hadoop-pig-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Santhosh Srinivasan (JIRA)" <j...@apache.org>
Subject [jira] Commented: (PIG-788) Proposal to remove float from Pig data types
Date Tue, 12 May 2009 00:52:45 GMT

    [ https://issues.apache.org/jira/browse/PIG-788?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12708274#action_12708274
] 

Santhosh Srinivasan commented on PIG-788:
-----------------------------------------

-1 on this jira for the following reasons:

1. floats take 4 bytes as opposed to doubles that take 8 bytes
2. Floating point operations are much faster than operations on doubles
3. Issue of breaking backward compatibility at the cost of slower performance (and not faster
performance)
4. A storage layer should not dictate how a higher layer evolves.

> Proposal to remove float from Pig data types
> --------------------------------------------
>
>                 Key: PIG-788
>                 URL: https://issues.apache.org/jira/browse/PIG-788
>             Project: Pig
>          Issue Type: Improvement
>          Components: impl
>    Affects Versions: 0.2.0
>            Reporter: Alan Gates
>            Assignee: Alan Gates
>
> Pig would like to use the new Hadoop Avro serialization package to pass data between
MR jobs, and eventually between Pig and UDFs that are not written in Java.  Avro will not
be supporting the float data type, but only double (see AVRO-17).  Pig currently support both
float and double.  Double is the default floating point type (so if the user says x + 1.0,
1.0 is taken to be a double, not a float).  Float was initially included in the list of Pig
types because Hadoop supported it as one of the Writable types, and we were trying to make
sure all of Hadoop's writable types could be represented in Pig.  
> In practice we do not see anyone using the float type.   In order to be able to easily
use Avro I propose dropping the float type.  
> Please speak up if you are using the float type and you have a compelling reason not
to use double.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message