How to generate variable names in multiple insert statement? - stored-procedures

Question

I have 18 different pairs of table column names like:
name_1, surname_1, ... name_18, surname_18
I would like to generate 18 inserts with Informix SPL using something like:
define _counter Int;
define _name_1 varchar(20);
define _surname_1 varchar(20);
...
define _name_18 varchar(20);
define _surname varchar(20);
select name_1, surname_1, ..., name_18, surname_18
into _name_1, _surname_1, ..., _name_18, _surname_18
from names where name_id = 1;
for _counter = 1 to 18 loop
insert into person(name, surname) values (_name_+_counter, _surname_+_counter);
end loop
If I try this I get syntax error. I am stuck with the terrible table design. Could you please advise if there is some similar/correct way of accomplishing this?

Answer 1

Given the clearer outline of the question, I think you have to forgo the loop. The best you can do is either 18 consecutive INSERT statements, or 18 calls to a stored procedure that executes one statement on each call.
Informix SPL does not have an array type, and you can only really use the loop with an array. (I have seen loops with a CASE statement inside, one case for each iteration of the loop; they're seldom a good solution to a problem, and it isn't a sensible solution to this situation.)
I will repeat an observation from my previous comments: the design of a table with 18 pairs of columns is very sub-optimal. However, it appears that you are trying to transfer data from this sub-optimal schema to a more sensible one with one row per name.
You could also consider using an 18-way UNION:
INSERT INTO Person(Name, Surname)
SELECT Name_1, Surname_1 FROM Names -- WHERE name_id = 1
UNION
SELECT Name_2, Surname_2 FROM Names -- …
UNION …
SELECT Name_18, Surname_18 FROM Names -- …
If the requirement is truly to have just the row where name_id = 1, you will need to add that criterion to each of the 18 SELECT clauses within the UNION SELECT statement. There are other ways to add that filter condition, with different sets of trade-offs at the source code level (and perhaps different trade-offs in the optimizer). Informix does not (yet) support CTEs (common table expressions, aka WITH clauses), which is a pity in this context.
Note that the code shown transfers all the data from Names into Person in a single SQL statement. This might well be the closest to optimal process overall.

Answer 2

Maybe something like this is what you want (using informix 12.10FC6, not sure if it will work on previous versions):
CREATE PROCEDURE copy_paste_names (p_name_id INTEGER);
DEFINE
l_query_string VARCHAR(255);
DEFINE
iter INT;
FOR iter IN (1 TO 18 STEP 1)
LET l_query_string = 'INSERT INTO person (name, surname) SELECT name_' || iter || ', surname_' || iter || ' FROM names WHERE name_id = ' || p_name_id || ';';
EXECUTE IMMEDIATE l_query_string;
END FOR;
END PROCEDURE;
I assume that the names table will always have 18 pairs of columns named name_? and surname_?.
This procedure will just blindly try to copy each pair of name_?, surname_? columns from the names table into a new row in the person table. There isn't any kind of checks to see if there are actually values to be copied or if