[ 
https://issues.apache.org/jira/browse/PIG-994?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Rekha updated PIG-994:
----------------------


Thanks Alan.

I am for 'option on store' mostly and definitely if they are exclusive 
possibilities.

However for arguments sake, a keyword approach can be considered, in addition.

This is because I am hoping append will open doors to be able to easily patch 
in update feature on similar lines into pig api, (and hopefully as part of same 
jira ticket)
My idea of update is a syntax like  "update <DS1> by (join_keys) from <DS2> by 
(join_keys) parallel $PARALLEL"
This will update dataset1(DS1) with data from dataset2(DS2) based on key joins.

{code}
update b by (jon_key1, join_key2) from c by (join_key1, join_key2); //this will 
update the DS b directly
//or alternatively
//x = update b by (jon_key1, join_key2) from c by (join_key1, join_key2); // 
making it two-step.
z = foreach b generate $0, $32, $50; // incase you are taking only few cols 
from main(b), new (c)
store z into 'bla' append; // appends the o/p data into 'bla' directly.
{code}

The append case, this below construct will be another way of doing it.
{code}
append b, c; // appends directly into b.
z = foreach b generate $0, $32, $50; // incase you are taking only few cols 
from main(b), new (c)
store z into 'bla';
{code}


> Provide 'append' keyword to allow appending to diferent dataset once the 
> feature is available in Hadoop
> -------------------------------------------------------------------------------------------------------
>
>                 Key: PIG-994
>                 URL: https://issues.apache.org/jira/browse/PIG-994
>             Project: Pig
>          Issue Type: New Feature
>          Components: impl
>    Affects Versions: 0.4.0
>         Environment: Grid clusters
>            Reporter: Rekha
>            Priority: Minor
>
> Provide 'append' keyword to allow appending to diferent dataset on pig 0.5.0 
> as it is now on hadoop 0.20(which has append feature)

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

Reply via email to