Mysql Trigger will not pass last_insert_id() to connection - mysql

Question

This is my schema:
I am trying to have an insert into "desktops" or "laptops" insert an id generated automatically from "computers". That works.
My issue is when I insert into either table, I can not select last_insert_id();
Is there something I am doing wrong? I am trying to pass the id all the way forward to my application, for further processing. Selecting MAX(id) is not a valid solution. My SQL connection makes one insert statement, and the trigger should not break that functionality...
Use test;
CREATE TABLE `laptops` (
`id` int(11) NOT NULL,
`name` varchar(45) DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=innodb DEFAULT CHARSET=utf8;
CREATE TABLE `desktops` (
`id` int(11) NOT NULL,
`name` varchar(45) DEFAULT NULL,
PRIMARY KEY (`id`),
) ENGINE=innodb DEFAULT CHARSET=utf8;
CREATE TABLE `computers` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`type` varchar(45) DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=innodb DEFAULT CHARSET=utf8;
CREATE TRIGGER `laptops_BINS` BEFORE INSERT ON `laptops` FOR EACH ROW
BEGIN
IF (EXISTS(SELECT id FROM laptops WHERE name = NEW.name)) THEN
SET NEW.id = NULL;
ELSE
INSERT INTO computers (type) VALUES ('laptop');
SET NEW.id = LAST_INSERT_ID();
SET NEW.id = LAST_INSERT_ID(NEW.id);
END IF;
END
CREATE TRIGGER `desktop_BINS` BEFORE INSERT ON `desktops` FOR EACH ROW
BEGIN
IF (EXISTS(SELECT id FROM desktops WHERE name = NEW.name)) THEN
SET NEW.id = NULL;
ELSE
INSERT INTO computers (type) VALUES ('desktop');
SET NEW.id = LAST_INSERT_ID();
SET NEW.id = LAST_INSERT_ID(NEW.id);
END IF;
END
INSERT INTO laptops (name) VALUES ('laptop1');
INSERT INTO laptops (desktop) VALUES ('desktop1');
INSERT INTO laptops (name) VALUES ('laptop2');
INSERT INTO laptops (desktop) VALUES ('desktop2');
SELECT last_insert_id();
Expecting 4, actually its 0.
Any thoughts as to how I can fix the trigger? Maybe someone can help me format the AFTER_INSERT statement to fix last_insert_id?
I tried setting the values to auto-increment, and unique in the laptops and desktops table, neither will fix the issue.

Answer 1

Rather than trying to deal with the 'confusion' of 'last_insert_id'. I decided to change the table structure to be a more 'common' format.
That is change the 'laptops' and 'desktops' tables to have the 'auto_increment' keys. This changes the 'computers' table to have a primary key of 'computer_id' from 'laptops' or 'desktops' and a 'computer_type'.
Here are the table structures and triggers.
It has been tested on mysql 5.5.16 on windows xp.
CREATE TABLE `laptops` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`name` varchar(45) DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=2 DEFAULT CHARSET=utf8;
CREATE TABLE `desktops` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`name` varchar(45) DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=4 DEFAULT CHARSET=utf8;
CREATE TABLE `computers` (
`computer_id` int(11) NOT NULL,
`computer_type` varchar(45) NOT NULL,
PRIMARY KEY (`computer_id`,`computer_type`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
DELIMITER $$
USE `testmysql`$$
DROP TRIGGER /*!50032 IF EXISTS */ `laptop_bins`$$
CREATE
/*!50017 DEFINER = 'test'#'localhost' */
TRIGGER `laptop_bins` AFTER INSERT ON `laptops`
FOR EACH ROW BEGIN
INSERT INTO computers (computer_id, computer_type ) VALUES (new.id, 'laptop');
END;
$$
DELIMITER ;
DELIMITER $$
USE `testmysql`$$
DROP TRIGGER /*!50032 IF EXISTS */ `desktop_bins`$$
CREATE
/*!50017 DEFINER = 'test'#'localhost' */
TRIGGER `desktop_bins` AFTER INSERT ON `desktops`
FOR EACH ROW BEGIN
INSERT INTO computers (computer_id, computer_type ) VALUES (new.id, 'desktop');
END;
$$
DELIMITER ;
Sample Queries and Output:
INSERT INTO laptops (NAME) VALUES ('laptop1');
INSERT INTO desktops (NAME) VALUES ('desktop1');
INSERT INTO laptops (NAME) VALUES ('laptop2');
INSERT INTO desktops (NAME) VALUES ('desktop2');
Laptops:
id name
------ ---------
1 laptop1
2 laptop2
Desktops:
id name
------ ----------
1 desktop1
2 desktop2
Computers:
computer_id computer_type
----------- ---------------
1 desktop
1 laptop
2 desktop
2 laptop

Answer 2

This more a possible approach to the requirement than an answer.
I can create the code if required. It is not a lot of code on top of what is here.
The problem is to maintain tables in an other database, in sync, without doing lots of repeat work.
My suggestion:
In the 'computers' database - have a 'computers_new' table that is inserted to by the 'after insert' trigger and holds the relevant key information. Including a 'unprocessed' column.
I would then run a script at regular intervals or was triggered when the 'computers_new' table changed. It would:
1) transfer the 'unprocessed' details to the 'laptops', 'desktops' tables in the other database.
2) mark the transferred records as processed.
Advantages:
Lots of small chunks of work.
By using transactions it is reliable.
Drawbacks.
Ensuring tables are in sync.