bonus.txt

We applied the concept fo serialization on the BPlus Tree also as the 
relation Pages to approach more to the real life application , we 
designed handling duplicates by making the reference of a value point 
to an overflow bucket that contains all page references that contains 
this value(in case no duplicates value has a direct reference to the page) 
respecting the maximum node size for each overflow bucket.

We did serialize each node in the B+ Tree and the descendant overflow 
buckets from the tree to the disk and only the root is in the memory.

Whenever we wanted to access a node in computation such as 
search/insertion/deletion we deserialize the node by its name, 
we identified all the nodes with a unique name, 
Node+(unique number) 
This happened for example in lines {
*BPTreeInnerNode.java:-
**in helper methods: 	40,62,72
**insertions: 		96
**deletion:		223,229-262,268
*search:			403-410-425-430-465
*BPTreeLeafNode.java:-
*helper methods:		35
*search:			506-518
}


When we finished computation if we adjusted in any node we re-serialize 
that node into the disk; such as insertion in lines {
*BPTree.java:-
**updateRef			46
**insert			69
*BPTreeInnerNode.java:-
**insert			110,111,120
**delete			225,231,239-264,270,278
***borrow(helper for delete)	312,316,329,333
***merge (helper for delete)	353,361
*BPTreeLeafNode.java:-
**insert			140
***borrow(helper for delete)	393,407
***merge (helper for delete)	428,436
**updateRef			565
}	

But if we didn't change anything we just leave it as the serialized object 
has the same value to reduce I/Os (such as search) in lines {
BPTree.java
[80:83],[105:107],[201:203],[204:206]
BPTreeInnerNode.java
[401:407],[408:414],[424:428],[429:433],[444:448]
BPTreeLeafNode.java
[232:238],[240:273][488:521]
}
The implementaion of reading and writing nodes to pages in disk is in
BPTreeNode.java
[223:238]	(write)		public void serializeNode() throws DBAppException 	
[239:258]	(read)		public BPTreeNode<T> deserializeNode(String name) throws DBAppException


The orginal implementation:
https://github.com/AhmadElsagheer/DBMS/tree/master/SagSolheeriman-A1/src/BPTree
((Many methods were added to that implementation to fit for our implementation; like different new methods to ease searching in the BPTree))


We added to it some new classes for handling duplicates 
*libs.General.OverflowReference.java
*libs.General.OverflowPage.java
*libs.General.GeneralReference.java
and implemented paging for the overflow buckets as well.