Undo Recovery Scheme:

Create a special relation named Log for logging purpose. The attributes of the log relation is as follows:

- type {START_TRANS = 0, COMMIT_TRANS = 1, ABORT_TRANS = 2, WRITE = 3, START_CHECKPOINT = 4, END_CHECKPOINT = 5} (byte) (if your design does not support byte data type, int can be used)
- transID (int)
- fileID (int)
- blockNum (int)
- offset (byte) (if your design does not support byte, int can be used)
- content Varchar(SIze) (SIZE is the maximum possible record length in any data, index file).

The operation on regular relation can be used to insert new log records into the log relation. In addition, this relation is built-in and can not be deleted.

Rules:
- When a new transaction starts:
	- a new record <Start_Trans,transID, 0, 0, 0, null> will be inserted to the log file.
- When an array of bytes X_OLD with length LEN (byte[]) of a block B from offset p to offset p+LEN to be updated with a new content X_NEW (byte[]) (applicable to both data and index blocks, even applicable to block/file headers and catalog). 
	- a new record <write, transID, fileID, blockNum, p, X_OLD> will be inserted to the log file prior to the update.
- If the block B is going to be written to disk, the log entry above must be flushed to disk first.
- When a transaction is aborted:
	- a new record <Abort_Trans,transID, 0, 0, 0, null> will be inserted to the log file.
- When a transaction is committed:
	- First, all blocks must be flushed to disk.
	- Second, a new record <Commit_Trans,transID, 0, 0, 0, null> will be inserted to the log file.
	- Then, the commit log record can be written to disk.
- Periodically:
	- First, insert into the log file a new log record: <START_CHK, ID of current active transaction, 0,0 ,0, null>.
	- Second, wait for the current active transaction is completed.
	- Third, insert into the log file a new log record: <END_CHK,0,0,0,0,null>.
- After the system is initialized, the log file will be examed. If a system crash from last running is detected, the log will be used to undo the changes by writing all the old values of the blocks from last crash. Refer to section 8.2.5 (nonquiescient check pointing) for more information.

Requirements:
1. Implement the logging scheme above.

Testing:
1. TA will ask you to run a script file which contains several insertion commands.The format of the script file will be as follows:
	Start
	insert ....
	insert ...
	.....
	Commit (or Abort)
2. Then, he will issue some SQL queries on the Log relation, i.e. select * from log, to exam the content of the log file.

Note: Pay attention to when the log records should be flushed to disk (follows the rules above).