An extremely fast bundler for the web
Support TypeScript experimental decorators on abstract class fields (#3684)
With this release, you can now use TypeScript experimental decorators on abstract class fields. This was silently compiled incorrectly in esbuild 0.19.7 and below, and was an error from esbuild 0.19.8 to esbuild 0.20.1. Code such as the following should now work correctly:
// Original code
const log = (x: any, y: string) => console.log(y)
abstract class Foo { @log abstract foo: string }
new class extends Foo { foo = '' }
// Old output (with --loader=ts --tsconfig-raw={\"compilerOptions\":{\"experimentalDecorators\":true}})
const log = (x, y) => console.log(y);
class Foo {
}
new class extends Foo {
foo = "";
}();
// New output (with --loader=ts --tsconfig-raw={\"compilerOptions\":{\"experimentalDecorators\":true}})
const log = (x, y) => console.log(y);
class Foo {
}
__decorateClass([
log
], Foo.prototype, "foo", 2);
new class extends Foo {
foo = "";
}();
JSON loader now preserves __proto__ properties (#3700)
Copying JSON source code into a JavaScript file will change its meaning if a JSON object contains the __proto__ key. A literal __proto__ property in a JavaScript object literal sets the prototype of the object instead of adding a property named __proto__, while a literal __proto__ property in a JSON object literal just adds a property named __proto__. With this release, esbuild will now work around this problem by converting JSON to JavaScript with a computed property key in this case:
// Original code
import data from 'data:application/json,{"__proto__":{"fail":true}}'
if (Object.getPrototypeOf(data)?.fail) throw 'fail'
// Old output (with --bundle)
(() => {
// <data:application/json,{"__proto__":{"fail":true}}>
var json_proto_fail_true_default = { __proto__: { fail: true } };
// entry.js
if (Object.getPrototypeOf(json_proto_fail_true_default)?.fail)
throw "fail";
})();
// New output (with --bundle)
(() => {
// <data:application/json,{"__proto__":{"fail":true}}>
var json_proto_fail_true_default = { ["__proto__"]: { fail: true } };
// example.mjs
if (Object.getPrototypeOf(json_proto_fail_true_default)?.fail)
throw "fail";
})();
Improve dead code removal of switch statements (#3659)
With this release, esbuild will now remove switch statements in branches when minifying if they are known to never be evaluated:
// Original code
if (true) foo(); else switch (bar) { case 1: baz(); break }
// Old output (with --minify)
if(1)foo();else switch(bar){case 1:}
// New output (with --minify)
foo();
Empty enums should behave like an object literal (#3657)
TypeScript allows you to create an empty enum and add properties to it at run time. While people usually use an empty object literal for this instead of a TypeScript enum, esbuild's enum transform didn't anticipate this use case and generated undefined instead of {} for an empty enum. With this release, you can now use an empty enum to generate an empty object literal.
// Original code
enum Foo {}
// Old output (with --loader=ts)
var Foo = /* @__PURE__ */ ((Foo2) => {
})(Foo || {});
// New output (with --loader=ts)
var Foo = /* @__PURE__ */ ((Foo2) => {
return Foo2;
})(Foo || {});
Handle Yarn Plug'n'Play edge case with tsconfig.json (#3698)
Previously a tsconfig.json file that extends another file in a package with an exports map failed to work when Yarn's Plug'n'Play resolution was active. This edge case should work now starting with this release.
Work around issues with Deno 1.31+ (#3682)
Version 0.20.0 of esbuild changed how the esbuild child process is run in esbuild's API for Deno. Previously it used Deno.run but that API is being removed in favor of Deno.Command. As part of this change, esbuild is now calling the new unref function on esbuild's long-lived child process, which is supposed to allow Deno to exit when your code has finished running even though the child process is still around (previously you had to explicitly call esbuild's stop() function to terminate the child process for Deno to be able to exit).
However, this introduced a problem for Deno's testing API which now fails some tests that use esbuild with error: Promise resolution is still pending but the event loop has already resolved. It's unclear to me why this is happening. The call to unref was recommended by someone on the Deno core team, and calling Node's equivalent unref API has been working fine for esbuild in Node for a long time. It could be that I'm using it incorrectly, or that there's some reference counting and/or garbage collection bug in Deno's internals, or that Deno's unref just works differently than Node's unref. In any case, it's not good for Deno tests that use esbuild to be failing.
In this release, I am removing the call to unref to fix this issue. This means that you will now have to call esbuild's stop() function to allow Deno to exit, just like you did before esbuild version 0.20.0 when this regression was introduced.
Note: This regression wasn't caught earlier because Deno doesn't seem to fail tests that have outstanding setTimeout calls, which esbuild's test harness was using to enforce a maximum test runtime. Adding a setTimeout was allowing esbuild's Deno tests to succeed. So this regression doesn't necessarily apply to all people using tests in Deno.
Fix a bug with the CSS nesting transform (#3648)
This release fixes a bug with the CSS nesting transform for older browsers where the generated CSS could be incorrect if a selector list contained a pseudo element followed by another selector. The bug was caused by incorrectly mutating the parent rule's selector list when filtering out pseudo elements for the child rules:
/* Original code */
.foo {
&:after,
& .bar {
color: red;
}
}
/* Old output (with --supported:nesting=false) */
.foo .bar,
.foo .bar {
color: red;
}
/* New output (with --supported:nesting=false) */
.foo:after,
.foo .bar {
color: red;
}
Constant folding for JavaScript inequality operators (#3645)
This release introduces constant folding for the < > <= >= operators. The minifier will now replace these operators with true or false when both sides are compile-time numeric or string constants:
// Original code
console.log(1 < 2, '🍕' > '🧀')
// Old output (with --minify)
console.log(1<2,"🍕">"🧀");
// New output (with --minify)
console.log(!0,!1);
Better handling of __proto__ edge cases (#3651)
JavaScript object literal syntax contains a special case where a non-computed property with a key of __proto__ sets the prototype of the object. This does not apply to computed properties or to properties that use the shorthand property syntax introduced in ES6. Previously esbuild didn't correctly preserve the "sets the prototype" status of properties inside an object literal, meaning a property that sets the prototype could accidentally be transformed into one that doesn't and vice versa. This has now been fixed:
// Original code
function foo(__proto__) {
return { __proto__: __proto__ } // Note: sets the prototype
}
function bar(__proto__, proto) {
{
let __proto__ = proto
return { __proto__ } // Note: doesn't set the prototype
}
}
// Old output
function foo(__proto__) {
return { __proto__ }; // Note: no longer sets the prototype (WRONG)
}
function bar(__proto__, proto) {
{
let __proto__2 = proto;
return { __proto__: __proto__2 }; // Note: now sets the prototype (WRONG)
}
}
// New output
function foo(__proto__) {
return { __proto__: __proto__ }; // Note: sets the prototype (correct)
}
function bar(__proto__, proto) {
{
let __proto__2 = proto;
return { ["__proto__"]: __proto__2 }; // Note: doesn't set the prototype (correct)
}
}
Fix cross-platform non-determinism with CSS color space transformations (#3650)
The Go compiler takes advantage of "fused multiply and add" (FMA) instructions on certain processors which do the operation x*y + z without intermediate rounding. This causes esbuild's CSS color space math to differ on different processors (currently ppc64le and s390x), which breaks esbuild's guarantee of deterministic output. To avoid this, esbuild's color space math now inserts a float64() cast around every single math operation. This tells the Go compiler not to use the FMA optimization.
Fix a crash when resolving a path from a directory that doesn't exist (#3634)
This release fixes a regression where esbuild could crash when resolving an absolute path if the source directory for the path resolution operation doesn't exist. While this situation doesn't normally come up, it could come up when running esbuild concurrently with another operation that mutates the file system as esbuild is doing a build (such as using git to switch branches). The underlying problem was a regression that was introduced in version 0.18.0.
This release deliberately contains backwards-incompatible changes. To avoid automatically picking up releases like this, you should either be pinning the exact version of esbuild in your package.json file (recommended) or be using a version range syntax that only accepts patch upgrades such as ^0.19.0 or ~0.19.0. See npm's documentation about semver for more information.
This time there is only one breaking change, and it only matters for people using Deno. Deno tests that use esbuild will now fail unless you make the change described below.
Work around API deprecations in Deno 1.40.x (#3609, #3611)
Deno 1.40.0 was just released and introduced run-time warnings about certain APIs that esbuild uses. With this release, esbuild will work around these run-time warnings by using newer APIs if they are present and falling back to the original APIs otherwise. This should avoid the warnings without breaking compatibility with older versions of Deno.
Unfortunately, doing this introduces a breaking change. The newer child process APIs lack a way to synchronously terminate esbuild's child process, so calling esbuild.stop() from within a Deno test is no longer sufficient to prevent Deno from failing a test that uses esbuild's API (Deno fails tests that create a child process without killing it before the test ends). To work around this, esbuild's stop() function has been changed to return a promise, and you now have to change esbuild.stop() to await esbuild.stop() in all of your Deno tests.
Reorder implicit file extensions within node_modules (#3341, #3608)
In version 0.18.0, esbuild changed the behavior of implicit file extensions within node_modules directories (i.e. in published packages) to prefer .js over .ts even when the --resolve-extensions= order prefers .ts over .js (which it does by default). However, doing that also accidentally made esbuild prefer .css over .ts, which caused problems for people that published packages containing both TypeScript and CSS in files with the same name.
With this release, esbuild will reorder TypeScript file extensions immediately after the last JavaScript file extensions in the implicit file extension order instead of putting them at the end of the order. Specifically the default implicit file extension order is .tsx,.ts,.jsx,.js,.css,.json which used to become .jsx,.js,.css,.json,.tsx,.ts in node_modules directories. With this release it will now become .jsx,.js,.tsx,.ts,.css,.json instead.
Why even rewrite the implicit file extension order at all? One reason is because the .js file is more likely to behave correctly than the .ts file. The behavior of the .ts file may depend on tsconfig.json and the tsconfig.json file may not even be published, or may use extends to refer to a base tsconfig.json file that wasn't published. People can get into this situation when they forget to add all .ts files to their .npmignore file before publishing to npm. Picking .js over .ts helps make it more likely that resulting bundle will behave correctly.
The "preserve" JSX mode now preserves JSX text verbatim (#3605)
The JSX specification deliberately doesn't specify how JSX text is supposed to be interpreted and there is no canonical way to interpret JSX text. Two most popular interpretations are Babel and TypeScript. Yes they are different (esbuild deliberately follows TypeScript by the way).
Previously esbuild normalized text to the TypeScript interpretation when the "preserve" JSX mode is active. However, "preserve" should arguably reproduce the original JSX text verbatim so that whatever JSX transform runs after esbuild is free to interpret it however it wants. So with this release, esbuild will now pass JSX text through unmodified:
// Original code
let el =
<a href={'/'} title=''"'> some text
{foo}
more text </a>
// Old output (with --loader=jsx --jsx=preserve)
let el = <a href="/" title={`'"`}>
{" some text"}
{foo}
{"more text "}
</a>;
// New output (with --loader=jsx --jsx=preserve)
let el = <a href={"/"} title=''"'> some text
{foo}
more text </a>;
Allow JSX elements as JSX attribute values
JSX has an obscure feature where you can use JSX elements in attribute position without surrounding them with {...}. It looks like this:
let el = <div data-ab=<><a/><b/></>/>;
I think I originally didn't implement it even though it's part of the JSX specification because it previously didn't work in TypeScript (and potentially also in Babel?). However, support for it was silently added in TypeScript 4.8 without me noticing and Babel has also since fixed their bugs regarding this feature. So I'm adding it to esbuild too now that I know it's widely supported.
Keep in mind that there is some ongoing discussion about removing this feature from JSX. I agree that the syntax seems out of place (it does away with the elegance of "JSX is basically just XML with {...} escapes" for something arguably harder to read, which doesn't seem like a good trade-off), but it's in the specification and TypeScript and Babel both implement it so I'm going to have esbuild implement it too. However, I reserve the right to remove it from esbuild if it's ever removed from the specification in the future. So use it with caution.
Fix a bug with TypeScript type parsing (#3574)
This release fixes a bug with esbuild's TypeScript parser where a conditional type containing a union type that ends with an infer type that ends with a constraint could fail to parse. This was caused by the "don't parse a conditional type" flag not getting passed through the union type parser. Here's an example of valid TypeScript code that previously failed to parse correctly:
type InferUnion<T> = T extends { a: infer U extends number } | infer U extends number ? U : never
Fix TypeScript-specific class transform edge case (#3559)
The previous release introduced an optimization that avoided transforming super() in the class constructor for TypeScript code compiled with useDefineForClassFields set to false if all class instance fields have no initializers. The rationale was that in this case, all class instance fields are omitted in the output so no changes to the constructor are needed. However, if all of this is the case and there are #private instance fields with initializers, those private instance field initializers were still being moved into the constructor. This was problematic because they were being inserted before the call to super() (since super() is now no longer transformed in that case). This release introduces an additional optimization that avoids moving the private instance field initializers into the constructor in this edge case, which generates smaller code, matches the TypeScript compiler's output more closely, and avoids this bug:
// Original code
class Foo extends Bar {
#private = 1;
public: any;
constructor() {
super();
}
}
// Old output (with esbuild v0.19.9)
class Foo extends Bar {
constructor() {
super();
this.#private = 1;
}
#private;
}
// Old output (with esbuild v0.19.10)
class Foo extends Bar {
constructor() {
this.#private = 1;
super();
}
#private;
}
// New output
class Foo extends Bar {
#private = 1;
constructor() {
super();
}
}
Minifier: allow reording a primitive past a side-effect (#3568)
The minifier previously allowed reordering a side-effect past a primitive, but didn't handle the case of reordering a primitive past a side-effect. This additional case is now handled:
// Original code
function f() {
let x = false;
let y = x;
const boolean = y;
let frag = $.template(`<p contenteditable="${boolean}">hello world</p>`);
return frag;
}
// Old output (with --minify)
function f(){const e=!1;return $.template(`<p contenteditable="${e}">hello world</p>`)}
// New output (with --minify)
function f(){return $.template('<p contenteditable="false">hello world</p>')}
Minifier: consider properties named using known Symbol instances to be side-effect free (#3561)
Many things in JavaScript can have side effects including property accesses and ToString operations, so using a symbol such as Symbol.iterator as a computed property name is not obviously side-effect free. This release adds a special case for known Symbol instances so that they are considered side-effect free when used as property names. For example, this class declaration will now be considered side-effect free:
class Foo {
*[Symbol.iterator]() {
}
}
Provide the stop() API in node to exit esbuild's child process (#3558)
You can now call stop() in esbuild's node API to exit esbuild's child process to reclaim the resources used. It only makes sense to do this for a long-lived node process when you know you will no longer be making any more esbuild API calls. It is not necessary to call this to allow node to exit, and it's advantageous to not call this in between calls to esbuild's API as sharing a single long-lived esbuild child process is more efficient than re-creating a new esbuild child process for every API call. This API call used to exist but was removed in version 0.9.0. This release adds it back due to a user request.
Fix glob imports in TypeScript files (#3319)
This release fixes a problem where bundling a TypeScript file containing a glob import could emit a call to a helper function that doesn't exist. The problem happened because esbuild's TypeScript transformation removes unused imports (which is required for correctness, as they may be type-only imports) and esbuild's glob import transformation wasn't correctly marking the imported helper function as used. This wasn't caught earlier because most of esbuild's glob import tests were written in JavaScript, not in TypeScript.
Fix require() glob imports with bundling disabled (#3546)
Previously require() calls containing glob imports were incorrectly transformed when bundling was disabled. All glob imports should only be transformed when bundling is enabled. This bug has been fixed.
Fix a panic when transforming optional chaining with define (#3551, #3554)
This release fixes a case where esbuild could crash with a panic, which was triggered by using define to replace an expression containing an optional chain. Here is an example:
// Original code
console.log(process?.env.SHELL)
// Old output (with --define:process.env={})
/* panic: Internal error (while parsing "<stdin>") */
// New output (with --define:process.env={})
var define_process_env_default = {};
console.log(define_process_env_default.SHELL);
This fix was contributed by @hi-ogawa.
Work around a bug in node's CommonJS export name detector (#3544)
The export names of a CommonJS module are dynamically-determined at run time because CommonJS exports are properties on a mutable object. But the export names of an ES module are statically-determined at module instantiation time by using import and export syntax and cannot be changed at run time.
When you import a CommonJS module into an ES module in node, node scans over the source code to attempt to detect the set of export names that the CommonJS module will end up using. That statically-determined set of names is used as the set of names that the ES module is allowed to import at module instantiation time. However, this scan appears to have bugs (or at least, can cause false positives) because it doesn't appear to do any scope analysis. Node will incorrectly consider the module to export something even if the assignment is done to a local variable instead of to the module-level exports object. For example:
// confuseNode.js
exports.confuseNode = function(exports) {
// If this local is called "exports", node incorrectly
// thinks this file has an export called "notAnExport".
exports.notAnExport = function() {
};
};
You can see that node incorrectly thinks the file confuseNode.js has an export called notAnExport when that file is loaded in an ES module context:
$ node -e 'import("./confuseNode.js").then(console.log)'
[Module: null prototype] {
confuseNode: [Function (anonymous)],
default: { confuseNode: [Function (anonymous)] },
notAnExport: undefined
}
To avoid this, esbuild will now rename local variables that use the names exports and module when generating CommonJS output for the node platform.
Fix the return value of esbuild's super() shim (#3538)
Some people write constructor methods that use the return value of super() instead of using this. This isn't too common because TypeScript doesn't let you do that but it can come up when writing JavaScript. Previously esbuild's class lowering transform incorrectly transformed the return value of super() into undefined. With this release, the return value of super() will now be this instead:
// Original code
class Foo extends Object {
field
constructor() {
console.log(typeof super())
}
}
new Foo
// Old output (with --target=es6)
class Foo extends Object {
constructor() {
var __super = (...args) => {
super(...args);
__publicField(this, "field");
};
console.log(typeof __super());
}
}
new Foo();
// New output (with --target=es6)
class Foo extends Object {
constructor() {
var __super = (...args) => {
super(...args);
__publicField(this, "field");
return this;
};
console.log(typeof __super());
}
}
new Foo();
Terminate the Go GC when esbuild's stop() API is called (#3552)
If you use esbuild with WebAssembly and pass the worker: false flag to esbuild.initialize(), then esbuild will run the WebAssembly module on the main thread. If you do this within a Deno test and that test calls esbuild.stop() to clean up esbuild's resources, Deno may complain that a setTimeout() call lasted past the end of the test. This happens when the Go is in the middle of a garbage collection pass and has scheduled additional ongoing garbage collection work. Normally calling esbuild.stop() will terminate the web worker that the WebAssembly module runs in, which will terminate the Go GC, but that doesn't happen if you disable the web worker with worker: false.
With this release, esbuild will now attempt to terminate the Go GC in this edge case by calling clearTimeout() on these pending timeouts.
Apply /* @__NO_SIDE_EFFECTS__ */ on tagged template literals (#3511)
Tagged template literals that reference functions annotated with a @__NO_SIDE_EFFECTS__ comment are now able to be removed via tree-shaking if the result is unused. This is a convention from Rollup. Here is an example:
// Original code
const html = /* @__NO_SIDE_EFFECTS__ */ (a, ...b) => ({ a, b })
html`<a>remove</a>`
x = html`<b>keep</b>`
// Old output (with --tree-shaking=true)
const html = /* @__NO_SIDE_EFFECTS__ */ (a, ...b) => ({ a, b });
html`<a>remove</a>`;
x = html`<b>keep</b>`;
// New output (with --tree-shaking=true)
const html = /* @__NO_SIDE_EFFECTS__ */ (a, ...b) => ({ a, b });
x = html`<b>keep</b>`;
Note that this feature currently only works within a single file, so it's not especially useful. This feature does not yet work across separate files. I still recommend using @__PURE__ annotations instead of this feature, as they have wider tooling support. The drawback of course is that @__PURE__ annotations need to be added at each call site, not at the declaration, and for non-call expressions such as template literals you need to wrap the expression in an IIFE (immediately-invoked function expression) to create a call expression to apply the @__PURE__ annotation to.
Publish builds for IBM AIX PowerPC 64-bit (#3549)
This release publishes a binary executable to npm for IBM AIX PowerPC 64-bit, which means that in theory esbuild can now be installed in that environment with npm install esbuild. This hasn't actually been tested yet. If you have access to such a system, it would be helpful to confirm whether or not doing this actually works.
Add support for transforming new CSS gradient syntax for older browsers
The specification called CSS Images Module Level 4 introduces new CSS gradient syntax for customizing how the browser interpolates colors in between color stops. You can now control the color space that the interpolation happens in as well as (for "polar" color spaces) control whether hue angle interpolation happens clockwise or counterclockwise. You can read more about this in Mozilla's blog post about new CSS gradient features.
With this release, esbuild will now automatically transform this syntax for older browsers in the target list. For example, here's a gradient that should appear as a rainbow in a browser that supports this new syntax:
/* Original code */
.rainbow-gradient {
width: 100px;
height: 100px;
background: linear-gradient(in hsl longer hue, #7ff, #77f);
}
/* New output (with --target=chrome99) */
.rainbow-gradient {
width: 100px;
height: 100px;
background:
linear-gradient(
#77ffff,
#77ffaa 12.5%,
#77ff80 18.75%,
#84ff77 21.88%,
#99ff77 25%,
#eeff77 37.5%,
#fffb77 40.62%,
#ffe577 43.75%,
#ffbb77 50%,
#ff9077 56.25%,
#ff7b77 59.38%,
#ff7788 62.5%,
#ff77dd 75%,
#ff77f2 78.12%,
#f777ff 81.25%,
#cc77ff 87.5%,
#7777ff);
}
You can now use this syntax in your CSS source code and esbuild will automatically convert it to an equivalent gradient for older browsers. In addition, esbuild will now also transform "double position" and "transition hint" syntax for older browsers as appropriate:
/* Original code */
.stripes {
width: 100px;
height: 100px;
background: linear-gradient(#e65 33%, #ff2 33% 67%, #99e 67%);
}
.glow {
width: 100px;
height: 100px;
background: radial-gradient(white 10%, 20%, black);
}
/* New output (with --target=chrome33) */
.stripes {
width: 100px;
height: 100px;
background:
linear-gradient(
#e65 33%,
#ff2 33%,
#ff2 67%,
#99e 67%);
}
.glow {
width: 100px;
height: 100px;
background:
radial-gradient(
#ffffff 10%,
#aaaaaa 12.81%,
#959595 15.62%,
#7b7b7b 21.25%,
#5a5a5a 32.5%,
#444444 43.75%,
#323232 55%,
#161616 77.5%,
#000000);
}
You can see visual examples of these new syntax features by looking at esbuild's gradient transformation tests.
If necessary, esbuild will construct a new gradient that approximates the original gradient by recursively splitting the interval in between color stops until the approximation error is within a small threshold. That is why the above output CSS contains many more color stops than the input CSS.
Note that esbuild deliberately replaces the original gradient with the approximation instead of inserting the approximation before the original gradient as a fallback. The latest version of Firefox has multiple gradient rendering bugs (including incorrect interpolation of partially-transparent colors and interpolating non-sRGB colors using the incorrect color space). If esbuild didn't replace the original gradient, then Firefox would use the original gradient instead of the fallback the appearance would be incorrect in Firefox. In other words, the latest version of Firefox supports modern gradient syntax but interprets it incorrectly.
Add support for color(), lab(), lch(), oklab(), oklch(), and hwb() in CSS
CSS has recently added lots of new ways of specifying colors. You can read more about this in Chrome's blog post about CSS color spaces.
This release adds support for minifying colors that use the color(), lab(), lch(), oklab(), oklch(), or hwb() syntax and/or transforming these colors for browsers that don't support it yet:
/* Original code */
div {
color: hwb(90deg 20% 40%);
background: color(display-p3 1 0 0);
}
/* New output (with --target=chrome99) */
div {
color: #669933;
background: #ff0f0e;
background: color(display-p3 1 0 0);
}
As you can see, colors outside of the sRGB color space such as color(display-p3 1 0 0) are mapped back into the sRGB gamut and inserted as a fallback for browsers that don't support the new color syntax.
Allow empty type parameter lists in certain cases (#3512)
TypeScript allows interface declarations and type aliases to have empty type parameter lists. Previously esbuild didn't handle this edge case but with this release, esbuild will now parse this syntax:
interface Foo<> {}
type Bar<> = {}
This fix was contributed by @magic-akari.
Add a treemap chart to esbuild's bundle analyzer (#2848)
The bundler analyzer on esbuild's website (https://esbuild.github.io/analyze/) now has a treemap chart type in addition to the two existing chart types (sunburst and flame). This should be more familiar for people coming from other similar tools, as well as make better use of large screens.
Allow decorators after the export keyword (#104)
Previously esbuild's decorator parser followed the original behavior of TypeScript's experimental decorators feature, which only allowed decorators to come before the export keyword. However, the upcoming JavaScript decorators feature also allows decorators to come after the export keyword. And with TypeScript 5.0, TypeScript now also allows experimental decorators to come after the export keyword too. So esbuild now allows this as well:
// This old syntax has always been permitted:
@decorator export class Foo {}
@decorator export default class Foo {}
// This new syntax is now permitted too:
export @decorator class Foo {}
export default @decorator class Foo {}
In addition, esbuild's decorator parser has been rewritten to fix several subtle and likely unimportant edge cases with esbuild's parsing of exports and decorators in TypeScript (e.g. TypeScript apparently does automatic semicolon insertion after interface and export interface but not after export default interface).
Pretty-print decorators using the same whitespace as the original
When printing code containing decorators, esbuild will now try to respect whether the original code contained newlines after the decorator or not. This can make generated code containing many decorators much more compact to read:
// Original code
class Foo {
@a @b @c abc
@x @y @z xyz
}
// Old output
class Foo {
@a
@b
@c
abc;
@x
@y
@z
xyz;
}
// New output
class Foo {
@a @b @c abc;
@x @y @z xyz;
}
Add support for bundling code that uses import attributes (#3384)
JavaScript is gaining new syntax for associating a map of string key-value pairs with individual ESM imports. The proposal is still a work in progress and is still undergoing significant changes before being finalized. However, the first iteration has already been shipping in Chromium-based browsers for a while, and the second iteration has landed in V8 and is now shipping in node, so it makes sense for esbuild to support it. Here are the two major iterations of this proposal (so far):
Import assertions (deprecated, will not be standardized)
assert keywordImport attributes (currently set to become standardized)
with keywordYou can already use esbuild to bundle code that uses import assertions (the first iteration). However, this feature is mostly useless for bundlers because import assertions are not allowed to affect module resolution. It's basically only useful as an annotation on external imports, which esbuild will then preserve in the output for use in a browser (which would otherwise refuse to load certain imports).
With this release, esbuild now supports bundling code that uses import attributes (the second iteration). This is much more useful for bundlers because they are allowed to affect module resolution, which means the key-value pairs can be provided to plugins. Here's an example, which uses esbuild's built-in support for the upcoming JSON module standard:
// On static imports
import foo from './package.json' with { type: 'json' }
console.log(foo)
// On dynamic imports
const bar = await import('./package.json', { with: { type: 'json' } })
console.log(bar)
One important consequence of the change in semantics between import assertions and import attributes is that two imports with identical paths but different import attributes are now considered to be different modules. This is because the import attributes are provided to the loader, which might then use those attributes during loading. For example, you could imagine an image loader that produces an image of a different size depending on the import attributes.
Import attributes are now reported in the metafile and are now provided to on-load plugins as a map in the with property. For example, here's an esbuild plugin that turns all imports with a type import attribute equal to 'cheese' into a module that exports the cheese emoji:
const cheesePlugin = {
name: 'cheese',
setup(build) {
build.onLoad({ filter: /.*/ }, args => {
if (args.with.type === 'cheese') return {
contents: `export default "🧀"`,
}
})
}
}
require('esbuild').build({
bundle: true,
write: false,
stdin: {
contents: `
import foo from 'data:text/javascript,' with { type: 'cheese' }
console.log(foo)
`,
},
plugins: [cheesePlugin],
}).then(result => {
const code = new Function(result.outputFiles[0].text)
code()
})
Warning: It's possible that the second iteration of this feature may change significantly again even though it's already shipping in real JavaScript VMs (since it has already happened once before). In that case, esbuild may end up adjusting its implementation to match the eventual standard behavior. So keep in mind that by using this, you are using an unstable upcoming JavaScript feature that may undergo breaking changes in the future.
Adjust TypeScript experimental decorator behavior (#3230, #3326, #3394)
With this release, esbuild will now allow TypeScript experimental decorators to access both static class properties and #private class names. For example:
const check =
<T,>(a: T, b: T): PropertyDecorator =>
() => console.log(a === b)
async function test() {
class Foo {
static #foo = 1
static bar = 1 + Foo.#foo
@check(Foo.#foo, 1) a: any
@check(Foo.bar, await Promise.resolve(2)) b: any
}
}
test().then(() => console.log('pass'))
This will now print true true pass when compiled by esbuild. Previously esbuild evaluated TypeScript decorators outside of the class body, so it didn't allow decorators to access Foo or #foo. Now esbuild does something different, although it's hard to concisely explain exactly what esbuild is doing now (see the background section below for more information).
Note that TypeScript's experimental decorator support is currently buggy: TypeScript's compiler passes this test if only the first @check is present or if only the second @check is present, but TypeScript's compiler fails this test if both checks are present together. I haven't changed esbuild to match TypeScript's behavior exactly here because I'm waiting for TypeScript to fix these bugs instead.
Some background: TypeScript experimental decorators don't have consistent semantics regarding the context that the decorators are evaluated in. For example, TypeScript will let you use await within a decorator, which implies that the decorator runs outside the class body (since await isn't supported inside a class body), but TypeScript will also let you use #private names, which implies that the decorator runs inside the class body (since #private names are only supported inside a class body). The value of this in a decorator is also buggy (the run-time value of this changes if any decorator in the class uses a #private name but the type of this doesn't change, leading to the type checker no longer matching reality). These inconsistent semantics make it hard for esbuild to implement this feature as decorator evaluation happens in some superposition of both inside and outside the class body that is particular to the internal implementation details of the TypeScript compiler.
Forbid --keep-names when targeting old browsers (#3477)
The --keep-names setting needs to be able to assign to the name property on functions and classes. However, before ES6 this property was non-configurable, and attempting to assign to it would throw an error. So with this release, esbuild will no longer allow you to enable this setting while also targeting a really old browser.
Fix a constant folding bug with bigint equality
This release fixes a bug where esbuild incorrectly checked for bigint equality by checking the equality of the bigint literal text. This is correct if the bigint doesn't have a radix because bigint literals without a radix are always in canonical form (since leading zeros are not allowed). However, this is incorrect if the bigint has a radix (e.g. 0x123n) because the canonical form is not enforced when a radix is present.
// Original code
console.log(!!0n, !!1n, 123n === 123n)
console.log(!!0x0n, !!0x1n, 123n === 0x7Bn)
// Old output
console.log(false, true, true);
console.log(true, true, false);
// New output
console.log(false, true, true);
console.log(!!0x0n, !!0x1n, 123n === 0x7Bn);
Add some improvements to the JavaScript minifier
This release adds more cases to the JavaScript minifier, including support for inlining String.fromCharCode and String.prototype.charCodeAt when possible:
// Original code
document.onkeydown = e => e.keyCode === 'A'.charCodeAt(0) && console.log(String.fromCharCode(55358, 56768))
// Old output (with --minify)
document.onkeydown=o=>o.keyCode==="A".charCodeAt(0)&&console.log(String.fromCharCode(55358,56768));
// New output (with --minify)
document.onkeydown=o=>o.keyCode===65&&console.log("🧀");
In addition, immediately-invoked function expressions (IIFEs) that return a single expression are now inlined when minifying. This makes it possible to use IIFEs in combination with @__PURE__ annotations to annotate arbitrary expressions as side-effect free without the IIFE wrapper impacting code size. For example:
// Original code
const sideEffectFreeOffset = /* @__PURE__ */ (() => computeSomething())()
use(sideEffectFreeOffset)
// Old output (with --minify)
const e=(()=>computeSomething())();use(e);
// New output (with --minify)
const e=computeSomething();use(e);
Automatically prefix the mask-composite CSS property for WebKit (#3493)
The mask-composite property will now be prefixed as -webkit-mask-composite for older WebKit-based browsers. In addition to prefixing the property name, handling older browsers also requires rewriting the values since WebKit uses non-standard names for the mask composite modes:
/* Original code */
div {
mask-composite: add, subtract, intersect, exclude;
}
/* New output (with --target=chrome100) */
div {
-webkit-mask-composite:
source-over,
source-out,
source-in,
xor;
mask-composite:
add,
subtract,
intersect,
exclude;
}
Avoid referencing this from JSX elements in derived class constructors (#3454)
When you enable --jsx=automatic and --jsx-dev, the JSX transform is supposed to insert this as the last argument to the jsxDEV function. I'm not sure exactly why this is and I can't find any specification for it, but in any case this causes the generated code to crash when you use a JSX element in a derived class constructor before the call to super() as this is not allowed to be accessed at that point. For example
// Original code
class ChildComponent extends ParentComponent {
constructor() {
super(<div />)
}
}
// Problematic output (with --loader=jsx --jsx=automatic --jsx-dev)
import { jsxDEV } from "react/jsx-dev-runtime";
class ChildComponent extends ParentComponent {
constructor() {
super(/* @__PURE__ */ jsxDEV("div", {}, void 0, false, {
fileName: "<stdin>",
lineNumber: 3,
columnNumber: 15
}, this)); // The reference to "this" crashes here
}
}
The TypeScript compiler doesn't handle this at all while the Babel compiler just omits this for the entire constructor (even after the call to super()). There seems to be no specification so I can't be sure that this change doesn't break anything important. But given that Babel is pretty loose with this and TypeScript doesn't handle this at all, I'm guessing this value isn't too important. React's blog post seems to indicate that this value was intended to be used for a React-specific migration warning at some point, so it could even be that this value is irrelevant now. Anyway the crash in this case should now be fixed.
Allow package subpath imports to map to node built-ins (#3485)
You are now able to use a subpath import in your package to resolve to a node built-in module. For example, with a package.json file like this:
{
"type": "module",
"imports": {
"#stream": {
"node": "stream",
"default": "./stub.js"
}
}
}
You can now import from node's stream module like this:
import * as stream from '#stream';
console.log(Object.keys(stream));
This will import from node's stream module when the platform is node and from ./stub.js otherwise.
No longer throw an error when a Symbol is missing (#3453)
Certain JavaScript syntax features use special properties on the global Symbol object. For example, the asynchronous iteration syntax uses Symbol.asyncIterator. Previously esbuild's generated code for older browsers required this symbol to be polyfilled. However, starting with this release esbuild will use Symbol.for() to construct these symbols if they are missing instead of throwing an error about a missing polyfill. This means your code no longer needs to include a polyfill for missing symbols as long as your code also uses Symbol.for() for missing symbols.
Parse upcoming changes to TypeScript syntax (#3490, #3491)
With this release, you can now use from as the name of a default type-only import in TypeScript code, as well as of as the name of an await using loop iteration variable:
import type from from 'from'
for (await using of of of) ;
This matches similar changes in the TypeScript compiler (#56376 and #55555) which will start allowing this syntax in an upcoming version of TypeScript. Please never actually write code like this.
The type-only import syntax change was contributed by @magic-akari.