unless you have large amounts of data, I'd go quick and dirty: untested

CREATE TABLE acmast (accode n(6), acname c(50),opamt n(10,2))
>SELECT acmast
INSERT INTO acmast (accode,acname,opamt) VALUES (1, 'party name', 54200)

CREATE TABLE sales (idno n(6), billno c(6), billdt d, accode n(6), qty n(7,2), amt n(10,2))
INSERT INTO sales (idno, billno , billdt , accode , qty , amt ) ;
VALUES (1, 'aa/101', DATE(2011,4,04),1,22,509)
INSERT INTO sales (idno, billno , billdt , accode , qty , amt ) ;
VALUES (1, 'aa/102', DATE(2011,4,14),1,28,852)
INSERT INTO sales (idno, billno , billdt , accode , qty , amt ) ;
VALUES (1, 'aa/103', DATE(2011,4,24),1,18,654)
>
SELECT sales.billno,sales.billdt, sales.accode, acmast.acname, acmast.opamt, sales.qty, sales.amt, cast(0 as I) as idTemp ;
FROM sales ;
LEFT JOIN acmast ON sales.accode=acmast.accode INTO CURSOR tempcurs READWRITE 

update tempcurs set idTemp = recno()
select acname, min(idTemp) from tempCurs into filtCurs group by 1
update tempcurs set opamt = 0 where idTemp not in ;
  Select idTemp from FiltCurs 

depending on the backend first selecting opamt as 0 and then setting only 1 line might be faster even if the update statement involves one more join. experiment ;-)