TABLE FILE CustomerRegistration
SUM COMPUTE CNT/I10 = CNT.CustomerId
BY CustomerId
WHERE TOTAL CNT EQ 1;
END

select CustomerId
  from CustomerRegistration R1
 where not exists (
   select 1
     from CustomerRegistration R2
    where R2.Customerid = R1.CustomerId and R2.RegistrationDate <> R1.RegistrationDate
)

quote:

Originally posted by Wep5622:

TABLE FILE CustomerRegistration
SUM COMPUTE CNT/I10 = CNT.CustomerId
BY CustomerId
WHERE TOTAL CNT EQ 1;
END

Is there no other field to distinguish multiple Registration records for the same customer? You seem to not actually be interested in the counts, but rather in the non-existence of other records for the same customer.

Something like:

select CustomerId
  from CustomerRegistration R1
 where not exists (
   select 1
     from CustomerRegistration R2
    where R2.Customerid = R1.CustomerId and R2.RegistrationDate <> R1.RegistrationDate
)

The principle is called a "correlated subquery". It is usually a faster query than one using COUNT, as COUNT has to visit every record per customer, while NOT EXISTS can stop looking as soon as a 2nd record is encountered.

WebFOCUS has an equivalent for that in the DB_INFILE function, where EXISTS can be tested using WHERE DB_INFILE(...) EQ 1; and NOT EXISTS using WHERE DB_INFILE(...) EQ 0;.
Leaving the equality comparison out translates to WHERE EXISTS too.

...
WHERE FIELD NE LAST FIELD THEN 1 ELSE CNTR + 1 ;
...