Informix select for update with order by - sql

Question

In ESQL/C I need to do a statement like this (I know that "order by" clause is not allowed in a "select ... for update")
SELECT FIRST 1 fd_foglio, fd_box_cod_soc, NVL(fd_id_subfoglio, 0)
FROM informix.foglioddt
WHERE
fd_box_data_all = TO_CHAR(CURRENT YEAR TO DAY, '%iy%m%d') AND
fd_box_cod_soc = '*'
ORDER BY fd_foglio DESC;
EXEC SQL
OPEN ifx_cursor_foglioddt_for_pending_crates;
EXEC SQL
FETCH ifx_cursor_foglioddt_for_pending_crates
INTO :fd_foglio, :fd_box_cod_soc, :local_fd_id_subfoglio;
in which I need an "order by" clause inside a "select for update" — I know that isn't allowed — in order to use a cursor successively to increment the field fd_id_subfogli (after I have used it in another part of the code) with a statement like:
UPDATE informix.foglioddt SET fd_id_subfoglio = :new_value WHERE
CURRENT OF ifx_cursor_foglioddt_for_pending_crates
How can I rewrite the previous code in order to select only the first ordered record and update a field in it?

Answer 1

The necessary information isn't in the question, but making a semi-educated guess, the foglioddt table has a UNIQUE constraint on columns fd_foglio and fd_id_subfoglio, but the latter can be NULL (in which case, Informix only allows one row in the table with a given fd_foglio value and a NULL fd_id_subfoglio).
Assuming that if there are any rows for a given fd_foglio value where the fd_id_subfoglio value is not null, then all the rows for that given fd_foglio value have a non-null value for the fd_id_subfoglio value, you could rephrase the UPDATE as:
UPDATE informix.foglioddt SET fd_id_subfoglio = :new_value
WHERE fd_foglio = :fd_foglio
AND (fd_id_subfoglio = :local_fd_id_subfoglio OR
fd_id_subfoglio IS NULL)
The first part of the OR condition deals with the normal case, where you've got both values that make up the primary key, and you simply change the identified row. The second part of the OR condition deals with the case where the only row in the table has fd_id_subfoglio set null.
Clearly, if my interpretation of the data structure is incorrect, then this won't work. The solution will then be to select the values from the primary key columns so that you can update using the primary key to identify the row to be updated — that's why primary keys are important! They're the way you can address individual rows by a set of values. Indexes on primary keys are also important to speed up such access (and to enforce uniqueness).