This repo contains example uses of CockroachDB with popular ORMs. Each example will implement the sample application spec presented below.

See the CockroachDB ORM Compatibility Plan for a roadmap towards supporting various ORMs.

Clone this repo into your $GOPATH manually, e.g.,

$ cd ~/go/src/github.com/cockroachdb
$ git clone https://github.com/cockroachdb/examples-orms

This is required because this project uses Go to drive the automated tests, so it will look for things in your $GOPATH. If you try to clone it to a non-$GOPATH directory, it will fail roughly as follows:

$ cd ~/some/random/dir/examples-orms
$ make test
go test -v -i ./testing
testing/api_handler.go:13:2: cannot find package "github.com/cockroachdb/examples-orms/go/gorm/model" in any of:
	/usr/local/Cellar/go/1.10/libexec/src/github.com/cockroachdb/examples-orms/go/gorm/model (from $GOROOT)
	/Users/rloveland/go/src/github.com/cockroachdb/examples-orms/go/gorm/model (from $GOPATH)
testing/api_handler.go:11:2: cannot find package "github.com/pkg/errors" in any of:
	/usr/local/Cellar/go/1.10/libexec/src/github.com/pkg/errors (from $GOROOT)
	/Users/rloveland/go/src/github.com/pkg/errors (from $GOPATH)
make: *** [test] Error 1

However, this is not actually a Go project, so go get -d will also fail (hence the need to manually clone).

$ go get -d github.com/cockroachdb/examples-orms
package github.com/cockroachdb/examples-orms: no Go files in /Users/rloveland/go/src/github.com/cockroachdb/examples-orms

To run automated testing against all ORMs using the latest binary of CockroachDB, run:

$ make test

To run automated testing against all ORMs using a custom CockroachDB binary, run:

$ make test COCKROACH_BINARY=/path/to/binary/cockroach

To run automated testing against a specific ORM, you can also specify the name with:

$ make test COCKROACH_BINARY=/path/to/binary/cockroach TESTS=TestSequelize/password

These tests require dependencies to be installed on your system. You can install them with:

$ make deps

While running tests locally, you may find it useful to comment out the lines in the Makefile that install the dependencies for a tool that you don't want to test.

A final option is to run using docker, so that a reproducible build environment is used.

$ make dockertest COCKROACH_BINARY=/path/to/binary/cockroach

The repository contains a set of directories named after programming languages. Beneath these language directories are sub-directories named after specific ORM used for example application implementations.

Each ORM example uses whatever build tool is standard for the language, but provides a standardized Makefile with a start rule, which will start an instance of the sample application.

For instance, the directory structure for an example application of the Hibernate ORM will look like:

java
└── hibernate
    ├── Makefile
    └── example_source

The sample application which each example implements is a JSON REST API modeling a company with customers, products, and orders. The API exposes access to the management of this company.

The purpose of the example application is to test common data access patterns so that we can stress various features and implementation details of each language/ORM.

An ideal schema diagram for the sample application looks like:

Customer
  |
  v
Order  <->  Product

The schema is implemented by each application using ORM-specific constructs to look as close as possible to:

CREATE DATABASE IF NOT EXISTS company_{language}_{ORM};

CREATE TABLE IF NOT EXISTS customers (
  id           SERIAL PRIMARY KEY,
  name         STRING
);

CREATE TABLE IF NOT EXISTS orders (
  id           SERIAL PRIMARY KEY,
  subtotal     DECIMAL(18,2)
  customer_id  INT    REFERENCES customers(id),
);

CREATE TABLE IF NOT EXISTS products (
  id           SERIAL PRIMARY KEY,
  name         STRING,
  price        DECIMAL(18,2)
);

CREATE TABLE IF NOT EXISTS product_orders (
  id           SERIAL PRIMARY KEY,
  product_id   INT    REFERENCES products(id),
  order_id     INT    REFERENCES orders(id) ON DELETE CASCADE
);

Each example will expose a RESTful JSON API. The endpoints and example curl command lines are:

GET    /customer
    curl http://localhost:6543/customer

GET    /customer/:id
    curl http://localhost:6543/customer/1

POST   /customer
    curl -X POST -d '{"id": 1, "name": "bob"}' http://localhost:6543/customer

PUT    /customer/:id
    curl -X PUT -d '{"id": 2, "name": "robert"}' http://localhost:6543/customer/1

DELETE /customer
    curl -X DELETE http://localhost:6543/customer/1

GET    /product
    curl http://localhost:6543/product

GET    /product/:id
    curl http://localhost:6543/product/1

POST   /product
    curl -X POST -d '{"id": 1, "name": "apple", "price": 0.30}' http://localhost:6543/product

PUT    /product
DELETE /product

GET    /order
    curl http://localhost:6543/order

GET    /order/:id
    curl http://localhost:6543/order/1

POST   /order
    curl -X POST -d '{"id": 1, "subtotal": 18.2, "customer": {"id": 1}}' http://localhost:6543/order

PUT    /order
DELETE /order

The semantics of each endpoint will be fleshed out when necessary.

• Can the schema be completely standardized across ORMs without too much of a hassle with overriding default type and naming conventions?