Vite's default browser support baseline is Native ESM, native ESM dynamic import, and import.meta
. This plugin provides support for legacy browsers that do not support those features when building for production.
By default, this plugin will:
-
Generate a corresponding legacy chunk for every chunk in the final bundle, transformed with @babel/preset-env and emitted as SystemJS modules (code splitting is still supported!).
-
Generate a polyfill chunk including SystemJS runtime, and any necessary polyfills determined by specified browser targets and actual usage in the bundle.
-
Inject
<script nomodule>
tags into generated HTML to conditionally load the polyfills and legacy bundle only in browsers without widely-available features support. -
Inject the
import.meta.env.LEGACY
env variable, which will only betrue
in the legacy production build, andfalse
in all other cases.
// vite.config.js
import legacy from '@vitejs/plugin-legacy'
export default {
plugins: [
legacy({
targets: ['defaults', 'not IE 11'],
}),
],
}
Terser must be installed because plugin-legacy uses Terser for minification.
npm add -D terser
-
Type:
string | string[] | { [key: string]: string }
-
Default:
'last 2 versions and not dead, > 0.3%, Firefox ESR'
If explicitly set, it's passed on to
@babel/preset-env
.The query is also Browserslist compatible. See Browserslist Best Practices for more details.
If it's not set, plugin-legacy will load the browserslist config sources and then fallback to the default value.
-
Type:
boolean | string[]
-
Default:
true
By default, a polyfills chunk is generated based on the target browser ranges and actual usage in the final bundle (detected via
@babel/preset-env
'suseBuiltIns: 'usage'
).Set to a list of strings to explicitly control which polyfills to include. See Polyfill Specifiers for details.
Set to
false
to avoid generating polyfills and handle it yourself (will still generate legacy chunks with syntax transformations).
-
Type:
string[]
Add custom imports to the legacy polyfills chunk. Since the usage-based polyfill detection only covers ES language features, it may be necessary to manually specify additional DOM API polyfills using this option.
Note: if additional polyfills are needed for both the modern and legacy chunks, they can simply be imported in the application source code.
-
Type:
boolean | string[]
-
Default:
false
Defaults to
false
. Enabling this option will generate a separate polyfills chunk for the modern build (targeting browsers that support widely-available features).Set to a list of strings to explicitly control which polyfills to include. See Polyfill Specifiers for details.
Note it is not recommended to use the
true
value (which uses auto-detection) becausecore-js@3
is very aggressive in polyfill inclusions due to all the bleeding edge features it supports. Even when targeting native ESM support, it injects 15kb of polyfills!If you don't have hard reliance on bleeding edge runtime features, it is not that hard to avoid having to use polyfills in the modern build altogether. Alternatively, consider using an on-demand service like Polyfill.io to only inject necessary polyfills based on actual browser user-agents (most modern browsers will need nothing!).
-
Type:
boolean
-
Default:
true
Set to
false
to disable legacy chunks. This is only useful if you are usingmodernPolyfills
, which essentially allows you to use this plugin for injecting polyfills to the modern build only:import legacy from '@vitejs/plugin-legacy' export default { plugins: [ legacy({ modernPolyfills: [ /* ... */ ], renderLegacyChunks: false, }), ], }
-
Type:
boolean
-
Default:
false
Defaults to
false
. Enabling this option will excludesystemjs/dist/s.min.js
inside polyfills-legacy chunk.
-
Type:
boolean
-
Default:
true
Set to
false
to only output the legacy bundles that support all target browsers.
The legacy plugin offers a way to use widely-available features natively in the modern build, while falling back to the legacy build in browsers with native ESM but without those features supported (e.g. Legacy Edge). This feature works by injecting a runtime check and loading the legacy bundle with SystemJs runtime if needed. There are the following drawbacks:
- Modern bundle is downloaded in all ESM browsers
- Modern bundle throws
SyntaxError
in browsers without those features support
The following syntax are considered as widely-available:
- dynamic import
import.meta
- async generator
Polyfill specifier strings for polyfills
and modernPolyfills
can be either of the following:
-
Any
core-js
3 sub import paths - e.g.es/map
will importcore-js/es/map
-
Any individual
core-js
3 modules - e.g.es.array.iterator
will importcore-js/modules/es.array.iterator.js
Example
import legacy from '@vitejs/plugin-legacy'
export default {
plugins: [
legacy({
polyfills: ['es.promise.finally', 'es/map', 'es/set'],
modernPolyfills: ['es.promise.finally'],
}),
],
}
The legacy plugin requires inline scripts for Safari 10.1 nomodule
fix, SystemJS initialization, and dynamic import fallback. If you have a strict CSP policy requirement, you will need to add the corresponding hashes to your script-src
list.
The hash values (without the sha256-
prefix) can be retrieved via:
import { cspHashes } from '@vitejs/plugin-legacy'
The current values are:
sha256-MS6/3FCg4WjP9gwgaBGwLpRCY6fZBgwmhVCdrPrNf3E=
sha256-tQjf8gvb2ROOMapIxFvFAYBeUJ0v1HCbOcSmDNXGtDo=
sha256-VA8O2hAdooB288EpSTrGLl7z3QikbWU9wwoebO/QaYk=
sha256-+5XkZFazzJo8n0iOP4ti/cLCMUudTf//Mzkb7xNPXIc=
Note that these values could change between minor versions. Thus, we recommend generating the CSP header from the exported cspHashes
variable. If you copy the values manually, then you should pin the minor version using ~
.
When using the regenerator-runtime
polyfill, it will attempt to use the globalThis
object to register itself. If globalThis
is not available (it is fairly new and not widely supported, including IE 11), it attempts to perform dynamic Function(...)
call which violates the CSP. To avoid dynamic eval
in the absence of globalThis
consider adding core-js/proposals/global-this
to additionalLegacyPolyfills
to define it.