### What problem were solved in this pull request?

Issue Number: ref #205 

Problem:
The build method is different with the current docker image, so we need
to update it.

### What is changed and how it works?
Install build tools and dependencies when building image.

### What problem were solved in this pull request?

Issue Number: close #230 

Problem:
logical plan的生成代码全放在了optimize中，看起来很不清晰

### What is changed and how it works?
将logical plan的代码提取出来放在logical plan generator中

### What problem were solved in this pull request?

Problem: close #229 
当前的词法分析语法分析非常简单，那当遇到比较复杂的词法分析语法分析题目，比如表达式，有些同学就难以入手

### What is changed and how it works?
增加一个支持四则运算的功能，关键字是calc，虽然与标准的SQL不同，但是可以作为学习表达式入门使用。
使用示例：
```
(base) build_debug $ ./bin/observer -f ../etc/observer.ini -P cli
Successfully load ../etc/observer.ini
miniob > calc 2*(2*(2+3));
2*(2*(2+3))
20

miniob > calc (1+2) * (2 * (20+ -(5*1)))
(1+2) * (2 * (20+ -(5*1)))
90

miniob > calc 1, 2, 3;
1 | 2 | 3
1 | 2 | 3

miniob > calc 1 + 2 - 3+4;
1 + 2 - 3+4
4

miniob > calc 2* 4;
2* 4
8

miniob > calc 1, 2, 3;
1 | 2 | 3
1 | 2 | 3

miniob > calc 1 + 2 - 3+4;
1 + 2 - 3+4
4

miniob > calc 2* 4;
2* 4
8
```

### What problem were solved in this pull request?

Issue Number: close #226

Problem:
记录管理部分中page_header_中record_capacity初始化有误，需要进行修正
由于record_capacity的计算并没有考虑8字节对齐，计算的是不对齐时页面下容纳的最大记录量。
但第一条记录偏移量first_record_offset需要8字节对齐，使得最后一条记录的数据可能会溢出页面。

### What is changed and how it works?
如果溢出页面，则减少record_capacity数值
同时去掉了部分冗余或含义不清的函数，修改了上取整的计算表达式
完善了记录管理模块成员函数的注释

### What problem were solved in this pull request?
Problem:
parse_defs中的一些变量命名不是很清晰

### What is changed and how it works?
重命名，增加SqlNode后缀，希望看起来更清晰，知道这些变量的意图
另外，增加了一个visual studio code的生成parser的任务

### What problem were solved in this pull request?

Issue Number: ref #205 

Problem:
build failed on docker container

### What is changed and how it works?
fix compile error

### Other information

### What problem were solved in this pull request?

Issue Number: close #217 

Problem:
compile error on gcc 12

### What is changed and how it works?

### Other information

### What problem were solved in this pull request?
ref #138 

Problem:

SEDA的线程池模式是每个处理环节分开，可能会放到不同的线程上，也可以调整不同stage对应的处理线程个数。但是会增加复杂性，比如session会切换线程、代码逻辑复杂。

### What is changed and how it works?
去掉了parse、resolve等stage，仅保留了session stage。代码中的名字依然叫stage，后续再调整。
在session
stage的处理逻辑中，直接调用其它流程处理，这样也使SQL处理流程变得更清晰。后续还可以考虑移除observer.ini，所有参数均通过命令行传递。

### Other information

### What problem were solved in this pull request?

Issue Number: close #178 

Problem:
sysbench is a powerful concurrency test tool and we should use it to
test our program.

### What is changed and how it works?
I create two sysbench lua scripts and a github workflow.

### Other information

### What problem were solved in this pull request?

Problem:
与训练营配合，可以在SQL命令请求过程中，添加调试信息

### What is changed and how it works?
可以参考 debug-output.md 文档。
增加 sql_debug 变量，使用set sql_debug=1; 可以设置。
在SQL命令执行过程中，调用sql_debug函数，增加调试信息。在普通文本通讯协议中，调试信息会以 '#'
开头的形式打印出来。但是注意调试信息中不要带换行符。

### Other information

### What problem were solved in this pull request?

Issue Number: close #142 

Problem:
value和tuple_cell的语义是重合的，所以代码时冗余的

### What is changed and how it works?
删除tuple_cell，都使用value来替换，并将tuple_cell中的接口都在value中实现

### Other information

### What problem were solved in this pull request?

Issue Number: close #191 

Problem:
doxy文档的首页没有描述

### What is changed and how it works?
增加首页描述信息，写在了main.cpp中

### Other information

### What problem were solved in this pull request?

Issue Number: ref #165 
ref #174 

Problem:
一些文档需要优化

### What problem were solved in this pull request?

Issue Number: close #162 

Problem:
当前的observer启动流程和调试方法比较繁琐，必须使用客户端服务端的方式，先启动服务端程序，再使用客户端启动调试

### What is changed and how it works?
observer可以直接启动，不监听tcp或unix socket，直接通过终端/控制台输入命令并执行，极大的方便了调试

### Other information
新的启动方法：
./bin/observer -P cli -f ../etc/observer.ini

### What problem were solved in this pull request?

Issue Number: close #151 

Problem:
当前在发送消息到客户端时，每个数据都会刷新socket buffer，效率很低

### What is changed and how it works?
实现BufferedWriter，通过buffered
writer将消息缓存在内存中，结果写完时或者buffer满时，才将结果真正的发送到客户端。

### Other information

### What problem were solved in this pull request?

Problem:
当前生成的github pages不太整洁方便，可以优化一下

### What is changed and how it works?
改用mdbook生成文档，并整理文档目录，调整链接

### What problem were solved in this pull request?

Issue Number: close #173 close #136 

ref #174 
ref #165 

Problem:
这个PR修改了多个问题，可以参考各个issue。包括observer编译两次、代码目录规划不合理、command executor未全部实现等

### What is changed and how it works?
做一些重构优化，包括CMakelist、一些代码目录、command executor

### What problem were solved in this pull request?

close issue: #186 

Problem:
实现MVCC代码后没有将持久化代码合并进来

### What is changed and how it works?
支持并发模式下落日志，并在异常停机后回复。

持久化是事务模块中最复杂的功能（没有之一），当前miniob实现的持久化简化了非常多，因此不能依赖持久化能力，仅仅作为一个学习测试的工具，在实际测试时也有一些限制。

### Other information

### What problem were solved in this pull request?

Issue Number: close #192 

Problem:
clang 14编译失败

### What is changed and how it works?
修复编译错误

### Other information

### What problem were solved in this pull request?

Issue Number: ref #165 

Problem:
record_manager 没有比较详细的注释

### What is changed and how it works?
加一些注释

### Other information

### What problem were solved in this pull request?

Issue Number: close #149 

Problem:
RC return code ，定义的东西太多，不太简洁，并且有很多错误码当前没有使用

### What is changed and how it works?
仅使用enum RC 定义错误码，删掉其它枚举定义。
删除当前没有使用的错误码。

### Other information

### What problem were solved in this pull request?

Issue Number: close #61 

Problem:
cmake 3.10 版本不支持 cmake_link_directories 命令，而且miniob也不需要

### What is changed and how it works?
将 cmake_link_directories 相关代码删除，可以正常编译

### What problem were solved in this pull request?

Issue Number: close #91 close #1 

Problem:
编译时编译器抛出的一些告警信息可以帮我们避免很多问题，但是现在并没有利用这个特性

### What is changed and how it works?
CMake中增加-Werror编译选项

### Other information

fix unittest;
add unittest to github action

### What problem were solved in this pull request?

Issue Number: close #176 close #163 

Problem:
单元测试执行失败；
github action 没有单元测试检查

### What problem were solved in this pull request?

Issue Number: close #155 close #135 

Problem:
实现了简单的事务处理。
当前支持并发事务数据.

### What is changed and how it works?
- 支持两种事务模式：Vacuous和MVCC。
  Vacuous 事务模式算是没有事务，它的事务接口什么都不做。
MVCC 是多版本并发控制(Multi-Version Concurrency
Control)，使用多个版本保留记录数据。启动miniob时增加运行时选项 -t mvcc可以选择mvcc。

- MVCC：简单模式的多版本并发控制。
  当前miniob仅包含insert和delete操作，因此数据最多包含两个版本，并且不需要在record中保留版本链表信息；
  不支持持久化；
  没有做垃圾回收；
  遗留一个BUG：在提交时没有保证提交的数据一次性对外可见；
  使用简单的写写冲突策略：检测到要修改的数据当前有人在修改，就回退

- 编译MINIOB时使用 -DCONCURRENCY=ON 才会支持并发
支持各个模块并发处理，包括buffer pool、bplus、record
manager。如果编译时没有使用CONCURRENCY选项，则保持原样，不支持并发，保持系统的简单性。

- 编译时增加 -DDEBUG=ON 还会增加并发调试日志与严格的运行时检测(ASSERT语句)
- 当前版本代码中包含了bplus tree和record manager的并发测试，参考benchmark目录下的代码。

### Other information

Fix problem:
1. the buffer pool and b tree is not thread safe;
2. github/workflow/build does not work

### What is changed and how it works?
1. thread-safe buffer pool
- I use a mutex in buffer pool and take a lock in buffer pool operations
such as allocate frame, dispose frame;
- The frame is locked while updating/reading the content of frame;
- Frame manager take a lock when allocate/free pages.

2. thread-safe b+tree
- Crabing protocol is used to support concurrent

3. github/workflow/build
- update the submodules;
- create a build script and run build.sh in build.yaml

logical plan & physical plan;
explain;
yacc rewrite;
join operator(nested loop join);
simple rewrite rules;
fix some issues of mysql protocol;

tested by mariadb ( Ver 15.1 Distrib 5.5.65-MariaDB, for Linux (x86_64) using readline 5.1)
but failed with obclient

如果使用readline，可以记录历史命令

2022 oceanbase competition

* support build with address sanitizer

using cmake options `-DENALBE_ASAN=ON` to enable address sanitizer

* fix some memory leaks detected by asan