我发现,尝试存储像状态事情通常是一个错误。例如,“当前”,“前任”,“重新雇佣”的就业状态通常更好地实施为具有开始日期和结束日期的就业表。
碎表和断开关系在数据库设计中不是正常的英文术语。我不确定你在这里的意思。
PostgreSQL代码如下。 SQL Server将使用日期时间数据类型来代替标准SQL的时间戳数据类型。可能还有其他小的差异。
-- Nothing surprising here.
create table schools (
school_id integer primary key,
school_name varchar(20) not null unique
-- other columns go here
);
-- Nothing surprising here.
create table tutors (
tutor_id integer primary key,
tutor_name varchar(20) not null
-- other columns go here
);
-- Nothing surprising here.
create table tutor_details (
tutor_id integer primary key references tutors (tutor_id),
tutor_phone varchar(15)
-- other columns go here
);
-- Predicate: School <school_id> employed <tutor_id>
-- starting on <start_date> and ending on <end_date>.
-- Allows multiple periods of employment.
create table school_tutors (
school_id integer not null references schools (school_id),
tutor_id integer not null references tutors (tutor_id),
start_date date not null default current_date,
end_date date not null default '9999-12-31',
-- You can make a good, practical argument for including end_date
-- in the primary key, but that's a different issue.
primary key (school_id, tutor_id, start_date)
);
-- Only makes sense in the context of employment, so a composite
-- foreign key references school_tutors. In production, I'd try
-- to use more check constraints on the timestamps.
create table tutor_office_hours (
school_id integer not null,
tutor_id integer not null,
start_date date not null,
foreign key (school_id, tutor_id, start_date)
references school_tutors (school_id, tutor_id, start_date),
office_hours_start_time timestamp not null,
office_hours_end_time timestamp not null
check (office_hours_end_time > office_hours_start_time),
primary key (school_id, tutor_id, office_hours_start_time)
);
感谢Zohar Peld先生纠正我的文章,我很快就写了,因为我的笔记本电池电量不足。 –