[
https://issues.apache.org/jira/browse/HIVE-9172?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14271943#comment-14271943
]
Jason Dere commented on HIVE-9172:
----------------------------------
Haven't really taken a detailed look yet, but a few questions:
- This changes a lot of the existing methods for the Lazy SerDe classes, which
affects a lot of the serializers in Hive and possibly 3rd party SerDes. Are
these interfaces something we can change like this, or do we need to
keep/deprecate the old methods for backward compatibility? If this does break
backward compatbility, what work would be involved to transition to the new API
- would it simply be the change from byte to byte[] for delimiters?
- This patch removes MultiDelimitSerde. Again on the subject of backward
compatibility, might be better to deprecate, then remove in a later release. If
a user created a table with MultiDelimitSerde and then cannot use the table
with the new version of Hive because it no longer contains this
MultiDelimitSerde, would they be able to switch the table SerDe to
LazySimpleSerDe and have the table work as it was before?
- HIVE-5871 mentions that unlike LazySimpleSerDe, MultiDelimitSerde will not
base64 encode binary data due to some concerns about the multi-byte delimiter.
Do the concerns here also affect LazySimpleSerDe if it uses multi-byte
delimiters?
> Merging HIVE-5871 into LazySimpleSerDe
> --------------------------------------
>
> Key: HIVE-9172
> URL: https://issues.apache.org/jira/browse/HIVE-9172
> Project: Hive
> Issue Type: Bug
> Components: Serializers/Deserializers
> Reporter: Navis
> Assignee: Navis
> Priority: Minor
> Attachments: HIVE-9172.1.patch.txt, HIVE-9172.2.patch.txt,
> HIVE-9172.3.patch.txt
>
>
> Merging multi character support for field delimiter to LazySimpleSerDe
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
