我会用两个 LEFT JOINS 用一个 UNION :

CREATE TABLE `a` (
    `id` INT(11) NOT NULL AUTO_INCREMENT,
    `name` VARCHAR(50) NOT NULL DEFAULT '0',
    `supplier` VARCHAR(50) NOT NULL DEFAULT '0',
    `value` INT(11) NOT NULL DEFAULT '0',
    PRIMARY KEY (`id`)
)
COLLATE='latin1_swedish_ci'
ENGINE=InnoDB
AUTO_INCREMENT=5
;

CREATE TABLE `b` (
    `id` INT(11) NOT NULL AUTO_INCREMENT,
    `name` VARCHAR(50) NOT NULL DEFAULT '0',
    `supplier` VARCHAR(50) NOT NULL DEFAULT '0',
    `value` INT(11) NOT NULL DEFAULT '0',
    PRIMARY KEY (`id`)
)
COLLATE='latin1_swedish_ci'
ENGINE=InnoDB
AUTO_INCREMENT=6
;

INSERT INTO `a` (`id`, `name`, `supplier`, `value`) VALUES (1, 'Alice', 'X', 100);
INSERT INTO `a` (`id`, `name`, `supplier`, `value`) VALUES (2, 'Bob', 'Y', 200);
INSERT INTO `a` (`id`, `name`, `supplier`, `value`) VALUES (3, 'Clare', 'Z', 300);
INSERT INTO `a` (`id`, `name`, `supplier`, `value`) VALUES (4, 'Desmond', 'X', 400);

INSERT INTO `b` (`id`, `name`, `supplier`, `value`) VALUES (1, 'Alice', 'X', 150);
INSERT INTO `b` (`id`, `name`, `supplier`, `value`) VALUES (2, 'Bob', 'X', 200);
INSERT INTO `b` (`id`, `name`, `supplier`, `value`) VALUES (3, 'Clare', 'Z', 350);
INSERT INTO `b` (`id`, `name`, `supplier`, `value`) VALUES (4, 'Desmond', 'X', 400);
INSERT INTO `b` (`id`, `name`, `supplier`, `value`) VALUES (5, 'Emily', 'X', 500);

SELECT a.name AS name, a.supplier AS a_supplier, a.value AS a_value, b.supplier AS b_supplier, b.value AS b_value FROM a
LEFT JOIN b ON a.name = b.name
WHERE (a.supplier ='X' OR b.supplier = 'X') AND (a.value <> b.value OR a.supplier <> b.supplier OR b.name IS NULL)
UNION 
SELECT b.name AS name, a.supplier AS a_supplier, a.value AS a_value, b.supplier AS b_supplier, b.value AS b_value  FROM b
LEFT JOIN a ON b.name = a.name
WHERE (a.supplier ='X' OR b.supplier = 'X') AND (a.value <> b.value OR a.supplier <> b.supplier OR a.name IS NULL)

首先，加入表 A 到表 B ，然后进行反向连接。
我不确定是否可以通过表的id连接这些表，因此在本例中，我使用名称作为连接列。
每个连接都包含一个 WHERE 使用cirteria筛选行的子句：“对涉及供应商x的任何行的更改”。
这是一把小提琴：http://sqlfiddle.com/#!9/46f213/1号楼

扩展lldar的答案，您还可以通过对列进行散列，然后查找更改来获得差异。

md5(concat(A.`Name`,A.`Supplier`, A.`Value`)) <> md5(concat(b.`Name`,b.`Supplier`,b.`Value`))

如果您有许多列，这会很有帮助。理想情况下，从长远来看，您可以编辑这些表，并将哈希添加为“computed/calculated”列。
那就简单了 A.hash <> b.hash

您可以在原始查询中添加一些where子句来检查supplier x，但我想我会采取稍微不同的方法，并使用join：

SELECT a.id, a.name, a.supplier, a.value, b.name, b.supplier, b.value
FROM a
INNER JOIN b ON (a.id = b.id AND (a.name != b.name OR a.value != b.value OR a.supplier != b.supplier))
WHERE a.supplier = 'X' OR b.supplier = 'X'
GROUP BY a.id;

这将获取已更改但仅与x相关的行。注意，这假设每个表中始终只有一个匹配的id。

只使用左连接就可以满足要求

SELECT b.NAME AS NAME, 
       a.supplier AS a_supplier, 
       a.value AS a_value, 
       b.supplier AS b_supplier, 
       b.value AS b_value
FROM   b 
       LEFT JOIN a 
              ON ( a.id = b.id ) 
WHERE  ( b.supplier = 'X' 
          OR a.supplier = 'X' ) 
       AND ( a.supplier != b.supplier 
              OR a.value != b.value 
              OR a.id IS NULL ) 
ORDER  BY b.id;