On 19 Aug 2010, at 17:24, Stack wrote:

> @ Andrew:  Your HTable mock sounds interesting.  Do you think it
> others would find it useful and if so, would you consider contributing
> it?

I just used Mockito in my tests, captured the Puts etc and checked that they 
had the data I expected them to have. 

I was really only testing that I'd turned my objects into byte arrays and was 
putting them in the right place. I left the workings of HBase to it's self and 
just assumed it was working. 

Andrew

> 
> @Imran, on 2.), James Kennedy and his cohort are working on making
> THBase work w/ 0.89/0.90 hbase.  See their github repo for the latest
> state.
> 
> St.Ack
> 
> On Wed, Aug 18, 2010 at 11:10 PM, Andrew McCall
> <[email protected]> wrote:
>> 
>> On 19 Aug 2010, at 05:57, Imran M Yousuf wrote:
>> 
>>> Hi,
>>> 
>>> I need some clarification on the following issues -
>>> 
>>> 1. TransactionalTable seems to be present in the current API docs,
>>> does that mean it will be present in the course of 0.89 and continue
>>> to 0.90?
>>> 
>>> 2. Is there a way to start up a single node in memory HBase server
>>> programatically? I want to do it from a JUnit class to use it from
>>> within some tests. Where can I find some sample code in this regard?
>>> 
>> 
>> I can't offer any insight on the first, but I encountered the same issues as 
>> you in the second. In the end there wasn't anyway to bring a HBase instance 
>> up I felt was fast enough to be useful in tests.
>> 
>> I went with mock objects, which is probably better anyway since you 
>> shouldn't really need to test the HBase code in your unit tests.
>> 
>> I mocked HTable in my tests and just made sure I was putting what I expected 
>> for writes and returned what I expected HBase to return for reads.
>> 
>> Andrew

Reply via email to