tree c82df1607a185ceafe973e0abb0b40f2fe4d0a63
parent 15f8bae2c1a346a760d93fa09e9d8b15bd6b6e63
author ramangrover29 <ramangrover29@123451ca-8445-de46-9d55-352943316053> 1286931410 +0000
committer ramangrover29 <ramangrover29@123451ca-8445-de46-9d55-352943316053> 1286931410 +0000

1) All Hadoop* operators now work with an instance of JobConf instead of working with 
   a hashmap. This is essential because JobConf has default settings (values) for non-configured keys. Using a map returns null values when using the JobConf could return valid defaults. 

2) All Hadoop* operators now set class loader for the JobConf instance they work with.
   This is required as the default class loader used inside JobConf cannot load application specific classes.

3) Hadoop Mapper needs to know specifics of the input split it is mapping. This is required to provide the similar environment as provided by Hadoop. Hence configuring a mapper now requires the input splits. At run time, the specific input split is chosen and its properties are set in the job conf before the mapper implementation is confgured.
   
3) Renamed HDFSWriteOperatorDescriptor to HadoopWriteOperatorDescriptor


git-svn-id: https://hyracks.googlecode.com/svn/trunk@144 123451ca-8445-de46-9d55-352943316053