st: update from long dataset using criteria

[Colin Campbell <col.campbell@rocketmail.com>]

Dear statalist,

I have a master dataset (1) which is a flat file with one record for each unique id. this contains missing data to be updated
from a second dataset (2) which is in long format with several observations per id.
I wish to update 1 with the result from 2 where the date difference between keydate(1) and test_date(2) is minimised.
i.e. for each unique id, update dataset 1 with result and test_date from dataset 2 when criteria X is satisfied

I have managed in dataset 2  to placer test_date in ascending order within each id, reshape wide and keep  only the first result 
and then merge, update  into the master to import the first result but cannot work out how to apply criteria to this.

Any help appreciated

Colin

dataset 1 

IdPacient  result    test_date            keydate
1                .                .                20/12/2010
2                .                .                01/01/1999

dataset 2
IdPacient    result    test_date            _n
1                1012      14/07/2010        2
1                1008      10/12/2010        3
1                 779       21/08/2009        1
2                 231       15/09/1999        3
2                195        01/06/1999        2
2                197        15/02/1999        1

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/