versions of the
same artifacts. And, again if what you state is really a problem, that problem
is not unique to flume-kafka integration but to all dependencies of flume
including hadoop, hive, hbase, avro and then some.
Long story short, if you don't handle that complexity in flume, you are
trickling that down to the user, and I don't personally think that's a great
idea. You do have a legitimate point though. Perhaps, another idea would be to
have an out of box experience with a pre-tested version of kafka bundled in
Flume and if the user wants to change it, they source out a variable in the
environment (e.g. FLUME_KAFKA_HOME) which puts the custom kafka jar before the
pre-bundled flume kafka jar in the classpath. So users don't have to muck with
lib dir, they have out of box integration and an easy ability to custom
substitute the kafka jar if they don't like it. What do you think of that,
Roshan?
> Kafka libs are being bundled into Flume distro
> ----------------------------------------------
>
> Key: FLUME-2819
> URL: https://issues.apache.org/jira/browse/FLUME-2819
> Project: Flume
> Issue Type: Bug
> Reporter: Roshan Naik
>
> Kafka dependency libs need to be marked as 'provided' in the pom.xml
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
