formal/orm

2.0.0 2024-02-25 14:13 UTC

Requires

• php: ~8.2
• formal/access-layer: ~2.15
• innmind/filesystem: ~7.4
• innmind/http-transport: ~7.2
• innmind/immutable: ~4.15|~5.0
• innmind/json: ~1.3
• innmind/reflection: ~5.1
• innmind/specification: ~3.0
• innmind/time-continuum: ~3.3
• innmind/type: ~1.2
• innmind/url-template: ~3.1
• innmind/validation: ^1.1.1
• ramsey/uuid: ~4.7

Requires (Dev)

• innmind/black-box: ~5.1
• innmind/coding-standard: ~2.0
• innmind/operating-system: ~4.2
• vimeo/psalm: ~5.13

MIT

• Baptiste Langlade

README

This ORM (Object-Relational Mapping) is focused to work in long living processes and being able to work in an asynchronous context.

This is achieved by:

• being memory efficient (objects are not kept in memory)
• relying on the Innmind platform
• using immutable objects

Installation

composer require formal/orm

Usage

use Formal\ORM\{
    Manager,
    Sort,
};
use Formal\AccessLayer\Connection\PDO;
use Innmind\Url\Url;

$manager = Manager::sql(
    PDO::of(Url::of('mysql://user:pwd@host:3306/database?charset=utf8mb4')),
);
$_ = $manager
    ->repository(YourAggregate::class)
    ->all()
    ->sort('someProperty', Sort::asc)
    ->drop(150)
    ->take(50)
    ->foreach(static fn($aggregate) => doStuff($aggregate));

This simple example will retrieve from the database 50 elements (from index 151 to 200) sorted by someProperty in ascending order and will call the function doStuff on each aggregate.

Note

The elements are streamed meaning only one aggregate is in memory at a time allowing you to deal with long lists of elements in a memory safe way.

Documentation

Full documentation available in the documentation folder.

Benchmark

A small benchmark as a reference point for the performance of this ORM consists in generating and persisting 100K users in a single transaction and then loading them.

time php benchmark/fill_storage.php
php benchmark/fill_storage.php  222.24s user 5.20s system 60% cpu 6:18.40 total
time php benchmark/load.php
Memory: 40.00 Mo
php benchmark/load.php  11.06s user 0.08s system 97% cpu 11.388 total

This means the ORM can load 1 aggregate in 0.1 millisecond.

This was run on a MacbookPro 16" with a M1 Max with the mariadb running inside Docker.

Note: If all the aggregates were to be stored in memory it would take around 2Go of RAM and 15 seconds to complete.