I stumbled on JCC, the Java->C++ (and Python) bridge that the PyLucene 
folk are using: http://svn.osafoundation.org/pylucene/trunk/jcc/jcc/README

Any thoughts on perhaps using that for the Hadoop family of projects?


Wes


Jonathan Gray wrote:
> I have not used the thrift interface, but know that many others are.  I
> doubt that they are seeing an order of magnitude decrease in performance.
> What have others seen with thrift?
> 
> There are talks of a new client-side API.  To my knowledge no one is working
> on this, so it is not currently scheduled for any upcoming release.  There
> are a number of people who would be interested in it, so please continue
> discussion on the issue if you have an interest in it.
> 
> Here is the open issue regarding this (Pure C/C++ client libraries):
> https://issues.apache.org/jira/browse/HBASE-1015
> 
> JG
>  
> 
>> -----Original Message-----
>> From: Slava Gorelik [mailto:slava.gorelik@gmail.com]
>> Sent: Tuesday, December 09, 2008 6:12 AM
>> To: hbase-user@hadoop.apache.org
>> Subject: Thrift API vs Java API
>>
>> Hi.Currently i'm trying to communicate with Hbase from .NET environment
>> and
>> the only option that i have is Thrift.
>> I did small comparison (execution time) between Java and Thrift (adding
>> 1000
>> rows with 1k data) and Java is 3.4 ms (average) vs. Thrift that is 44
>> ms
>> average.
>> It seems that Thrift option is not usable from performance aspect.
>>
>> Now he question : some one tried to implement Hbase client side in C++
>> /
>> .NET ?  Is it feasible ?
>>
>> Best Regards.
>