[
https://issues.apache.org/jira/browse/TEZ-2959?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Siddharth Seth updated TEZ-2959:
--------------------------------
Attachment: TEZ-2959.1.wip.txt
The patch adds an interface which can be implemented by InputFormats.
InputFormats which are in the Hadoop project - Text, CombineFile etc, will not
be able to use this. That should not be a problem though since they have no use
of the InputContext.
Downstream InputFormats could however implement the interface and make use of
the InputCotnext. e.g. HiveInputFormat and OrcInputFormat.
[~hitesh] - this is a wip patch. Could you please take a look and provide some
early feedback.
An alternate would be to set a value unique to the Input (uniqueIdentifier) in
the configuration passed to the InputFormat. Also create a static access class
- something like MRInputInputContextAccess, which can provide inputContext
instances based on the unique IDs. This relies on statics however.
> Make InputContext available to InputFormats
> -------------------------------------------
>
> Key: TEZ-2959
> URL: https://issues.apache.org/jira/browse/TEZ-2959
> Project: Apache Tez
> Issue Type: Improvement
> Reporter: Siddharth Seth
> Assignee: Siddharth Seth
> Attachments: TEZ-2959.1.wip.txt
>
>
> Currently an InputFormat can only be provided with a Configuration instance,
> which very often is not adequate.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
