crunch-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Josh Wills (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CRUNCH-329) Re-add type info to TupleWritable to make fields sort correctly
Date Thu, 23 Jan 2014 21:41:39 GMT

    [ https://issues.apache.org/jira/browse/CRUNCH-329?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13880396#comment-13880396
] 

Josh Wills commented on CRUNCH-329:
-----------------------------------

Yeah, storing the key schema in the Configuration might work, since it's the only one we really
need to have access to during the shuffle phase. As you said, at that point we've basically
re-invented Avro for writables, except it would be slower since we couldn't do direct binary
comparisons using the schemas (well, I mean we could, but then we would have really-really
reinvented Avro.)

Since we're talking about re-inventing Avro anyway, we could theoretically force all shuffles
to be done in terms of Avro schemas. We have a hack in place for supporting Writables through
Avro, and of course all of the other primitive types have natural Avro correspondences. Not
sure which of these options is the least insane at this point. ;-)

> Re-add type info to TupleWritable to make fields sort correctly
> ---------------------------------------------------------------
>
>                 Key: CRUNCH-329
>                 URL: https://issues.apache.org/jira/browse/CRUNCH-329
>             Project: Crunch
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 0.10.0, 0.8.3
>            Reporter: Josh Wills
>            Assignee: Josh Wills
>             Fix For: 0.10.0, 0.8.3
>
>         Attachments: fix-ss-writables.patch
>
>
> Secondary sorts aren't currently working correctly for Writable types after we hacked
the TupleWritable impl to make all of the fields BytesWritables (e.g., secondary IntWritable
values will no longer be sorted correctly, even though everything is still grouped correctly.)
> The least-bad way that I came up with to fix this is to use integer codes for each possible
WritableComparable type in a pipeline that we can use to decode what Writable type each tuple
field corresponds to. This allows us to keep the various fields sortable while still doing
a reasonable job of minimizing the serialization required to pass the type information along.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message