Skip to content

DEVELOPING.md

Latest commit

 

History

History
302 lines (204 loc) · 9.65 KB

DEVELOPING.md

File metadata and controls

302 lines (204 loc) · 9.65 KB
Raw

Developing

Prerequisites

Node.js

nvm is recommended for managing your Node.js installation. We provide a .nvmrc file so when you have nvm install simply run nvm use to switch to our currently recommended version of Node.js.

If you want to manually install Node.js, we recommend the current LTS version.

Yarn

Once you have Node.js installed, you should globally install Yarn by running the following:

npm install -g yarn

Cloning the code

If you just want to snoop around, clone our repository using:

https://github.com/neo4j/graphql.git

If you want to make a contribution, fork our repository and perform a clone, preferably over SSH:

git@github.com:USERNAME/graphql.git

You will then need to add our repository as an upstream:

git remote add upstream git@github.com:neo4j/graphql.git

You can then fetch and merge from the upstream to keep in sync.

Getting started

After cloning the repository, simply run the following from the root to get up and running:

yarn install

Code editor

Visual Studio Code comes highly recommended for working in this repository, and we additionally recommend the following extensions:

The Jest extension should automatically detect the tests for this repository and watch them in the Status Bar.

Using local Cypher-Builder

In some cases, it may be worth testing @neo4j/graphql with a local version of @neo4j/cypher-builder, to do this, you'll need a version of the Cypher Builder repository cloned locally.

In the Cypher Builder folder run:

  • yarn link
  • yarn build - This step needs to be done each time a change is done in the Cypher Builder

In the root of the neo4j/graphql monorepo run:

  • `yarn link -p [path-to-local-cypher-builder]

To unlink, in the graphql project:

  • yarn unlink @neo4j/cypher-builder

Testing

Testing locally

In order to run all of the tests, you will need to have a local instance of Neo4j running! We highly recommend Neo4j Desktop to easily get up and running with a local Neo4j instance.

  1. Create and start a new DBMS with a database named neo4j (default).
  2. Install APOC plugin for that DB.
  3. Run tests with yarn test.

This might cause you errors with running the tests!

We used to require an admin user to be created and further assigned permissions. We have since settled for the default neo4j user, which already has the required permissions.

Tests are run using Jest, which has been configured to allow for execution of test suites at any level in the project.

You can execute tests with a different database, user and password with the following command:

NEO_URL=neo4j://localhost:7687 NEO_USER=neo4j NEO_PASSWORD=password yarn test

The above command can additionally be run from packages/graphql, or any directory where there is a jest.config.js file!

Alternatively, you can put these environment variables in a .env file in the root of the repo which will automatically get picked up:

NEO_URL=neo4j://localhost:7687
NEO_USER=neo4j
NEO_PASSWORD=password

The above command can additionally be run from packages/graphql, or any directory where there is a jest.config.js file!

Additionally, for projects which have the appropriate Yarn scripts setup, you can run individual test suites. For instance, to run the TCK test suite of @neo4j/graphql, run the following from packages/graphql:

yarn test:tck

Verify TCK Tests

You can run all the TCK tests against the database to check that the Cypher generated is valid. This can be done with the env variable VERIFY_TCK

VERIFY_TCK=true yarn test:tck

Performance

packages/graphql has several performance benchmarks built in. To run the benchmarks:

  1. Go to packages/graphql
  2. Run yarn performance (by default, Database Query Performance test will be run)

Database Query Performance

The Database query benchmarks will translate several GraphQL queries and run the generated Cypher against a test dataset in Neo4j, measuring time and dbHits.

These will be run by default when executing yarn performance.

All .graphql files in tests/performance/graphql are part of the performance suite. To skip or run a test, append _skip or _only to the query, e.g.:

query SimpleUnionQuery_only {
    users {
        name
        liked {
            ... on Person {
                name
            }
            ... on Movie {
                title
            }
        }
    }
}

Saving metrics With the option -u a file performance.json will be generated, that can be used to compare metrics between runs. This is only available for database query benchmark.

Running Cypher queries

The performance tests can also run raw Cypher, to enable it, run yarn performance --cypher. Cypher queries must be located at tests/performance/databaseQuery/cypher. This allows for comparison between GraphQL and an alternative Cypher query.

Generating markdown

With the option --markdown the output will be formatted in markdown instead of the CLI. This is only available for database query benchmark.

Schema Generation

This benchmark runs the schema generation for a large GraphQL Schema and outputs the time that took to generate it. No database needed.

To execute this, run yarn performance --schema.

Subgraph Schema Alternatively, the schema can be run ready for subgraph by executing yarn performance --subgraph-schema.

Translation

This benchmark will measure the time it takes to translate a certain GraphQL query (without hitting the database). All of the queries in the graphql folder will be used, along with the queries located in translation/graphql _skip and _only can be used to limit how many queries are run.

This can be run with yarn performance --translation.

Note that the output will be rounded to milliseconds.

Change test runs By default, each query will be translated 100 times, and the total time will be shown.

With the option --runs you can change how many runs to do per query (e.g. yarn performance --translation --runs 1000).

With the flag --single each query translation will only run once. Note that this option makes the tests faster and logging easier but will yield less accurate results.

Make runs asynchronous With the flag --async, all the runs for each test will be done asynchronously, this makes the tests slightly faster, and it may yield a more accurate depiction of a real server, but makes it harder to get reliable results and debugging.

Increase Schema Size The option --schemaSize allows to artificially increase the schema size x times.

For instance --schemaSize 500 will generate a schema roughly 500x. This is done by appending a test default schema to the base schema

Linting/formatting

We use ESLint for linting and Prettier for code formatting. Contributions must adhere to our linting and formatting rules.

Adding a new project to the monorepo

For the sake of completeness, add an entry for the new project into the following files in the root of the repository:

  • tsconfig.json (references entry)
  • jest.config.base.js (moduleNameMapper entry)

Dependencies within the monorepo

Adding dependencies within the monorepo is a little bit tricky because of the fact that we need to use uncompiled TypeScript code.

This section will contain a couple of example use cases, one for production dependencies and one for test dependencies. They will use an example project with name "project" in packages/project, and the dependency in question will be @neo4j/graphql.

Production dependencies

First things first, install the dependency as you normally would:

yarn add @neo4j/graphql

Now, inside packages/project/src/tsconfig.json, this will need to look something like:

{
    "extends": "../../../tsconfig.base.json",
    "compilerOptions": {
        "baseUrl": "./",
        "outDir": "../dist",
        "paths": {
            "@neo4j/graphql": ["../../graphql/src"]
        }
    },
    "references": [{ "path": "../../graphql/src/tsconfig.json" }]
}

The real key entries here are:

  • baseUrl - for all of the relative references in this file, this will tell tsc where to start from
  • paths - this will translate import statements in code to the relative dependency
  • references - gives TypeScript "permission" to accesss the projects at these paths

Finally, it is highly likely that Jest will also need access to this internal dependency, so packages/project/jest.config.js will need to look like:

const globalConf = require("../../jest.config.base");

module.exports = {
    ...globalConf,
    displayName: "@neo4j/graphql-project",
    roots: ["<rootDir>/packages/project/src", "<rootDir>/packages/project/tests"],
    coverageDirectory: "<rootDir>/packages/project/coverage/",
    globals: {
        "ts-jest": {
            tsconfig: "<rootDir>/packages/project/src/tsconfig.json",
        },
    },
};

The magic sauce here is globals/ts-jest/tsconfig, which tells Jest which TypeScript configuration to use.

Test dependencies

Let's say you just need an internal dependency for testing purposes. You would install this as a dev dependency:

yarn add --dev @neo4j/graphql

You then need to follow the steps above, but using packages/project/tests/tsconfig.json instead of the production tsconfig.json file.