You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-issues@hadoop.apache.org by "Konstantin Weitz (JIRA)" <ji...@apache.org> on 2014/01/04 18:40:52 UTC

[jira] [Updated] (MAPREDUCE-5708) Duplicate String.format in getSpillFileForWrite

     [ https://issues.apache.org/jira/browse/MAPREDUCE-5708?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Konstantin Weitz updated MAPREDUCE-5708:
----------------------------------------

    Description: 
The code responsible for formatting the spill file name (namely _gettSpillFileForWrite_) unnecessarily calls _String.format_ twice. This does not only affect performance, but leads to the weird requirement that task attempt ids cannot contain _%_ characters (because these would be interpreted as format specifiers in the outside _String.format_ call).

I assume this was done by mistake, as it could only be useful if task attempt ids contained _%n_.

  was:
The code responsible for formatting the spill file name (namely {code}getSpillFileForWrite{code}) unnecessarily calls {code}String.format{code} twice. This does not only affect performance, but leads to the weird requirement that task attempt ids cannot contain {code}%{code} characters (because these would be interpreted as format specifiers in the outside {code}String.format{code} call).

I assume this was done by mistake, as it could only be useful if task attempt ids contained {code}%n{code}.


> Duplicate String.format in getSpillFileForWrite
> -----------------------------------------------
>
>                 Key: MAPREDUCE-5708
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5708
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>            Reporter: Konstantin Weitz
>            Priority: Minor
>   Original Estimate: 10m
>  Remaining Estimate: 10m
>
> The code responsible for formatting the spill file name (namely _gettSpillFileForWrite_) unnecessarily calls _String.format_ twice. This does not only affect performance, but leads to the weird requirement that task attempt ids cannot contain _%_ characters (because these would be interpreted as format specifiers in the outside _String.format_ call).
> I assume this was done by mistake, as it could only be useful if task attempt ids contained _%n_.



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