为什么SQL子查询结果没有加GROUP BY还能按分组结果显示

CREATE TABLE t_book (
FId int(11) NOT NULL,
FName varchar(50) DEFAULT NULL,
FYearPublished int(11) DEFAULT NULL,
FCategoryId int(11) DEFAULT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

-- Records of t_book

INSERT INTO t_book VALUES ('1', 'About J2EE', '2005', '4');
INSERT INTO t_book VALUES ('2', 'Learning Hibernate', '2003', '4');
INSERT INTO t_book VALUES ('3', 'Two Cites', '1999', '1');
INSERT INTO t_book VALUES ('4', 'Jane Eyre', '2001', '1');
INSERT INTO t_book VALUES ('5', 'Oliver Twist', '2002', '1');
INSERT INTO t_book VALUES ('6', 'History of China', '1982', '2');
INSERT INTO t_book VALUES ('7', 'History of England', '1860', '2');
INSERT INTO t_book VALUES ('8', 'History of America', '1700', '2');
INSERT INTO t_book VALUES ('9', 'History of The World', '2008', '2');
INSERT INTO t_book VALUES ('10', 'Atom', '1930', '3');
INSERT INTO t_book VALUES ('11', 'RELATIVITY', '1945', '3');
INSERT INTO t_book VALUES ('12', 'Computer', '1970', '3');
INSERT INTO t_book VALUES ('13', 'Astronomy', '1971', '3');
INSERT INTO t_book VALUES ('14', 'How To Singing', '1771', '5');
INSERT INTO t_book VALUES ('15', 'DaoDeJing', '2001', '6');
INSERT INTO t_book VALUES ('16', 'Obedience to Authority', '1995', '6');

CREATE TABLE t_category (
FId int(11) NOT NULL,
FName varchar(50) DEFAULT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

-- Records of t_category

INSERT INTO t_category VALUES ('1', 'Story');
INSERT INTO t_category VALUES ('2', 'History');
INSERT INTO t_category VALUES ('3', 'Theory');
INSERT INTO t_category VALUES ('4', 'Technology');
INSERT INTO t_category VALUES ('5', 'Art');
INSERT INTO t_category VALUES ('6', 'Philosophy');

任务：查询出每种类目的类目名称、类目id，最晚的出版年份
想要得到的结果类似下图

原文提供了两种写法

第一种：
select c.*,A.maxPub
from t_category c
left join (
select b.FCategoryId cid ,max(b.fyearpublished) maxPub from t_book b group by b.FCategoryId)A
on c.fid=A.cid

第二种

select c.fid,c.FName,(select max(b.FYearPublished) from t_book b where b.FCategoryId=c.FId) maxPub from t_category c

第二种仅按照代码来理解，子查询中得到的结果，难道不是用条件b.FCategoryId=c.FId连接起来的所有类目中最晚的发布年份吗？为什么第二种写法子查询中没有加GROUP BY 还能运行出来呢？

原文链接
SELECT列表中的标量子查询 - 我俩绝配 - 博客园发现了一种表连接新的写法，以前还没有这样写过或者见别人写过。跟同学聊天他们公司却很多人这样写，看来真的要学学sql了表 CREATE TABLE `t_book` ( `FId` int(11) N https://www.cnblogs.com/lihaoyang/p/8648955.html

针对你这个问题我大概解释一下啊，子查询select max(b.FYearPublished) from t_book b where b.FCategoryId=c.FId的意思是已经过滤后的的结果再进行max函数取最大值，过滤条件就是当类目已经确定是某一个的时候，所以不用再group by。
举例说明，如不加子查询的结果为
1 Story
2 History
3 Theory
4 Technology
5 Art
6 Philosophy
增加子查询后，
针对第一行，子查询就等于select max(b.FYearPublished) from t_book b where b.FCategoryId=1
结果为：
1 Story 2002
2 History
3 Theory
4 Technology
5 Art
6 Philosophy

第二行为select max(b.FYearPublished) from t_book b where b.FCategoryId=2，结果为：
1 Story 2002
2 History 2008
3 Theory
4 Technology
5 Art
6 Philosophy
依次类推
就是
1 Story 2002
2 History 2008
3 Theory 1971
4 Technology 2005
5 Art 1771
6 Philosophy 2001