Persistence

This requires latest rupy in hosted mode.

For true redundancy you need to use some custom cluster database since we host across multiple colocations! Ordinary databases are not secure or redundant in a distributed cluster environment out of the box, but most of all:

Database clients don't use async HTTP; you can't scale them easily.

On the other hand modern file systems are now cheap and fast (ext4) and async is becoming common place; See the entry on FUSE to learn more on how you can scale any µSOA cluster with async HTTP.

ROOT takes a little bit from all NoSQL databases: document, key-value and graph; It's basically a key-value store (Base58-JSON) with indexing (that we call sort) on custom fields (including full text search) in the JSON and graph relationships between the key-values.

The JSON objects and indexes are stored as a plain file (symbolic hard link). The relations and full text search are just binary files with lists of longs (the hashed Base58 key). Simple, powerful and in line with both Moore's and Murphy's laws.

Try the current version: root.rupy.se

TODO:

1. new async client
2. edit node
3. sort node text index
4. link node relation graph
5. find node full text search
6. binary file table data
7. pagination in lists
8. trim node & link

WISH:

1. link node many-to-many-to-many... relations, you can do this by folder path.
2. replicate any file, images would be nice!

Source: User, Root