cuprate/storage/database/README.md
hinto-janai 59adf6dcf8
Some checks failed
Audit / audit (push) Has been cancelled
CI / fmt (push) Has been cancelled
CI / typo (push) Has been cancelled
CI / ci (macos-latest, stable, bash) (push) Has been cancelled
CI / ci (ubuntu-latest, stable, bash) (push) Has been cancelled
CI / ci (windows-latest, stable-x86_64-pc-windows-gnu, msys2 {0}) (push) Has been cancelled
Deny / audit (push) Has been cancelled
Doc / build (push) Has been cancelled
Doc / deploy (push) Has been cancelled
std::mem::{size,align}_of -> {size,align}_of (#254)
2024-08-10 00:09:25 +01:00

156 lines
No EOL
5.4 KiB
Markdown

Cuprate's database abstraction.
This documentation is mostly for practical usage of `cuprate-database`.
For a high-level overview, see the database section in
[Cuprate's architecture book](https://architecture.cuprate.org).
If you need blockchain specific capabilities, consider using the higher-level
[`cuprate-blockchain`](https://doc.cuprate.org/cuprate_blockchain) crate which builds upon this one.
# Purpose
This crate abstracts various database backends with traits.
All backends have the following attributes:
- [Embedded](https://en.wikipedia.org/wiki/Embedded_database)
- [Multiversion concurrency control](https://en.wikipedia.org/wiki/Multiversion_concurrency_control)
- [ACID](https://en.wikipedia.org/wiki/ACID)
- Are `(key, value)` oriented and have the expected API (`get()`, `insert()`, `delete()`)
- Are table oriented (`"table_name" -> (key, value)`)
- Allows concurrent readers
The currently implemented backends are:
- [`heed`](https://github.com/meilisearch/heed) (LMDB)
- [`redb`](https://github.com/cberner/redb)
# Terminology
To be more clear on some terms used in this crate:
| Term | Meaning |
|------------------|--------------------------------------|
| `Env` | The 1 database environment, the "whole" thing
| `DatabaseR{o,w}` | A _actively open_ readable/writable `key/value` store
| `Table` | Solely the metadata of a `Database` (the `key` and `value` types, and the name)
| `TxR{o,w}` | A read/write transaction
| `Storable` | A data type that can be stored in the database
The flow is `Env` -> `Tx` -> `Database`
Which reads as:
1. You have a database `Environment`
1. You open up a `Transaction`
1. You open a particular `Table` from that `Environment`, getting a `Database`
1. You can now read/write data from/to that `Database`
# Concrete types
You should _not_ rely on the concrete type of any abstracted backend.
For example, when using the `heed` backend, [`Env`]'s associated [`TxRw`] type
is `RefCell<heed::RwTxn<'_>>`. In order to ensure compatibility with other backends
and to not create backend-specific code, you should _not_ refer to that concrete type.
Use generics and trait notation in these situations:
- `impl<T: TxRw> Trait for Object`
- `fn() -> impl TxRw`
# `ConcreteEnv`
This crate exposes [`ConcreteEnv`], which is a non-generic/non-dynamic,
concrete object representing a database [`Env`]ironment.
The actual backend for this type is determined via feature flags.
This object existing means `E: Env` doesn't need to be spread all through the codebase,
however, it also means some small invariants should be kept in mind.
As `ConcreteEnv` is just a re-exposed type which has varying inner types,
it means some properties will change depending on the backend used.
For example:
- [`size_of::<ConcreteEnv>`]
- [`align_of::<ConcreteEnv>`]
Things like these functions are affected by the backend and inner data,
and should not be relied upon. This extends to any `struct/enum` that contains `ConcreteEnv`.
`ConcreteEnv` invariants you can rely on:
- It implements [`Env`]
- Upon [`Drop::drop`], all database data will sync to disk
Note that `ConcreteEnv` itself is not a clonable type,
it should be wrapped in [`std::sync::Arc`].
<!-- SOMEDAY: replace `ConcreteEnv` with `fn Env::open() -> impl Env`/
and use `<E: Env>` everywhere it is stored instead. This would allow
generic-backed dynamic runtime selection of the database backend, i.e.
the user can select which database backend they use. -->
# Defining tables
Most likely, your crate building on-top of `cuprate_database` will
want to define all tables used at compile time.
If this is the case, consider using the [`define_tables`] macro
to bulk generate zero-sized marker types that implement [`Table`].
This macro also generates other convenient traits specific to _your_ tables.
# Feature flags
Different database backends are enabled by the feature flags:
- `heed` (LMDB)
- `redb`
The default is `heed`.
`tracing` is always enabled and cannot be disabled via feature-flag.
<!-- FIXME: tracing should be behind a feature flag -->
# Examples
The below is an example of using `cuprate-database`.
```rust
use cuprate_database::{
ConcreteEnv,
config::ConfigBuilder,
Env, EnvInner,
DatabaseRo, DatabaseRw, TxRo, TxRw,
};
# fn main() -> Result<(), Box<dyn std::error::Error>> {
// Create a configuration for the database environment.
let tmp_dir = tempfile::tempdir()?;
let db_dir = tmp_dir.path().to_owned();
let config = ConfigBuilder::new(db_dir.into()).build();
// Initialize the database environment.
let env = ConcreteEnv::open(config)?;
// Define metadata for a table.
struct Table;
impl cuprate_database::Table for Table {
// The name of the table is "table".
const NAME: &'static str = "table";
// The key type is a `u8`.
type Key = u8;
// The key type is a `u64`.
type Value = u64;
}
// Open up a transaction + tables for writing.
let env_inner = env.env_inner();
let tx_rw = env_inner.tx_rw()?;
// We must create the table first or the next line will error.
env_inner.create_db::<Table>(&tx_rw)?;
let mut table = env_inner.open_db_rw::<Table>(&tx_rw)?;
// Write data to the table.
table.put(&0, &1)?;
// Commit the data written.
drop(table);
TxRw::commit(tx_rw)?;
// Read the data, assert it is correct.
let tx_ro = env_inner.tx_ro()?;
let table = env_inner.open_db_ro::<Table>(&tx_ro)?;
assert_eq!(table.first()?, (0, 1));
# Ok(()) }
```