Nested Load Statement - Different Results Depending on KEEP placement

[Anonymous]

Consider this and where the KEEP statement is. This produces a table with 2.2million rows which "seems" correct.

If however I move the keep statement to the final load statement I get circa 750k rows.

IN the UI this results in slightly different claim_counters.

For the life of me I cannot think why this would be different. As you would expect placing the keep on the first load statement would simply mean that all rows would be dropped from T_CLAIM2 before the calculated fields were derived, whereas placing it last would mean the calculations would simply be done on all rows and then subsequently dropped.

Any ideas?

T_CLAIMS2:
LOAD *,
IF([Actual Date] = 'NA','$(vNaDesc)',IF(LEN([Actual Date])>2,'$(vClaimDesc)',IF(LEN([Measure Date])=0 OR [Measure Date] >=TODAY(),'$(vExpDesc)','$(vOsDesc)'))) AS [Measure Type]
;
LOAD *,
IF(LEN([Actual Date])>=2,[Actual Date],IF(PTYPE = 'IS Delivery' AND (msname = 'pre-production complete' OR msname = 'feasibility study required'),[Forecast Date],[Expected Date])) AS [Measure Date]
;
INNER KEEP(D_PROJECT)
LOAD
  msname,
  AUTONUMBER(msname,'%_MSID') AS %_MSID,
  %_PROJECTID,
  %_SITEID,
  [Expected Date],
  [Forecast Date],
  [Actual Date],
  PTYPE,
  TTYPE,
  %_TARGPROGID,
  1 AS #_CLAIM_COUNTER
RESIDENT T_CLAIMS;
DROP TABLE T_CLAIMS;

[swuehl]

Think I need to back to source and check my expected results.

Yes. And maybe check for NULL in your key field.