A data type is assigned to every attribute (field) in a table, and it does two jobs at once: it restricts which VALUES the field will accept, and it determines which OPERATIONS are actually permitted on that field, arithmetic, date comparisons, and so on.

数据类型会被赋予表中的每一个属性(字段),它同时承担两项工作:限制该字段可以接受哪些值,并决定该字段实际允许执行哪些操作,例如算术运算、日期比较等。

Storage and indexing efficiency. When a field's length genuinely never changes, a fixed-width UserID is the classic case, CHARACTER(n) can reserve exactly that much space and be done with it. VARCHAR(n) has to keep track of how long each individual value actually is, and that bookkeeping is work a truly fixed-length field simply never has to do.

存储与索引效率。当一个字段的长度确实永远不变时(固定宽度的 UserID 就是典型例子),CHARACTER(n) 可以精确预留这么多空间,不再多做任何事。VARCHAR(n) 则必须记录每个值实际的长度,而这项记录工作对于真正的定长字段来说根本不需要。

Which operations are even possible. A value's data type is what tells the database whether an operation makes sense at all. Give it a price or a quantity typed as plain text instead of REAL or INTEGER, and adding two of them together stops being something the database can do, the application is left converting that text back into a real number itself, on every single calculation, forever.

哪些操作根本上是可行的。值的数据类型决定了某项操作对数据库而言是否有意义。如果把价格或数量当作纯文本而非 REAL 或 INTEGER 来存储,那么把两个值相加就不再是数据库能做的事,应用程序只能自行把这些文本转换回真正的数字,而且每次计算都要重新转换一次,永无止境。

Insertion errors. A field's declared type is a promise the DBMS actually enforces, not just a suggestion. Try to insert "twenty" into a field typed INTEGER and the insert is refused outright, loudly, at the moment it happens, rather than being quietly accepted and causing confusion somewhere down the line.

插入错误。字段声明的类型是 DBMS 真正会强制执行的约定,而不仅仅是一种建议。如果尝试把"twenty"插入一个类型为 INTEGER 的字段,插入操作会在发生的当下就被直接、明确地拒绝,而不是被悄悄接受,并在之后的某个环节造成混乱。

Data type consistency is easy to break without noticing, since the database and the application code that talks to it are two separate places the same attribute's type has to agree. Let them drift apart, one side treating a field as an integer, the other as text, and the failure often isn't a clean error at all, it's a query that runs and returns something quietly wrong.

数据类型一致性很容易在不知不觉中被破坏,因为数据库与和它交互的应用程序代码是两个独立的地方,它们对同一属性的类型必须保持一致。一旦两者出现偏差,一边把某字段当作整数处理,另一边当作文本处理,由此产生的故障往往不是一个干净利落的错误,而是一条能正常执行、却悄悄返回错误结果的查询。