Change in precision for results obtained in Step2

[vinay]

Hi,

I have made code changes to run Step1 and Step2 of ccc-gistemp in
MapReduce (MR). When both Step1 and 2 run as MR job, I see that the
output of Step2 has slightly different temperature values than when
run separately. I believe its because, when the output of Step1 is
written to the file only 1 significant digit is used after the decimal
point. When it is re-read as the input to step2, there is loss of
precision (rounding error).

What I am doing is using the output of Step 1 with all its precision
as an input to Step2 within the MR job, without using intermediate
file in between. Is there something wrong in doing this? Can someone
please throw some light on this scenario?

Thanks,
VS