[
https://issues.apache.org/jira/browse/HIVE-29?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12652082#action_12652082
]
Ashish Thusoo commented on HIVE-29:
-----------------------------------
Some of the things that we had discussed at FB a while back that could be a
step in this direction include:
1. Support variables and variable substitution in Hive cli to support things
like @@logfile@@ as shown int the example above.
2. Support the ability to create data flows using data variables such as
mentioned in the example above. We also mentioned that it would be cool to
support such a syntax natively in a programming language like python, perl etc.
and provide client side libraries to compile such chains into optimal hive
queries. Note that writing the operations as shown in the example above it
another way of composing queries that appeals well to programmers as opposed to
nesting in the from clause that appeals well to SQL programmers. There is no
reason why these cannot be combined and together, so that the first approach
rewrites to nested SQL which is what hive understands. Key thing would be to do
this natively as python/perl - other libraries so programmers do not have to
learn yet another language...
> Add functionality similar to SCOPE's "virtual" views to Hive
> ------------------------------------------------------------
>
> Key: HIVE-29
> URL: https://issues.apache.org/jira/browse/HIVE-29
> Project: Hadoop Hive
> Issue Type: Bug
> Components: Query Processor
> Reporter: Jeff Hammerbacher
>
> SCOPE has many nice features, and the ability to IMPORT/EXPORT parameterized
> scripts and store partial queries in named variables is one of them. Section
> 3.5 of the SCOPE paper has the details, and there are several examples
> throughout the paper. Perhaps we can choose an alternative delimiter for
> PARAMTER imports, however (SCOPE uses "@@...@@").
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
