[
https://issues.apache.org/jira/browse/HBASE-1626?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-1626:
-------------------------
Attachment: deletes.patch
Doğacan had interesting idea where hbase subclassed the new hadoop context.
A lesser suggestion was to add a Marker interface to Delete and Put.
Playing around with it, a Marker interface would be a bit messy. It'd have to
be in the client package and be public. It'd have to be something like Update
or UpdateMarker (yuck). But maybe this is the way to go?
Otherwise, make the hbase reducer generic so can pass anything -- e.g. a Delete
or a Put -- and not specify types which might not be that bad. Here's why.
Looking at the TableOutputFormat, turns out the key is not used at all so no
need to specify a Type; i.e. ImmutableByteWritable. And for the value, we
could leave it as Writable. If not a Put or Delete, throw an IOE. Means we
lose some of power of generics.
This is what the attached patch does.
Lars, any chance you'd take a look? Do we even need a TableReducer if its all
generic types? I can see point of the TableMapper because value will be Result
(Here too, Key is redundant...). Is it wrong to specialize the
TableOutputFormat? Rather, leave it generic (though value must be a Delete or
a Put)?
> Allow emitting Deletes out of new TableReducer
> ----------------------------------------------
>
> Key: HBASE-1626
> URL: https://issues.apache.org/jira/browse/HBASE-1626
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Fix For: 0.20.0
>
> Attachments: deletes.patch
>
>
> Doğacan Güney (nutch) wants to emit Delete from TableReduce. Currently we
> only do Put.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
