Thank for your reply^^
but question is how we can write map reduce in C# without using Java
abstract classes and interfaces of Hadoop, should we only
write C# code according to map/reduce app logic?
what do you mean of app can run standalone? you mean its should be .exe
did you do map/reduce by C# & Hadoop before?
Oh, lot of question. sorry ...
On 12/27/2011 06:13 PM, Harsh J wrote:
> I haven't used Mono but if your written program can run as a standalone and read from stdin and write to stdout, then streaming is sufficient to run your C# MR programs.
> On 27-Dec-2011, at 2:31 PM, Masoud wrote:
>> Dear All,
>> any one did it before: map/reduce by C#& Hadoop ???
>> As you know for developing map/reduce app in hadoop we should extend and implement special map and reduce abstract classes and interfaces,
>> and Hadoop pipes is for C++ not C#. The question is what we should do for C#?
>> *IS IT RIGHT THAT *
>> 1- Just develop our C# code (maybe its better with MonoDevelop) according to map/reduce abstract logic by developing map and reduce classes
>> 2- Then introduce the map and reduce classes to the hadoop streaming.
>> Does hadoop streaming work with .ddl file?
>> Thanks for your help.