Gets most recent tree committed to the store

Load a specific snapshot from the store, 0th version = load most recent version as a special case note: 0th tree is not stored in disk also note tha

put a key value in the tree, if the value exists, it's overwritten. ToDO: it should ignore duplicate key value, if first using a get and then a put

commit the tree to disk, the current version

start a new memory backed store which may be useful for testing and other temporaray use cases.

open/create a disk based store, if the directory pre-exists, it is used as is. Since we are an append only keyvalue store, we do not delete any data.

Get current version number of tree

Commit the tree (or a number of trees) to persistance, write a new snapshot which can be accessed henceforth without any modifications Commiting multi

all puts must be checked with deduplication and skipped if duplicate

4 billion files each of 4 GB seems to be enough for quite some time, we will run out of handles much earlier note that the structure is independant o

get Cursor which is used as an iterator that can traverse over all key/value pairs in a tree in hash sorted order.

we are here means we have a currently open file this function is single threaded

Next moves the cursor to the next item in the tree and returns its key and value.If the tree is empty then an error is returned.If the cursor is at th

this function will descend and reach the next or previous value

we can get away with runtime type detection

This function can be used to diff 2 trees and thus find all the keys which have been deleted, modified, inserted. The algorithm is linear time in the

First moves the cursor to the first item in the tree and returns its key and value. If the tree is empty then an error is returned. The returned key a

Check whether the tree is currently dirty or not

should we return a copy

Load a versioned tree from the store all trees have there own version number

versions are 1 based

prepare the structure for reuse

Generate proof of any key, which can be used to prove whether the key exists or not.Please note that the tree root hash (tree.Hash()) is not part of t

Get parent version number of tree from which this tree was derived, they might not be sequential but they will be monotonically increasing this can be

Prev moves the cursor to the prev item in the tree and returns its key and value.If the tree is empty then an error is returned.If the cursor is at th

leafs return nil,false, inner returns nil, false if both children are present or absent, if single child is present, it is returned nodes can only be

load tree using the specfic global version

Gets highest stored version number of the specific tree

Unmarshal follows reverse of marshal to deserialize the array of bytes to proof for verification.

todo we need to take care to prune single branches to achieve same root hash insertion/deletion the returns are in this order empty, changed, err

Gets the tree which has specific roothash

Gets the tree which has specific tag NOTE: same tags might point to different trees in different snapshots of db

Last moves the cursor to the last item in the tree and returns its key and value. If the tree is empty then an error is returned. The returned key and

first byte is skipped and processed elsewhere

this is mostly recursive and must skip non modified branches reusing them and must skip dirty parts

all empty nodes have this hash

versions are 1 based

Reload the tree from the disk, causing all current changes to be discarded,

Serialize the proof to a byte array

Serialize the proof to a bytes Buffer the following are the size requirements for proof 1 byte for version 1 byte for type varint trace length

if the proof is for existence for a key, it's associated value can be read here

add paths

extract changes one bye one

this is never recursive leaf marshalling is done at only one place while committing this is done here avoid an allocation which can be done from the

encode findex,fpos

init and load some items from the store

load all files from the disk we may need to increase file handles

store highest version of tree

Get a specifically value associated with a key TODO, we need to expose this in other forms so as memory allocations and better error detection could b

should we return a copy

Give the merkle hash of the entire tree

this always assummes that keyhash already matches to new keyhash this function is only used once , in node_inner.go insert

each step consists of generating pseudorandom data, which is first committed and then verified after each step