这篇文章的目的是记录CS350 Assignment2中, 我编写各种System Calls时所采用的思路. 实际coding的时候, 同一种System Call的实现方式很可能不止一种, 但殊途同归.
注1: 文章顺序和实际coding顺序并不一定一致, 请参考Assignment中的Strategy部分.
注2: 如果没有仔细读过code和Assignment...这里很有些东西你可能读的似懂非懂.

欢迎在评论中提问...

General Tips

1. 别管写什么function, 第一步永远是检查parameter是否有效!!!
   (比如, pointer是不是NULL, string是不是空, etc)
2. 不要放过任何一个warning...需要explicit cast的时候, 千万不要偷懒.
3. function一般使用parameter中的pointer进行value return, 正常的return用来return errno.
4. 开始coding之前, 除了读Assignment, 读code, 还要好好读Assignment Hint! 解答了很多FAQ
5. 注意在header file里用#ifndef, 保证header不会出现重复include
6. kmalloc了的东西...在destroy的时候一定要kfree. 不然A3里, 你会发现各种memory leak
   (Create某些data structure的中间, 如果出现error, 则需要首先free已经allocated的内存, 然后返回error)
7. 善用spl解决mutex问题. 当然, 首先你要明白什么样的操作需要mutex!

我会边写这个Note边添加Tips...

Read more...

Definition实在不知道怎么概括= =...

• Intention of a relation: relation definition and constrains
• Extention of a relation: the actual data, tuples

Relation properties

• Set Theory
  
  
  • Attributes has order (not necessary)
  • Value are used to identify tuples
  • Tuples dont have order
  • Tuples cant have duplication
• Attribute value are atomic
• Degree: How many attributes in the schema
• Cardinality: How many tuples in an instance

Constraints

• Let the DBMS to ensure the entry or modification on data are legal
• Give applications' bug no chance to ruin the data

Three Constraint Level

• Tuple-level
  
  
  • Domain restrictions (datatype)
  • Attribute comparisons (好像是指check)
• Relation-level
  
  
  • Special type of keys
    
    
    • Superkey: 能uniquely identifies a tuple的一个or一组attribute
    • Candidate key: Superkey的最小集, (最简单的一个or一组attribute)
    • Primary key: 被选中, 用来identity tuples的Candidate key
• Database-level
  
  
  • Referential integrity
    
    
    • Foreign key: 其他relation中的primary key
    • Referential integrity: 不允许出现和对应relation中的记录不符的情况
  • Inclusion dependencies: = =`不懂

Definitions

• Database: is a large and persistent collection of pieces of information organized in a way that facilitate efficient retrieval and modification.
• DBMS: is a program that manages details related to storage and access for a database.
• Schema: is a description of the data interface to the database.
• Instance: is a database that conforms to a given schema.
• Transaction: is an application-specified atomic and durable unit of work.
• View: is a relation in the external schema whose instance is determined by the instances of the relations in the conceptual schema.
• Trigger: is a procedure executed by the database in response to a change to the database instance

Three Level Schema Architecture

• External schema (view): visualized data
• Conceptual schema: logical structure
• Physical schema: file/storage devices

Data Independence

• Application access the data through an abstract data model provided by the DBMS, rather than direct access
• Physical: applications cant touch the storage stucture
• Logical: applications cant change the data organization

Four Transaction Properties

NoSQL据说就是要break these properties= =`IDK why...

• Atomic: once the transaction happens, it happened. No intermediate stage!
• Consistency: transaction follows the database consitences
• Isolated: concurrent transactions wont effect each other
• Durable: transaction results permanently

What DBMS do

• Remoce common code from applications
• Provide uniform access to data
• Guarantee data integrity
• manage concurrent access
• Protect against system failure
• Set access policies for data

SELECT

SELECT attribute-expression-list FROM relation-list [ WHERE condition ];

attribute-expression-list:

• [relation-name.]attribute
• [relation-name.]attribute [arithmatic computation] AS another name
  
  
  • i.e., E.Salary - 40000 AS SalaryDiff
• CASE WHEN ... THEN ...
  ELSE ... END
  
  
  • i.e., CASE WHEN E.Salary < 40000 THEN 0
    ELSE E.Salary - 40000 END

relation-list:

• list of table names
• seperate by comma

condition:

• arithmetic operation +, -, *, /
• comparisions =, <>, <, <=, >. >=
• logical connectives AND, OR, NOT
• attribute IN (Q)
• attribute NOT IN (Q)
• attribute op SOME (Q)
• attribute op ALL (Q)
• EXISTS (Q)
• NOT EXISTS (Q)
• IS [NOT] NULL

Read more...