[
https://issues.apache.org/jira/browse/CRUNCH-581?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15045501#comment-15045501
]
John Leacox commented on CRUNCH-581:
------------------------------------
What are the thoughts around non-passive changes required to the public APIs of
crunch in order to remove Guava? I haven't looked through all uses of Guava
yet, but I have already found one public method that depends on the Supplier
class from Guava.
The below method on PTypes takes a SerializableSuppler instance, which is a
crunch extension of the Guava Supplier that adds the Serializable interface.
The Serializable interface is simple and could be duplicated easily, but it
would still require changes for consumers to update.
{code}
PTypes.protos(
Class<T> clazz,
PTypeFamily typeFamily,
SerializableSupplier<ExtensionRegistry> supplier)
{code}
> Remove dependency on Guava
> --------------------------
>
> Key: CRUNCH-581
> URL: https://issues.apache.org/jira/browse/CRUNCH-581
> Project: Crunch
> Issue Type: Improvement
> Components: Core
> Reporter: Micah Whitacre
> Assignee: Josh Wills
>
> As mentioned on CRUNCH0-356, we should look at removing all dependencies on
> Guava to help simplify Crunch's dependency tree and to not constrain
> consumers on the versions of Guava they want to consume.
> Looking at other projects like Spark, Hadoop, and HBase they all do not
> depend on Guava but have copied a class or two internally that they depend on.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
