[
https://issues.apache.org/jira/browse/PIG-911?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12742239#action_12742239
]
Alan Gates commented on PIG-911:
--------------------------------
Dmitry,
First this is great. We've had requests to read Sequence files. Being able to
write them also would be great.
A few thoughts:
1) This should not extend UTF8StorageConverter. This loader will be returning
actual data types, not bytes that need to be interpreted. I would think
instead that it should implement the bytesToX() methods itself and just throw
an exception saying it didn't expect to do any conversion.
2) The getSampledTuple looks fine if skip is handling getting the stream to the
point that reading the next tuple is viable.
3) In the bindTo call, where you obtain the key and value by reflection, should
there be a try/catch block there in case the cast to Writable fails? In the
same way, in describe schema you're asking how to suppress warnings from the
cast in reader.getKeyClass(). But don't you want to check that what you got
really is a writable, since there is no guarantee?
> [Piggybank] SequenceFileLoader
> -------------------------------
>
> Key: PIG-911
> URL: https://issues.apache.org/jira/browse/PIG-911
> Project: Pig
> Issue Type: New Feature
> Reporter: Dmitriy V. Ryaboy
> Attachments: pig_sequencefile.patch
>
>
> The proposed piggybank contribution adds a SequenceFileLoader to the
> piggybank.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
