Informix 4gl details

To all users/developers of Informix 4gl 2.10 or higher: My company is still using Informix 1.10, and I am trying to determine if the following limitation in our compiler is limited to 1.10: DEFINE i INTEGER DEFINE my_arr ARRAY[100] OF RECORD LIKE my_table DECLARE my_cursor CURSOR FOR SELECT * FROM my_tableLET i = 1 FOREACH my_cursor INTO my_arr[i].* LET i = i + 1 END FOREACH This block of code will produce a run-time error if the table contains exactly 100 rows, because the Operating System attempts to calculate the addresses of my_arr[101].* before the database engine can justify ending the FOREACH loop with the non-existence of another row. Although this problem is not difficult to work around, it is annoying for 3 reasons: 1) The code above is so human-readable that any other solution is awfully cumbersome by comparison. 2) The reference manual is silent about the order of operations in a FOREACH loop, whereas most reference manuals easily plumb this level of detail to avoid off-by-one errors by their end-users. 3) The unnatural order of operations in the FOREACH loop is reminiscent of programming languages that eschew short-circuit evaluation of boolean conditions in favor of evaluating both sides of an AND or an OR essentially in parallel, thus leading to convoluted coding. Now for the questions: does any version of Informix 4gl later than 1.10 perform the FOREACH end-of-loop check by querying the database before calculating the destination memory addresses? -Richard H.