MySQL查询IN GROUP_CONCAT不起作用

Question

Having problems with a MySQL query. Only returning the results for the first item in the resulting GROUP_CONCAT array. ie resulting array is [1,3,4], the quesitionnaires for only user id 1 is returned.

$query_search = "SELECT questionnaires_index.id, questionnaires_index.ea_num, questionnaires_index.address, questionnaires_index.status, questionnaires_index.json_stored, users.username FROM questionnaires_index INNER JOIN users ON users.id = questionnaires_index.interviewer_id WHERE questionnaires_index.interviewer_id IN (SELECT GROUP_CONCAT(id) FROM users WHERE supervisor = (SELECT id FROM users WHERE username = '".$username."'))";

Am I using GROUP_CONCAT incorrectly? Is there a better way of doing this?

EDIT 1 Here are the SQL tables used in this query

CREATE TABLE IF NOT EXISTS `questionnaires_index` (
  `id` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `ea_num` int(10) unsigned NOT NULL,
  `address` varchar(100) NOT NULL,
  `interviewer_id` int(10) unsigned NOT NULL,
  `status` varchar(30) NOT NULL DEFAULT 'Available',
  `json_stored` tinyint(1) NOT NULL DEFAULT '0',
  PRIMARY KEY (`id`)
) ENGINE=MyISAM  DEFAULT CHARSET=latin1 AUTO_INCREMENT=8 ;

--
-- Dumping data for table `questionnaires_index`
--

INSERT INTO `questionnaires_index` (`id`, `ea_num`, `address`, `interviewer_id`, `status`, `json_stored`) VALUES
(1, 101, '29 De Havilland Crescent Pro Park, Building 1 Persequor Technopark Pretoria 0020', 1, 'Non Contact', 1),
(2, 102, '5th Floor, Imperial Bank Terraces Carl Cronje Drive Tyger Waterfront Bellville 7530', 1, 'Available', 0),
(3, 101, '29 De Havilland Crescent Pro Park, Building 1 Persequor Technopark Pretoria 0020', 3, 'Partially Completed', 0),
(4, 102, '5th Floor, Imperial Bank Terraces Carl Cronje Drive Tyger Waterfront Bellville 7530', 3, 'Available', 0),
(5, 201, '101 test address', 4, 'Available', 0);

And for users:

CREATE TABLE IF NOT EXISTS `users` (
  `id` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `username` varchar(20) NOT NULL,
  `password` varchar(20) NOT NULL,
  `supervisor` int(10) unsigned NOT NULL,
  `version_code` varchar(10) NOT NULL,
  `is_interviewer` tinyint(1) NOT NULL DEFAULT '0',
  `is_supervisor` tinyint(1) NOT NULL DEFAULT '0',
  `surname` varchar(50) NOT NULL,
  UNIQUE KEY `id` (`id`)
) ENGINE=MyISAM  DEFAULT CHARSET=latin1 AUTO_INCREMENT=5 ;

--
-- Dumping data for table `users`
--

INSERT INTO `users` (`id`, `username`, `password`, `supervisor`, `version_code`, `is_interviewer`, `is_supervisor`, `surname`) VALUES
(1, 'Rynardt', 'q', 2, '2.2.0', 1, 0, ''),
(2, 'Herholdt', 'q', 0, '2.2.0', 0, 1, ''),
(3, 'test', 'test', 2, '2.2.0', 1, 0, ''),
(4, 'Botha', 'q', 2, '', 1, 0, '');

EDIT 2 More info on what I am trying to do:

If you look at the EDIT I made you will see the SQL tables are now included. u.supervisor is op type integer to indicate id of supervisor entry in users table. $username is of type string and indicates the name of the supervisor. I therefore first have to get the id of the supervisor, then get all the id's of users that has a supervisor with the id found previously. Then using this array of users.id get all the questionnaires associated with these users and display it to the supervisor when he logs in.

Answer 1

SELECT GROUP_CONCAT(id) FROM users WHERE supervisor = 
   (SELECT id FROM users WHERE username = '".$username."') 
GROUP BY supervisor

to work with aggregate functions like group_concat , you need to have an aggregated result set ;)

edit...hmm... in this special case

 WHERE ... IN (SELECT id FROM ...)

should be sufficent, no need to concatenate the result sets

Answer 2

Well firstly, you could skip the GROUP_CONCAT(ID) here completely and simply use interviewer_id IN ( SELECT ID FROM users [...] )

Edit: since you updated your post with the tables, yep - then most likely you would need that Subquery. You could also extract the ID of the supervisor in an additional query and could skip the nested-sql here. Or, since you said the supervisor logs in and has to see the data aquired... maybe also save the user-id into your session ( I guess it comes from there? ) and skip the subquery and additional query completely just by adding the user-id to your login-script creating session vars.

And just on a side note... you should at least hash the passwords with md5 or sha. Good idea would be to save the time the user registred and use the timestamp + a constant to salt the password before hashing and saving password to the table.