Follow us on Twitter! @nodepractices
Read in a different language: CN, FR, BR, RU, PL, JA, EU (ES, HE, KR and TR in progress! )
-
๐ฐ Modernized to 2024: Tons of text edits, new recommended libraries, and some new best practices
-
โจ Easily focus on new content: Already visited before? Search for
#new
or#updated
tags for new content only -
๐ Curious to see examples? We have a starter: Visit Practica.js, our application example and boilerplate (beta) to see some practices in action
1. You are reading dozens of the best Node.js articles - this repository is a summary and curation of the top-ranked content on Node.js best practices, as well as content written here by collaborators
2. It is the largest compilation, and it is growing every week - currently, more than 80 best practices, style guides, and architectural tips are presented. New issues and pull requests are created every day to keep this live book updated. We'd love to see you contributing here, whether that is fixing code mistakes, helping with translations, or suggesting brilliant new ideas. See our writing guidelines here
3. Best practices have additional info - most bullets include a ๐Read More link that expands on the practice with code examples, quotes from selected blogs, and more information
Learn with me: As a consultant, I engage with worldwide teams on various activities like workshops and code reviews. ๐AND... Hold on, I've just launched my beyond-the-basics testing course, which is on a ๐ limited-time sale until August 7th
1. Project Architecture Practices (6)
โโ1.1 Structure your solution by components #strategic
#updated
โโ1.2 Layer your components, keep the web layer within its boundaries #strategic
#updated
โโ1.3 Wrap common utilities as packages, consider publishing
โโ1.4 Use environment aware, secure and hierarchical config #updated
โโ1.5 Consider all the consequences when choosing the main framework #new
โโ1.6 Use TypeScript sparingly and thoughtfully #new
2. Error Handling Practices (12)
โโ2.1 Use Async-Await or promises for async error handling
โโ2.2 Extend the built-in Error object #strategic
#updated
โโ2.3 Distinguish operational vs programmer errors #strategic
#updated
โโ2.4 Handle errors centrally, not within a middleware #strategic
โโ2.5 Document API errors using OpenAPI or GraphQL
โโ2.6 Exit the process gracefully when a stranger comes to town #strategic
โโ2.7 Use a mature logger to increase errors visibility #updated
โโ2.8 Test error flows using your favorite test framework #updated
โโ2.9 Discover errors and downtime using APM products
โโ2.10 Catch unhandled promise rejections #updated
โโ2.11 Fail fast, validate arguments using a dedicated library
โโ2.12 Always await promises before returning to avoid a partial stacktrace #new
โโ2.13 Subscribe to event emitters 'error' event #new
3. Code Style Practices (12)
โโ3.1 Use ESLint #strategic
โโ3.2 Use Node.js eslint extension plugins #updated
โโ3.3 Start a Codeblock's Curly Braces on the Same Line
โโ3.4 Separate your statements properly
โโ3.5 Name your functions
โโ3.6 Use naming conventions for variables, constants, functions and classes
โโ3.7 Prefer const over let. Ditch the var
โโ3.8 Require modules first, not inside functions
โโ3.9 Set an explicit entry point to a module/folder #updated
โโ3.10 Use the === operator
โโ3.11 Use Async Await, avoid callbacks #strategic
โโ3.12 Use arrow function expressions (=>)
โโ3.13 Avoid effects outside of functions #new
4. Testing And Overall Quality Practices (13)
โโ4.1 At the very least, write API (component) testing #strategic
โโ4.2 Include 3 parts in each test name #new
โโ4.3 Structure tests by the AAA pattern #strategic
โโ4.4 Ensure Node version is unified #new
โโ4.5 Avoid global test fixtures and seeds, add data per-test #strategic
โโ4.6 Tag your tests #advanced
โโ4.7 Check your test coverage, it helps to identify wrong test patterns
โโ4.8 Use production-like environment for e2e testing
โโ4.9 Refactor regularly using static analysis tools
โโ4.10 Mock responses of external HTTP services #advanced #new
#advanced
โโ4.11 Test your middlewares in isolation
โโ4.12 Specify a port in production, randomize in testing #new
โโ4.13 Test the five possible outcomes #strategic #new
5. Going To Production Practices (19)
โโ5.1. Monitoring #strategic
โโ5.2. Increase the observability using smart logging #strategic
โโ5.3. Delegate anything possible (e.g. gzip, SSL) to a reverse proxy #strategic
โโ5.4. Lock dependencies
โโ5.5. Guard process uptime using the right tool
โโ5.6. Utilize all CPU cores
โโ5.7. Create a โmaintenance endpointโ
โโ5.8. Discover the unknowns using APM products #advanced
#updated
โโ5.9. Make your code production-ready
โโ5.10. Measure and guard the memory usage #advanced
โโ5.11. Get your frontend assets out of Node
โโ5.12. Strive to be stateless #strategic
โโ5.13. Use tools that automatically detect vulnerabilities
โโ5.14. Assign a transaction id to each log statement #advanced
โโ5.15. Set NODE_ENV=production
โโ5.16. Design automated, atomic and zero-downtime deployments #advanced
โโ5.17. Use an LTS release of Node.js
โโ5.18. Log to stdout, avoid specifying log destination within the app #updated
โโ5.19. Install your packages with npm ci #new
6. Security Practices (25)
โโ6.1. Embrace linter security rules
โโ6.2. Limit concurrent requests using a middleware
โโ6.3 Extract secrets from config files or use packages to encrypt them #strategic
โโ6.4. Prevent query injection vulnerabilities with ORM/ODM libraries #strategic
โโ6.5. Collection of generic security best practices
โโ6.6. Adjust the HTTP response headers for enhanced security
โโ6.7. Constantly and automatically inspect for vulnerable dependencies #strategic
โโ6.8. Protect Users' Passwords/Secrets using bcrypt or scrypt #strategic
โโ6.9. Escape HTML, JS and CSS output
โโ6.10. Validate incoming JSON schemas #strategic
โโ6.11. Support blocklisting JWTs
โโ6.12. Prevent brute-force attacks against authorization #advanced
โโ6.13. Run Node.js as non-root user
โโ6.14. Limit payload size using a reverse-proxy or a middleware
โโ6.15. Avoid JavaScript eval statements
โโ6.16. Prevent evil RegEx from overloading your single thread execution
โโ6.17. Avoid module loading using a variable
โโ6.18. Run unsafe code in a sandbox
โโ6.19. Take extra care when working with child processes #advanced
โโ6.20. Hide error details from clients
โโ6.21. Configure 2FA for npm or Yarn #strategic
โโ6.22. Modify session middleware settings
โโ6.23. Avoid DOS attacks by explicitly setting when a process should crash #advanced
โโ6.24. Prevent unsafe redirects
โโ6.25. Avoid publishing secrets to the npm registry
โโ6.26. 6.26 Inspect for outdated packages
โโ6.27. Import built-in modules using the 'node:' protocol #new
7. Performance Practices (2) (Work In Progress๏ธ โ๏ธ)
โโ7.1. Don't block the event loop
โโ7.2. Prefer native JS methods over user-land utils like Lodash
8. Docker Practices (15)
โโ8.1 Use multi-stage builds for leaner and more secure Docker images #strategic
โโ8.2. Bootstrap using node command, avoid npm start
โโ8.3. Let the Docker runtime handle replication and uptime #strategic
โโ8.4. Use .dockerignore to prevent leaking secrets
โโ8.5. Clean-up dependencies before production
โโ8.6. Shutdown smartly and gracefully #advanced
โโ8.7. Set memory limits using both Docker and v8 #advanced
#strategic
โโ8.8. Plan for efficient caching
โโ8.9. Use explicit image reference, avoid latest tag
โโ8.10. Prefer smaller Docker base images
โโ8.11. Clean-out build-time secrets, avoid secrets in args #strategic #new
โโ8.12. Scan images for multi layers of vulnerabilities #advanced
โโ8.13 Clean NODE_MODULE cache
โโ8.14. Generic Docker practices
โโ8.15. Lint your Dockerfile #new
TL;DR: The root of a system should contain folders or repositories that represent reasonably sized business modules. Each component represents a product domain (i.e., bounded context), like 'user-component', 'order-component', etc. Each component has its own API, logic, and logical database. What is the significant merit? With an autonomous component, every change is performed over a granular and smaller scope - the mental overload, development friction, and deployment fear are much smaller and better. As a result, developers can move much faster. This does not necessarily demand physical separation and can be achieved using a Monorepo or with a multi-repo
my-system
โโ apps (components)
โ โโ orders
โ โโ users
โ โโ payments
โโ libraries (generic cross-component functionality)
โ โโ logger
โ โโ authenticator
Otherwise: when artifacts from various modules/topics are mixed together, there are great chances of a tightly-coupled 'spaghetti' system. For example, in an architecture where 'module-a controller' might call 'module-b service', there are no clear modularity borders - every code change might affect anything else. With this approach, developers who code new features struggle to realize the scope and impact of their change. Consequently, they fear breaking other modules, and each deployment becomes slower and riskier
๐ Read More: structure by components
TL;DR: Each component should contain 'layers' - a dedicated folder for common concerns: 'entry-point' where controller lives, 'domain' where the logic lives, and 'data-access'. The primary principle of the most popular architectures is to separate the technical concerns (e.g., HTTP, DB, etc) from the pure logic of the app so a developer can code more features without worrying about infrastructural concerns. Putting each concern in a dedicated folder, also known as the 3-Tier pattern, is the simplest way to meet this goal
my-system
โโ apps (components)
โ โโ component-a
โ โโ entry-points
โ โ โโ api # controller comes here
โ โ โโ message-queue # message consumer comes here
โ โโ domain # features and flows: DTO, services, logic
โ โโ data-access # DB calls w/o ORM
Otherwise: It's often seen that developer pass web objects like request/response to functions in the domain/logic layer - this violates the separation principle and makes it harder to access later the logic code by other clients like testing code, scheduled jobs, message queues, etc
๐ Read More: layer your app
TL;DR: Place all reusable modules in a dedicated folder, e.g., "libraries", and underneath each module in its own folder, e.g., "/libraries/logger". Make the module an independent package with its own package.json file to increase the module encapsulation, and allows future publishing to a repository. In a Monorepo setup, modules can be consumed by 'npm linking' to their physical paths, using ts-paths or by publishing and installing from a package manager repository like the npm registry
my-system
โโ apps (components)
โ โโ component-a
โโ libraries (generic cross-component functionality)
โ โโ logger
โ โ โโ package.json
โ โ โโ src
โ โ โ โโ index.js
Otherwise: Clients of a module might import and get coupled to internal functionality of a module. With a package.json at the root, one can set a package.json.main or package.json.exports to explicitly tell which files and functions are part of the public interface
๐ Read More: Structure by feature
TL;DR: A flawless configuration setup should ensure (a) keys can be read from file AND from environment variable (b) secrets are kept outside committed code (c) config is hierarchical for easier findability (d) typing support (e) validation for failing fast (f) Specify default for each key. There are a few packages that can help tick most of those boxes like convict, env-var, zod, and others
Otherwise: Consider a mandatory environment variable that wasn't provided. The app starts successfully and serve requests, some information is already persisted to DB. Then, it's realized that without this mandatory key the request can't complete, leaving the app in a dirty state
๐ Read More: configuration best practices
TL;DR: When building apps and APIs, using a framework is mandatory. It's easy to overlook alternative frameworks or important considerations and then finally land on a sub optimal option. As of 2023/2024, we believe that these four frameworks are worth considering: Nest.js, Fastify, express, and Koa. Click read more below for a detailed pros/cons of each framework. Simplistically, we believe that Nest.js is the best match for teams who wish to go OOP and/or build large-scale apps that can't get partitioned into smaller autonomous components. Fastify is our recommendation for apps with reasonably-sized components (e.g., Microservices) that are built around simple Node.js mechanics. Read our full considerations guide here
Otherwise: Due to the overwhelming amount of considerations, it's easy to make decisions based on partial information and compare apples with oranges. For example, it's believed that Fastify is a minimal web-server that should get compared with express only. In reality, it's a rich framework with many official plugins that cover many concerns
๐ Read More: Choosing the right framework
TL;DR: Coding without type safety is no longer an option, TypeScript is the most popular option for this mission. Use it to define variables and functions return types. With that, it is also a double edge sword that can greatly encourage complexity with its additional ~ 50 keywords and sophisticated features. Consider using it sparingly, mostly with simple types, and utilize advanced features only when a real need arises
Otherwise: Researches show that using TypeScript can help in detecting ~20% bugs earlier. Without it, also the developer experience in the IDE is intolerable. On the flip side, 80% of other bugs were not discovered using types. Consequently, typed syntax is valuable but limited. Only efficient tests can discover the whole spectrum of bugs, including type-related bugs. It might also defeat its purpose: sophisticated code features are likely to increase the code complexity, which by itself increases both the amount of bugs and the average bug fix time
๐ Read More: TypeScript considerations
TL;DR: Handling async errors in callback style is probably the fastest way to hell (a.k.a the pyramid of doom). The best gift you can give to your code is using Promises with async-await which enables a much more compact and familiar code syntax like try-catch
Otherwise: Node.js callback style, function(err, response), is a promising way to un-maintainable code due to the mix of error handling with casual code, excessive nesting, and awkward coding patterns
๐ Read More: avoiding callbacks
TL;DR: Some libraries throw errors as a string or as some custom type โ this complicates the error handling logic and the interoperability between modules. Instead, create app error object/class that extends the built-in Error object and use it whenever rejecting, throwing or emitting an error. The app error should add useful imperative properties like the error name/code and isCatastrophic. By doing so, all errors have a unified structure and support better error handling. There is no-throw-literal
ESLint rule that strictly checks that (although it has some limitations which can be solved when using TypeScript and setting the @typescript-eslint/no-throw-literal
rule)
Otherwise: When invoking some component, being uncertain which type of errors come in return โ it makes proper error handling much harder. Even worse, using custom types to describe errors might lead to loss of critical error information like the stack trace!
๐ Read More: using the built-in error object
TL;DR: Operational errors (e.g. API received an invalid input) refer to known cases where the error impact is fully understood and can be handled thoughtfully. On the other hand, catastrophic error (also known as programmer errors) refers to unusual code failures that dictate to gracefully restart the application
Otherwise: You may always restart the application when an error appears, but why let ~5000 online users down because of a minor, predicted, operational error? The opposite is also not ideal โ keeping the application up when an unknown catastrophic issue (programmer error) occurred might lead to an unpredicted behavior. Differentiating the two allows acting tactfully and applying a balanced approach based on the given context
๐ Read More: operational vs programmer error
TL;DR: Error handling logic such as logging, deciding whether to crash and monitoring metrics should be encapsulated in a dedicated and centralized object that all entry-points (e.g. APIs, cron jobs, scheduled jobs) call when an error comes in
Otherwise: Not handling errors within a single place will lead to code duplication and probably to improperly handled errors
๐ Read More: handling errors in a centralized place
TL;DR: Let your API callers know which errors might come in return so they can handle these thoughtfully without crashing. For RESTful APIs, this is usually done with documentation frameworks like OpenAPI. If you're using GraphQL, you can utilize your schema and comments as well
Otherwise: An API client might decide to crash and restart only because it received back an error it couldnโt understand. Note: the caller of your API might be you (very typical in a microservice environment)
๐ Read More: documenting API errors in Swagger or GraphQL
TL;DR: When an unknown error occurs (catastrophic error, see best practice 2.3) - there is uncertainty about the application healthiness. In this case, there is no escape from making the error observable, shutting off connections and exiting the process. Any reputable runtime framework like Dockerized services or cloud serverless solutions will take care to restart
Otherwise: When an unfamiliar exception occurs, some object might be in a faulty state (e.g. an event emitter which is used globally and not firing events anymore due to some internal failure) and all future requests might fail or behave crazily
๐ Read More: shutting the process
TL;DR: A robust logging tools like Pino or Winston increases the errors visibility using features like log-levels, pretty print coloring and more. Console.log lacks these imperative features and should be avoided. The best in class logger allows attaching custom useful properties to log entries with minimized serialization performance penalty. Developers should write logs to stdout
and let the infrastructure pipe the stream to the appropriate log aggregator
Otherwise: Skimming through console.logs or manually through messy text file without querying tools or a decent log viewer might keep you busy at work until late
๐ Read More: using a mature logger
TL;DR: Whether professional automated QA or plain manual developer testing โ Ensure that your code not only satisfies positive scenarios but also handles and returns the right errors. On top of this, simulate deeper error flows like uncaught exceptions and ensure that the error handler treat these properly (see code examples within the "read more" section)
Otherwise: Without testing, whether automatically or manually, you canโt rely on your code to return the right errors. Without meaningful errors โ thereโs no error handling
๐ Read More: testing error flows
TL;DR: Monitoring and performance products (a.k.a APM) proactively gauge your codebase or API so they can automagically highlight errors, crashes, and slow parts that you were missing
Otherwise: You might spend great effort on measuring API performance and downtimes, probably youโll never be aware which are your slowest code parts under real-world scenario and how these affect the UX
๐ Read More: using APM products
TL;DR: Any exception thrown within a promise will get swallowed and discarded unless a developer didnโt forget to explicitly handle it. Even if your code is subscribed to process.uncaughtException
! Overcome this by registering to the event process.unhandledRejection
Otherwise: Your errors will get swallowed and leave no trace. Nothing to worry about
๐ Read More: catching unhandled promise rejection
TL;DR: Assert API input to avoid nasty bugs that are much harder to track later. The validation code is usually tedious unless you are using a modern validation library like ajv, zod, or typebox
Otherwise: Consider this โ your function expects a numeric argument โDiscountโ which the caller forgets to pass, later on, your code checks if Discount!=0 (amount of allowed discount is greater than zero), then it will allow the user to enjoy a discount. OMG, what a nasty bug. Can you see it?
TL;DR: Always do return await
when returning a promise to benefit full error stacktrace. If a
function returns a promise, that function must be declared as async
function and explicitly
await
the promise before returning it
Otherwise: The function that returns a promise without awaiting won't appear in the stacktrace. Such missing frames would probably complicate the understanding of the flow that leads to the error, especially if the cause of the abnormal behavior is inside of the missing function
๐ Read More: returning promises
TL;DR: Unlike typical functions, a try-catch clause won't get errors that originate from Event Emitters and anything inherited from it (e.g., streams). Instead of try-catch, subscribe to an event emitter's 'error' event so your code can handle the error in context. When dealing with EventTargets (the web standard version of Event Emitters) there are no 'error' event and all errors end in the process.on('error) global event - in this case, at least ensure that the process crash or not based on the desired context. Also, mind that error originating from asynchronous event handlers are not get caught unless the event emitter is initialized with {captureRejections: true}
Otherwise: Event emitters are commonly used for global and key application functionality such as DB or message queue connection. When this kind of crucial objects throw an error, at best the process will crash due to unhandled exception. Even worst, it will stay alive as a zombie while a key functionality is turned off
TL;DR: ESLint is the de-facto standard for checking possible code errors and fixing code style, not only to identify nitty-gritty spacing issues but also to detect serious code anti-patterns like developers throwing errors without classification. Though ESLint can automatically fix code styles, other tools like prettier are more powerful in formatting the fix and work in conjunction with ESLint
Otherwise: Developers will focus on tedious spacing and line-width concerns and time might be wasted overthinking the project's code style
๐ Read More: Using ESLint and Prettier
TL;DR: On top of ESLint standard rules that cover vanilla JavaScript, add Node.js specific plugins like eslint-plugin-node, eslint-plugin-mocha and eslint-plugin-node-security, eslint-plugin-require, /eslint-plugin-jest and other useful rules
Otherwise: Many faulty Node.js code patterns might escape under the radar. For example, developers might require(variableAsPath) files with a variable given as a path which allows attackers to execute any JS script. Node.js linters can detect such patterns and complain early
TL;DR: The opening curly braces of a code block should be on the same line as the opening statement
// Do
function someFunction() {
// code block
}
// Avoid
function someFunction()
{
// code block
}
Otherwise: Deferring from this best practice might lead to unexpected results, as seen in the StackOverflow thread below:
๐ Read more: "Why do results vary based on curly brace placement?" (StackOverflow)
No matter if you use semicolons or not to separate your statements, knowing the common pitfalls of improper linebreaks or automatic semicolon insertion, will help you to eliminate regular syntax errors.
TL;DR: Use ESLint to gain awareness about separation concerns. Prettier or Standardjs can automatically resolve these issues.
Otherwise: As seen in the previous section, JavaScript's interpreter automatically adds a semicolon at the end of a statement if there isn't one, or considers a statement as not ended where it should, which might lead to some undesired results. You can use assignments and avoid using immediately invoked function expressions to prevent most of the unexpected errors.
// Do
function doThing() {
// ...
}
doThing()
// Do
const items = [1, 2, 3]
items.forEach(console.log)
// Avoid โ throws exception
const m = new Map()
const a = [1,2,3]
[...m.values()].forEach(console.log)
> [...m.values()].forEach(console.log)
> ^^^
> SyntaxError: Unexpected token ...
// Avoid โ throws exception
const count = 2 // it tries to run 2(), but 2 is not a function
(function doSomething() {
// do something amazing
}())
// put a semicolon before the immediate invoked function, after the const definition, save the return value of the anonymous function to a variable or avoid IIFEs altogether
๐ Read more: "Semi ESLint rule" ๐ Read more: "No unexpected multiline ESLint rule"
TL;DR: Name all functions, including closures and callbacks. Avoid anonymous functions. This is especially useful when profiling a node app. Naming all functions will allow you to easily understand what you're looking at when checking a memory snapshot
Otherwise: Debugging production issues using a core dump (memory snapshot) might become challenging as you notice significant memory consumption from anonymous functions
TL;DR: Use lowerCamelCase when naming constants, variables and functions, UpperCamelCase (capital first letter as well) when naming classes and UPPER_SNAKE_CASE when naming global or static variables. This will help you to easily distinguish between plain variables, functions, classes that require instantiation and variables declared at global module scope. Use descriptive names, but try to keep them short
Otherwise: JavaScript is the only language in the world that allows invoking a constructor ("Class") directly without instantiating it first. Consequently, Classes and function-constructors are differentiated by starting with UpperCamelCase
// for global variables names we use the const/let keyword and UPPER_SNAKE_CASE
let MUTABLE_GLOBAL = "mutable value";
const GLOBAL_CONSTANT = "immutable value";
const CONFIG = {
key: "value",
};
// examples of UPPER_SNAKE_CASE convention in nodejs/javascript ecosystem
// in javascript Math.PI module
const PI = 3.141592653589793;
// https://github.com/nodejs/node/blob/b9f36062d7b5c5039498e98d2f2c180dca2a7065/lib/internal/http2/core.js#L303
// in nodejs http2 module
const HTTP_STATUS_OK = 200;
const HTTP_STATUS_CREATED = 201;
// for class name we use UpperCamelCase
class SomeClassExample {
// for static class properties we use UPPER_SNAKE_CASE
static STATIC_PROPERTY = "value";
}
// for functions names we use lowerCamelCase
function doSomething() {
// for scoped variable names we use the const/let keyword and lowerCamelCase
const someConstExample = "immutable value";
let someMutableExample = "mutable value";
}
TL;DR: Using const
means that once a variable is assigned, it cannot be reassigned. Preferring const
will help you to not be tempted to use the same variable for different uses, and make your code clearer. If a variable needs to be reassigned, in a for loop, for example, use let
to declare it. Another important aspect of let
is that a variable declared using it is only available in the block scope in which it was defined. var
is function scoped, not block-scoped, and shouldn't be used in ES6 now that you have const
and let
at your disposal
Otherwise: Debugging becomes way more cumbersome when following a variable that frequently changes
๐ Read more: JavaScript ES6+: var, let, or const?
TL;DR: Require modules at the beginning of each file, before and outside of any functions. This simple best practice will not only help you easily and quickly tell the dependencies of a file right at the top but also avoids a couple of potential problems
Otherwise: Requires are run synchronously by Node.js. If they are called from within a function, it may block other requests from being handled at a more critical time. Also, if a required module or any of its dependencies throw an error and crash the server, it is best to find out about it as soon as possible, which might not be the case if that module is required from within a function
TL;DR: When developing a module/library, set an explicit root file that exports the public and interesting code. Discourage the client code from importing deep files and becoming familiar with the internal structure. With commonjs (require), this can be done with an index.js file at the folder's root or the package.json.main field. With ESM (import), if a package.json exists on the root, the field "exports" allow specifying the module's root file. If no package.json exists, you may put an index.js file on the root which re-exports all the public functionality
Otherwise: Having an explicit root file acts like a public 'interface' that encapsulates the internal, directs the caller to the public code and facilitates future changes without breaking the contract
// Avoid: client has deep familiarity with the internals
// Client code
const SMSWithMedia = require("./SMSProvider/providers/media/media-provider.js");
// Better: explicitly export the public functions
//index.js, module code
module.exports.SMSWithMedia = require("./SMSProvider/providers/media/media-provider.js");
// Client code
const { SMSWithMedia } = require("./SMSProvider");
TL;DR: Prefer the strict equality operator ===
over the weaker abstract equality operator ==
. ==
will compare two variables after converting them to a common type. There is no type conversion in ===
, and both variables must be of the same type to be equal
Otherwise: Unequal variables might return true when compared with the ==
operator
"" == "0"; // false
0 == ""; // true
0 == "0"; // true
false == "false"; // false
false == "0"; // true
false == undefined; // false
false == null; // false
null == undefined; // true
" \t\r\n " == 0; // true
All statements above will return false if used with ===
TL;DR: Async-await is the simplest way to express an asynchronous flow as it makes asynchronous code look synchronous. Async-await will also result in much more compact code and support for try-catch. This technique now supersedes callbacks and promises in most cases. Using it in your code is probably the best gift one can give to the code reader
Otherwise: Handling async errors in callback style are probably the fastest way to hell - this style forces to check errors all over, deal with awkward code nesting, and makes it difficult to reason about the code flow
๐Read more: Guide to async-await 1.0
TL;DR: Though it's recommended to use async-await and avoid function parameters when dealing with older APIs that accept promises or callbacks - arrow functions make the code structure more compact and keep the lexical context of the root function (i.e. this
)
Otherwise: Longer code (in ES5 functions) is more prone to bugs and cumbersome to read
๐ Read more: Itโs Time to Embrace Arrow Functions
TL;DR: Avoid putting code with effects like network or DB calls outside of functions. Such a code will be executed immediately when another file requires the file. This 'floating' code might get executed when the underlying system is not ready yet. It also comes with a performance penalty even when this module's functions will finally not be used in runtime. Last, mocking these DB/network calls for testing is harder outside of functions. Instead, put this code inside functions that should get called explicitly. If some DB/network code must get executed right when the module loads, consider using the factory or revealing module patterns
Otherwise: A typical web framework sets error handler, environment variables and monitoring. When DB/network calls are made before the web framework is initialized, they won't be monitored or fail due to a lack of configuration data
_We have dedicated guides for testing, see below. The best practices list here is a brief summary of these guides
a. JavaScript testing best practices b. Node.js testing - beyond the basics _
TL;DR: Most projects just don't have any automated testing due to short timetables or often the 'testing project' ran out of control and was abandoned. For that reason, prioritize and start with API testing which is the easiest way to write and provides more coverage than unit testing (you may even craft API tests without code using tools like Postman). Afterwards, should you have more resources and time, continue with advanced test types like unit testing, DB testing, performance testing, etc
Otherwise: You may spend long days on writing unit tests to find out that you got only 20% system coverage
TL;DR: Make the test speak at the requirements level so it's self-explanatory also to QA engineers and developers who are not familiar with the code internals. State in the test name what is being tested (unit under test), under what circumstances, and what is the expected result
Otherwise: A deployment just failed, a test named โAdd productโ failed. Does this tell you what exactly is malfunctioning?
๐ Read More: Include 3 parts in each test name
TL;DR: Structure your tests with 3 well-separated sections: Arrange, Act & Assert (AAA). The first part includes the test setup, then the execution of the unit under test, and finally the assertion phase. Following this structure guarantees that the reader spends no brain CPU on understanding the test plan
Otherwise: Not only you spend long daily hours on understanding the main code, but now also what should have been the simple part of the day (testing) stretches your brain
๐ Read More: Structure tests by the AAA pattern
TL;DR: Use tools that encourage or enforce the same Node.js version across different environments and developers. Tools like nvm, and Volta allow specifying the project's version in a file so each team member can run a single command to conform with the project's version. Optionally, this definition can be replicated to CI and the production runtime (e.g., copy the specified value to .Dockerfile build and to the CI declaration file)
Otherwise: A developer might face or miss an error because she uses a different Node.js version than her teammates. Even worse - the production runtime might be different than the environment where tests were executed
TL;DR: To prevent test coupling and easily reason about the test flow, each test should add and act on its own set of DB rows. Whenever a test needs to pull or assume the existence of some DB data - it must explicitly add that data and avoid mutating any other records
Otherwise: Consider a scenario where deployment is aborted due to failing tests, team is now going to spend precious investigation time that ends in a sad conclusion: the system works well, the tests however interfere with each other and break the build
๐ Read More: Avoid global test fixtures
TL;DR: Different tests must run on different scenarios: quick smoke, IO-less, tests should run when a developer saves or commits a file, full end-to-end tests usually run when a new pull request is submitted, etc. This can be achieved by tagging tests with keywords like #cold #api #sanity so you can grep with your testing harness and invoke the desired subset. For example, this is how you would invoke only the sanity test group with Mocha: mocha --grep 'sanity'
Otherwise: Running all the tests, including tests that perform dozens of DB queries, any time a developer makes a small change can be extremely slow and keeps developers away from running tests
TL;DR: Code coverage tools like Istanbul/NYC are great for 3 reasons: it comes for free (no effort is required to benefit this reports), it helps to identify a decrease in testing coverage, and last but not least it highlights testing mismatches: by looking at colored code coverage reports you may notice, for example, code areas that are never tested like catch clauses (meaning that tests only invoke the happy paths and not how the app behaves on errors). Set it to fail builds if the coverage falls under a certain threshold
Otherwise: There won't be any automated metric telling you when a large portion of your code is not covered by testing
TL;DR: End to end (e2e) testing which includes live data used to be the weakest link of the CI process as it depends on multiple heavy services like DB. Use an environment which is as close to your real production environment as possible like a-continue (Missed -continue here, needs content. Judging by the Otherwise clause, this should mention docker-compose)
Otherwise: Without docker-compose, teams must maintain a testing DB for each testing environment including developers' machines, keep all those DBs in sync so test results won't vary across environments
TL;DR: Using static analysis tools helps by giving objective ways to improve code quality and keeps your code maintainable. You can add static analysis tools to your CI build to fail when it finds code smells. Its main selling points over plain linting are the ability to inspect quality in the context of multiple files (e.g. detect duplications), perform advanced analysis (e.g. code complexity), and follow the history and progress of code issues. Two examples of tools you can use are Sonarqube (2,600+ stars) and Code Climate (1,500+ stars).
Otherwise: With poor code quality, bugs and performance will always be an issue that no shiny new library or state of the art features can fix
TL;DR: Use network mocking tools to simulate responses of external collaborators' services that are approached over the network (e.g., REST, Graph). This is imperative not only to isolate the component under test but mostly to simulate non-happy path flows. Tools like nock (in-process) or Mock-Server allow defining a specific response of external service in a single line of code. Remember to simulate also errors, delays, timeouts, and any other event that is likely to happen in production
Otherwise: Allowing your component to reach real external services instances will likely result in naive tests that mostly cover happy paths. The tests might also be flaky and slow
๐ Read More: Mock external services
TL;DR: When a middleware holds some immense logic that spans many requests, it is worth testing it in isolation without waking up the entire web framework. This can be easily achieved by stubbing and spying on the {req, res, next} objects
Otherwise: A bug in Express middleware === a bug in all or most requests
๐ Read More: Test middlewares in isolation
TL;DR: When testing against the API, it's common and desirable to initialize the web server inside the tests. Let the server randomize the web server port in testing to prevent collisions. If you're using Node.js http server (used by most frameworks), doing so demands nothing but passing a port number zero - this will randomize an available port
Otherwise: Specifying a fixed port will prevent two testing processes from running at the same time. Most of the modern test runners run with multiple processes by default
๐ Read More: Randomize a port for testing
TL;DR: When testing a flow, ensure to cover five potential categories. Any time some action is triggered (e.g., API call), a reaction occurs, a meaningful outcome is produced and calls for testing. There are five possible outcome types for every flow: a response, a visible state change (e.g., DB), an outgoing API call, a new message in a queue, and an observability call (e.g., logging, metric). See a checklist here. Each type of outcome comes with unique challenges and techniques to mitigate those challenges - we have a dedicated guide about this topic: Node.js testing - beyond the basics
Otherwise: Consider a case when testing the addition of a new product to the system. It's common to see tests that assert on a valid response only. What if the product was failed to persist regardless of the positive response? what if when adding a new product demands calling some external service, or putting a message in the queue - shouldn't the test assert these outcomes as well? It's easy to overlook various paths, this is where a checklist comes handy
๐ Read More: Test five outcomes
TL;DR: Monitoring is a game of finding out issues before customers do โ obviously this should be assigned unprecedented importance. The market is overwhelmed with offers thus consider starting with defining the basic metrics you must follow (my suggestions inside), then go over additional fancy features and choose the solution that ticks all boxes. In any case, the 4 layers of observability must be covered: uptime, metrics with focus on user-facing symptoms and Node.js technical metrics like event loop lag, distributed flows measurement with Open Telemetry and logging. Click โRead Moreโ below for an overview of the solutions
Otherwise: Failure === disappointed customers. Simple
TL;DR: Logs can be a dumb warehouse of debug statements or the enabler of a beautiful dashboard that tells the story of your app. Plan your logging platform from day 1: how logs are collected, stored and analyzed to ensure that the desired information (e.g. error rate, following an entire transaction through services and servers, etc) can really be extracted
Otherwise: You end up with a black box that is hard to reason about, then you start re-writing all logging statements to add additional information
๐ Read More: Increase transparency using smart logging
TL;DR: Node is quite bad at doing CPU intensive tasks like gzipping, SSL termination, etc. You should use specialized infrastructure like nginx, HAproxy or cloud vendor services instead
Otherwise: Your poor single thread will stay busy doing infrastructural tasks instead of dealing with your application core and performance will degrade accordingly
๐ Read More: Delegate anything possible (e.g. gzip, SSL) to a reverse proxy
TL;DR: Your code must be identical across all environments, but without a special lockfile npm lets dependencies drift across environments. Ensure to commit your package-lock.json so all the environments will be identical
Otherwise: QA will thoroughly test the code and approve a version that will behave differently in production. Even worse, different servers in the same production cluster might run different code
๐ Read More: Lock dependencies
TL;DR: The process must go on and get restarted upon failures. Modern runtime platforms like Docker-ized platforms (e.g. Kubernetes), and Serverless take care for this automatically. When the app is hosted on a bare metal server, one must take care for a process management tools like systemd. Avoid including a custom process management tool in a modern platform that monitors an app instance (e.g., Kubernetes) - doing so will hide failures from the infrastructure. When the underlying infrastructure is not aware of errors, it can't perform useful mitigation steps like re-placing the instance in a different location
Otherwise: Running dozens of instances without a clear strategy and too many tools together (cluster management, docker, PM2) might lead to DevOps chaos
๐ Read More: Guard process uptime using the right tool
TL;DR: At its basic form, a Node app runs on a single CPU core while all others are left idling. Itโs your duty to replicate the Node process and utilize all CPUs. Most of the modern run-times platform (e.g., Kubernetes) allow replicating instances of the app but they won't verify that all cores are utilized - this is your duty. If the app is hosted on a bare server, it's also your duty to use some process replication solution (e.g. systemd)
Otherwise: Your app will likely utilize only 25% of its available resources(!) or even less. Note that a typical server has 4 CPU cores or more, naive deployment of Node.js utilizes only 1 (even using PaaS services like AWS beanstalk!)
๐ Read More: Utilize all CPU cores
TL;DR: Expose a set of system-related information, like memory usage and REPL, etc in a secured API. Although itโs highly recommended to rely on standard and battle-tested tools, some valuable information and operations are easier done using code
Otherwise: Youโll find that youโre performing many โdiagnostic deploysโ โ shipping code to production only to extract some information for diagnostic purposes
๐ Read More: Create a โmaintenance endpointโ
TL;DR: Consider adding another safety layer to the production stack - APM. While the majority of symptoms and causes can be detected using traditional monitoring techniques, in a distributed system there is more than meets the eye. Application monitoring and performance products (a.k.a. APM) can auto-magically go beyond traditional monitoring and provide additional layer of discovery and developer-experience. For example, some APM products can highlight a transaction that loads too slow on the end-user's side while suggesting the root cause. APMs also provide more context for developers who try to troubleshoot a log error by showing what was the server busy with when the error occurred. To name a few example
Otherwise: You might spend great effort on measuring API performance and downtimes, probably youโll never be aware which is your slowest code parts under real-world scenario and how these affect the UX
๐ Read More: Discover errors and downtime using APM products
TL;DR: Code with the end in mind, plan for production from day 1. This sounds a bit vague so Iโve compiled a few development tips that are closely related to production maintenance (click 'Read More')
Otherwise: A world champion IT/DevOps guy wonโt save a system that is badly written
๐ Read More: Make your code production-ready
TL;DR: Node.js has controversial relationships with memory: the v8 engine has soft limits on memory usage (1.4GB) and there are known paths to leak memory in Nodeโs code โ thus watching Nodeโs process memory is a must. In small apps, you may gauge memory periodically using shell commands but in medium-large apps consider baking your memory watch into a robust monitoring system
Otherwise: Your process memory might leak a hundred megabytes a day like how it happened at Walmart
๐ Read More: Measure and guard the memory usage
TL;DR: Serve frontend content using a specialized infrastructure (nginx, S3, CDN) because Node performance gets hurt when dealing with many static files due to its single-threaded model. One exception to this guideline is when doing server-side rendering
Otherwise: Your single Node thread will be busy streaming hundreds of html/images/angular/react files instead of allocating all its resources for the task it was born for โ serving dynamic content
๐ Read More: Get your frontend assets out of Node