Tree Shaking
Learn how to reduce Sentry bundle size by tree shaking unused code.
The Sentry SDK supports tree shaking in various ways. To fully utilize the tree shaking capabilities of modern bundlers like webpack or Rollup, some additional configurations must be applied. If you want to minimize the bundle size of the Sentry SDK, we recommend reading through this page and applying the tree shaking configurations as shown.
The Sentry SDK ships with code that is not strictly required for it to collect your errors. This includes code to debug your Sentry configuration or code to enable tracing, for example. While debug code can be very useful in development environments, it's not typically necessary to include it in your production bundles where it takes up valuable space. The JavaScript SDK includes special flags in its CommonJS and ESM distributions, which can be used to facilitate tree shaking (removal) of this kind of code during the build process.
Anything that you don't import and use will automatically be tree shaken away when using any modern bundler like Webpack, Rollup, Vite, or similar. This means that optional integrations like Replay, BrowserTracing, BrowserProfiling, and any unused utility methods won't be included in your bundle unless you import and use them yourself. The rest of this page relates to ways to tree shake internal SDK code, which isn't strictly required unless you use certain features.
To tree shake Sentry debug code in Next.js projects, use webpack's DefinePlugin in your Next.js configuration like in the example below:
next.config.(js|mjs)
const nextConfig = {
webpack: (config, { webpack }) => {
config.plugins.push(
new webpack.DefinePlugin({
__SENTRY_DEBUG__: false,
__SENTRY_TRACING__: false,
__RRWEB_EXCLUDE_IFRAME__: true,
__RRWEB_EXCLUDE_SHADOW_DOM__: true,
__SENTRY_EXCLUDE_REPLAY_WORKER__: true,
}),
);
// return the modified config
return config;
},
};
For more information on custom webpack configurations in Next.js, see Custom Webpack Config in the Next.js docs.
If you want to tree shake optional code, remove the code from your build output by replacing various flags in the Sentry SDK.
The following flags are available:
__SENTRY_DEBUG__
Replacing this flag with false
will tree shake any SDK code that's related to debug logging.
__SENTRY_TRACING__
Replacing this flag with false
will tree shake any SDK code that's related to tracing.
__SENTRY_TRACING__
must not be replaced with false
when you're using any tracing-related SDK features (for example,Sentry.startTransaction()
). This flag is intended to be used in combination with packages like @sentry/next
or @sentry/sveltekit
, which automatically include the tracing functionality.
__RRWEB_EXCLUDE_IFRAME__
Replacing this flag with true
will tree shake any SDK code related to capturing iframe content with Session Replay. It's only relevant when using Session Replay. Enable this flag if you don't want to record any iframes.
__RRWEB_EXCLUDE_SHADOW_DOM__
Replacing this flag with true
will tree shake any SDK code related to capturing shadow dom elements with Session Replay. It's only relevant when using Session Replay. Enable this flag if you don't want to record any shadow dom elements.
__SENTRY_EXCLUDE_REPLAY_WORKER__
Replacing this flag with true
will tree shake any SDK code that's related to the included compression web worker for Session Replay. It's only relevant when using Session Replay. Enable this flag if you want to host a compression worker yourself. See Using a Custom Compression Worker for details. We don't recommend enabling this flag unless you provide a custom worker URL.
By default, the Sentry SDK sets up a list of default integrations that extend your SDK functionality. You can also add additional or custom integrations to your SDK configuration. If you don't want to include default integrations in your config, you can disable them and add your custom array of integrations. However, if you also want to tree shake the unused default integrations, you can do so by creating a Client
yourself. By doing this, you essentially bypass Sentry.init
which normally creates a Client
for you.
The following example shows how to create and bind a Client
which enables tree shaking of unused default integrations:
main.js
import {
BrowserClient,
breadcrumbsIntegration,
dedupeIntegration,
defaultStackParser,
getCurrentScope,
globalHandlersIntegration,
makeFetchTransport,
linkedErrorsIntegration,
} from "@sentry/browser";
const client = new BrowserClient({
// all options you normally pass to Sentry.init
dsn: "your DSN",
// ...
transport: makeFetchTransport,
stackParser: defaultStackParser,
// Only the integrations listed here will be used
integrations: [
breadcrumbsIntegration(),
globalHandlersIntegration(),
linkedErrorsIntegration(),
dedupeIntegration(),
],
});
getCurrentScope().setClient(client);
client.init();
In contrast to Sentry.init
, the Client
constructor expects options of type ClientOptions
instead of Options
. This means that the ClientOptions.integrations
property is the final array of all integrations that will be used.
Our documentation is open source and available on GitHub. Your contributions are welcome, whether fixing a typo (drat!) or suggesting an update ("yeah, this would be better").