debug vs winston vs pino vs loglevel vs log4js vs bunyan vs @datadog/browser-logs vs js-logger
JavaScript Logging Libraries Comparison
1 Year
debugwinstonpinologlevellog4jsbunyan@datadog/browser-logsjs-loggerSimilar Packages:
What's JavaScript Logging Libraries?

JavaScript logging libraries provide developers with tools to track, record, and manage logs in their applications. These libraries help in debugging, monitoring application performance, and maintaining logs for production environments. Each library offers unique features, performance characteristics, and use cases, making it essential to choose the right one based on specific project requirements and development practices.

Package Weekly Downloads Trend
Github Stars Ranking
Stat Detail
Package
Downloads
Stars
Size
Issues
Publish
License
debug401,179,08311,29942.8 kB782 months agoMIT
winston14,926,47623,797271 kB5138 months agoMIT
pino11,818,49115,763754 kB1412 months agoMIT
loglevel11,245,7482,68886.2 kB1710 months agoMIT
log4js6,163,4385,826160 kB962 years agoApache-2.0
bunyan2,297,9977,201-2945 years agoMIT
@datadog/browser-logs1,283,490354350 kB10214 days agoApache-2.0
js-logger55,011644-115 years agoMIT
Feature Comparison: debug vs winston vs pino vs loglevel vs log4js vs bunyan vs @datadog/browser-logs vs js-logger

Performance

  • debug:

    Extremely lightweight and does not impact performance significantly. Ideal for development environments where performance is less of a concern compared to log visibility.

  • winston:

    Performance is generally good, but can vary based on the number of transports and configurations. Offers a balance between features and performance.

  • pino:

    One of the fastest logging libraries available, designed for high throughput and low overhead. Ideal for applications that require extensive logging without sacrificing performance.

  • loglevel:

    Minimal impact on performance due to its simple API and lightweight nature. Suitable for applications where logging is not a critical aspect.

  • log4js:

    Performance can vary based on configuration and the number of appenders used. Offers flexibility but may introduce overhead if not configured properly.

  • bunyan:

    Designed for high performance with a focus on structured logging. Outputs logs in JSON format, making it easy to parse and analyze, which can be beneficial for performance-sensitive applications.

  • @datadog/browser-logs:

    Optimized for sending logs to Datadog, but may introduce some latency due to network calls. Best used in conjunction with Datadog's monitoring tools for real-time insights.

  • js-logger:

    Lightweight and straightforward, making it easy to integrate without impacting performance. Suitable for small to medium-sized applications.

Log Format

  • debug:

    Logs are typically unstructured and printed to the console, which may not be ideal for production environments but is sufficient for development purposes.

  • winston:

    Supports multiple log formats, including JSON and plain text, allowing for flexibility in how logs are recorded and displayed.

  • pino:

    Logs are output in a JSON format that is highly efficient for parsing and analysis, making it suitable for production environments where structured logs are preferred.

  • loglevel:

    Logs are output as simple strings with log levels, making it easy to understand but not structured for advanced analysis.

  • log4js:

    Supports various log formats and layouts, allowing for customization based on project needs. Can output logs in both plain text and JSON formats.

  • bunyan:

    Outputs logs in JSON format, which is easily parsable and ideal for structured logging. This format is beneficial for log analysis and integration with log management tools.

  • @datadog/browser-logs:

    Logs are sent in a structured format compatible with Datadog's logging service, making it easy to analyze and visualize logs in the Datadog dashboard.

  • js-logger:

    Logs are output as simple strings, making it easy to read but lacking structure for advanced log analysis.

Configuration Flexibility

  • debug:

    Configuration is minimal, primarily controlled through environment variables, making it easy to use but less flexible for advanced setups.

  • winston:

    Extremely flexible with support for multiple transports and custom formatting, making it suitable for a wide range of logging scenarios.

  • pino:

    Highly configurable with options for serializers and transport, allowing for extensive customization while maintaining performance.

  • loglevel:

    Simple configuration with predefined log levels, making it easy to use but limited in customization options.

  • log4js:

    Extensively configurable with support for multiple appenders and layouts, making it suitable for complex logging requirements.

  • bunyan:

    Highly configurable with support for multiple streams and serializers, allowing developers to customize log output based on their needs.

  • @datadog/browser-logs:

    Configuration is primarily focused on integration with Datadog, with limited customization options for log output.

  • js-logger:

    Offers basic configuration options for log levels and output, suitable for small projects but may lack advanced features for larger applications.

Use Case Suitability

  • debug:

    Perfect for development environments where dynamic control over log visibility is needed without affecting production performance.

  • winston:

    Suitable for applications that need a comprehensive logging solution with advanced features and flexibility.

  • pino:

    Ideal for high-performance applications that require extensive logging without compromising speed.

  • loglevel:

    Best for projects that need a simple logging solution without the overhead of more complex libraries.

  • log4js:

    Suitable for applications that require detailed logging configurations and flexibility in log output formats.

  • bunyan:

    Ideal for server-side applications that need structured logging and integration with various log management tools.

  • @datadog/browser-logs:

    Best suited for applications that are already using Datadog for monitoring and require real-time log management.

  • js-logger:

    Great for small to medium-sized applications that require straightforward logging without complex configurations.

Community and Ecosystem

  • debug:

    Has a large user base and is widely adopted in the Node.js community, making it easy to find support and resources.

  • winston:

    One of the most popular logging libraries in the Node.js ecosystem, with a large community and extensive resources available.

  • pino:

    Rapidly growing community with a focus on performance, providing extensive documentation and support for users.

  • loglevel:

    Widely used in smaller projects, with a decent community for support and resources.

  • log4js:

    Established library with a solid community, providing good documentation and support for various configurations.

  • bunyan:

    Well-supported with a strong community, and integrates well with various logging tools and services.

  • @datadog/browser-logs:

    Part of the Datadog ecosystem, with strong community support for users integrating with Datadog services.

  • js-logger:

    Smaller community but sufficient documentation for basic use cases. Less common in larger projects.

How to Choose: debug vs winston vs pino vs loglevel vs log4js vs bunyan vs @datadog/browser-logs vs js-logger
  • debug:

    Use Debug for lightweight debugging during development. It allows you to enable/disable logs dynamically via environment variables, making it perfect for applications where you want to control log verbosity without affecting production performance.

  • winston:

    Choose Winston for a comprehensive logging solution that supports multiple transports and formats. It's suitable for applications that require advanced logging features, including log rotation, custom formatting, and integration with various storage backends.

  • pino:

    Use Pino if performance is a priority, as it is designed to be one of the fastest logging libraries available. It's perfect for high-throughput applications where logging speed and efficiency are critical.

  • loglevel:

    Select Loglevel for a minimalistic logging library that provides a simple API and different log levels. It's ideal for projects that require basic logging functionality without the overhead of more complex libraries.

  • log4js:

    Choose Log4js if you are looking for a versatile logging framework that supports multiple appenders and layouts. It's suitable for applications that need detailed logging configurations and flexibility in log output formats.

  • bunyan:

    Select Bunyan if you prefer a structured logging format (JSON) and want to integrate with various transports. It's suitable for server-side applications that need efficient logging with a focus on performance and log parsing.

  • @datadog/browser-logs:

    Choose this package if you need to integrate logging with Datadog for monitoring and analytics. It's ideal for applications that require real-time log management and visualization in a centralized dashboard.

  • js-logger:

    Opt for js-logger if you need a simple, lightweight logging solution that works in both browser and Node.js environments. It's great for projects that require minimal setup and straightforward logging capabilities.

README for debug

debug

OpenCollective OpenCollective

A tiny JavaScript debugging utility modelled after Node.js core's debugging technique. Works in Node.js and web browsers.

Installation

$ npm install debug

Usage

debug exposes a function; simply pass this function the name of your module, and it will return a decorated version of console.error for you to pass debug statements to. This will allow you to toggle the debug output for different parts of your module as well as the module as a whole.

Example app.js:

var debug = require('debug')('http')
  , http = require('http')
  , name = 'My App';

// fake app

debug('booting %o', name);

http.createServer(function(req, res){
  debug(req.method + ' ' + req.url);
  res.end('hello\n');
}).listen(3000, function(){
  debug('listening');
});

// fake worker of some kind

require('./worker');

Example worker.js:

var a = require('debug')('worker:a')
  , b = require('debug')('worker:b');

function work() {
  a('doing lots of uninteresting work');
  setTimeout(work, Math.random() * 1000);
}

work();

function workb() {
  b('doing some work');
  setTimeout(workb, Math.random() * 2000);
}

workb();

The DEBUG environment variable is then used to enable these based on space or comma-delimited names.

Here are some examples:

screen shot 2017-08-08 at 12 53 04 pm screen shot 2017-08-08 at 12 53 38 pm screen shot 2017-08-08 at 12 53 25 pm

Windows command prompt notes

CMD

On Windows the environment variable is set using the set command.

set DEBUG=*,-not_this

Example:

set DEBUG=* & node app.js
PowerShell (VS Code default)

PowerShell uses different syntax to set environment variables.

$env:DEBUG = "*,-not_this"

Example:

$env:DEBUG='app';node app.js

Then, run the program to be debugged as usual.

npm script example:

  "windowsDebug": "@powershell -Command $env:DEBUG='*';node app.js",

Namespace Colors

Every debug instance has a color generated for it based on its namespace name. This helps when visually parsing the debug output to identify which debug instance a debug line belongs to.

Node.js

In Node.js, colors are enabled when stderr is a TTY. You also should install the supports-color module alongside debug, otherwise debug will only use a small handful of basic colors.

Web Browser

Colors are also enabled on "Web Inspectors" that understand the %c formatting option. These are WebKit web inspectors, Firefox (since version 31) and the Firebug plugin for Firefox (any version).

Millisecond diff

When actively developing an application it can be useful to see when the time spent between one debug() call and the next. Suppose for example you invoke debug() before requesting a resource, and after as well, the "+NNNms" will show you how much time was spent between calls.

When stdout is not a TTY, Date#toISOString() is used, making it more useful for logging the debug information as shown below:

Conventions

If you're using this in one or more of your libraries, you should use the name of your library so that developers may toggle debugging as desired without guessing names. If you have more than one debuggers you should prefix them with your library name and use ":" to separate features. For example "bodyParser" from Connect would then be "connect:bodyParser". If you append a "*" to the end of your name, it will always be enabled regardless of the setting of the DEBUG environment variable. You can then use it for normal output as well as debug output.

Wildcards

The * character may be used as a wildcard. Suppose for example your library has debuggers named "connect:bodyParser", "connect:compress", "connect:session", instead of listing all three with DEBUG=connect:bodyParser,connect:compress,connect:session, you may simply do DEBUG=connect:*, or to run everything using this module simply use DEBUG=*.

You can also exclude specific debuggers by prefixing them with a "-" character. For example, DEBUG=*,-connect:* would include all debuggers except those starting with "connect:".

Environment Variables

When running through Node.js, you can set a few environment variables that will change the behavior of the debug logging:

| Name | Purpose | |-----------|-------------------------------------------------| | DEBUG | Enables/disables specific debugging namespaces. | | DEBUG_HIDE_DATE | Hide date from debug output (non-TTY). | | DEBUG_COLORS| Whether or not to use colors in the debug output. | | DEBUG_DEPTH | Object inspection depth. | | DEBUG_SHOW_HIDDEN | Shows hidden properties on inspected objects. |

Note: The environment variables beginning with DEBUG_ end up being converted into an Options object that gets used with %o/%O formatters. See the Node.js documentation for util.inspect() for the complete list.

Formatters

Debug uses printf-style formatting. Below are the officially supported formatters:

| Formatter | Representation | |-----------|----------------| | %O | Pretty-print an Object on multiple lines. | | %o | Pretty-print an Object all on a single line. | | %s | String. | | %d | Number (both integer and float). | | %j | JSON. Replaced with the string '[Circular]' if the argument contains circular references. | | %% | Single percent sign ('%'). This does not consume an argument. |

Custom formatters

You can add custom formatters by extending the debug.formatters object. For example, if you wanted to add support for rendering a Buffer as hex with %h, you could do something like:

const createDebug = require('debug')
createDebug.formatters.h = (v) => {
  return v.toString('hex')
}

// …elsewhere
const debug = createDebug('foo')
debug('this is hex: %h', new Buffer('hello world'))
//   foo this is hex: 68656c6c6f20776f726c6421 +0ms

Browser Support

You can build a browser-ready script using browserify, or just use the browserify-as-a-service build, if you don't want to build it yourself.

Debug's enable state is currently persisted by localStorage. Consider the situation shown below where you have worker:a and worker:b, and wish to debug both. You can enable this using localStorage.debug:

localStorage.debug = 'worker:*'

And then refresh the page.

a = debug('worker:a');
b = debug('worker:b');

setInterval(function(){
  a('doing some work');
}, 1000);

setInterval(function(){
  b('doing some work');
}, 1200);

In Chromium-based web browsers (e.g. Brave, Chrome, and Electron), the JavaScript console will—by default—only show messages logged by debug if the "Verbose" log level is enabled.

Output streams

By default debug will log to stderr, however this can be configured per-namespace by overriding the log method:

Example stdout.js:

var debug = require('debug');
var error = debug('app:error');

// by default stderr is used
error('goes to stderr!');

var log = debug('app:log');
// set this namespace to log via console.log
log.log = console.log.bind(console); // don't forget to bind to console!
log('goes to stdout');
error('still goes to stderr!');

// set all output to go via console.info
// overrides all per-namespace log settings
debug.log = console.info.bind(console);
error('now goes to stdout via console.info');
log('still goes to stdout, but via console.info now');

Extend

You can simply extend debugger

const log = require('debug')('auth');

//creates new debug instance with extended namespace
const logSign = log.extend('sign');
const logLogin = log.extend('login');

log('hello'); // auth hello
logSign('hello'); //auth:sign hello
logLogin('hello'); //auth:login hello

Set dynamically

You can also enable debug dynamically by calling the enable() method :

let debug = require('debug');

console.log(1, debug.enabled('test'));

debug.enable('test');
console.log(2, debug.enabled('test'));

debug.disable();
console.log(3, debug.enabled('test'));

print :

1 false
2 true
3 false

Usage :
enable(namespaces)
namespaces can include modes separated by a colon and wildcards.

Note that calling enable() completely overrides previously set DEBUG variable :

$ DEBUG=foo node -e 'var dbg = require("debug"); dbg.enable("bar"); console.log(dbg.enabled("foo"))'
=> false

disable()

Will disable all namespaces. The functions returns the namespaces currently enabled (and skipped). This can be useful if you want to disable debugging temporarily without knowing what was enabled to begin with.

For example:

let debug = require('debug');
debug.enable('foo:*,-foo:bar');
let namespaces = debug.disable();
debug.enable(namespaces);

Note: There is no guarantee that the string will be identical to the initial enable string, but semantically they will be identical.

Checking whether a debug target is enabled

After you've created a debug instance, you can determine whether or not it is enabled by checking the enabled property:

const debug = require('debug')('http');

if (debug.enabled) {
  // do stuff...
}

You can also manually toggle this property to force the debug instance to be enabled or disabled.

Usage in child processes

Due to the way debug detects if the output is a TTY or not, colors are not shown in child processes when stderr is piped. A solution is to pass the DEBUG_COLORS=1 environment variable to the child process.
For example:

worker = fork(WORKER_WRAP_PATH, [workerPath], {
  stdio: [
    /* stdin: */ 0,
    /* stdout: */ 'pipe',
    /* stderr: */ 'pipe',
    'ipc',
  ],
  env: Object.assign({}, process.env, {
    DEBUG_COLORS: 1 // without this settings, colors won't be shown
  }),
});

worker.stderr.pipe(process.stderr, { end: false });

Authors

  • TJ Holowaychuk
  • Nathan Rajlich
  • Andrew Rhyne
  • Josh Junon

Backers

Support us with a monthly donation and help us continue our activities. [Become a backer]

Sponsors

Become a sponsor and get your logo on our README on Github with a link to your site. [Become a sponsor]

License

(The MIT License)

Copyright (c) 2014-2017 TJ Holowaychuk <tj@vision-media.ca> Copyright (c) 2018-2021 Josh Junon

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the 'Software'), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED 'AS IS', WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.