chore(deps): update dependency esbuild to ^0.18.0
This MR contains the following updates:
Package | Type | Update | Change |
---|---|---|---|
esbuild | devDependencies | minor | ^0.17.0 -> ^0.18.0 |
MR created with the help of gitlab-org/frontend/renovate-gitlab-bot
Release Notes
evanw/esbuild
v0.18.0
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 ^0.17.0
or ~0.17.0
. See npm's documentation about semver for more information.
The breaking changes in this release mainly focus on fixing some long-standing issues with esbuild's handling of tsconfig.json
files. Here are all the changes in this release, in detail:
-
Add a way to try esbuild online (#797)
There is now a way to try esbuild live on esbuild's website without installing it: https://esbuild.github.io/try/. In addition to being able to more easily evaluate esbuild, this should also make it more efficient to generate esbuild bug reports. For example, you can use it to compare the behavior of different versions of esbuild on the same input. The state of the page is stored in the URL for easy sharing. Many thanks to @hyrious for creating https://hyrious.me/esbuild-repl/, which was the main inspiration for this addition to esbuild's website.
Two forms of build options are supported: either CLI-style (example) or JS-style (example). Both are converted into a JS object that's passed to esbuild's WebAssembly API. The CLI-style argument parser is a custom one that simulates shell quoting rules, and the JS-style argument parser is also custom and parses a superset of JSON (basically JSON5 + regular expressions). So argument parsing is an approximate simulation of what happens for real but hopefully it should be close enough.
-
Changes to esbuild's
tsconfig.json
support (#3019):This release makes the following changes to esbuild's
tsconfig.json
support:-
Using experimental decorators now requires
"experimentalDecorators": true
(#104)Previously esbuild would always compile decorators in TypeScript code using TypeScript's experimental decorator transform. Now that standard JavaScript decorators are close to being finalized, esbuild will now require you to use
"experimentalDecorators": true
to do this. This new requirement makes it possible for esbuild to introduce a transform for standard JavaScript decorators in TypeScript code in the future. Such a transform has not been implemented yet, however. -
TypeScript's
target
no longer affects esbuild'starget
(#2628)Some people requested that esbuild support TypeScript's
target
setting, so support for it was added (in version 0.12.4). However, esbuild supports reading from multipletsconfig.json
files within a single build, which opens up the possibility that different files in the build have different language targets configured. There isn't really any reason to do this and it can lead to unexpected results. So with this release, thetarget
setting intsconfig.json
will no longer affect esbuild's owntarget
setting. You will have to use esbuild's own target setting instead (which is a single, global value). -
TypeScript's
jsx
setting no longer causes esbuild to preserve JSX syntax (#2634)TypeScript has a setting called
jsx
that controls how to transform JSX into JS. The tool-agnostic transform is calledreact
, and the React-specific transform is calledreact-jsx
(orreact-jsxdev
). There is also a setting calledpreserve
which indicates JSX should be passed through untransformed. Previously people would run esbuild with"jsx": "preserve"
in theirtsconfig.json
files and then be surprised when esbuild preserved their JSX. So with this release, esbuild will now ignore"jsx": "preserve"
intsconfig.json
files. If you want to preserve JSX syntax with esbuild, you now have to use--jsx=preserve
.Note: Some people have suggested that esbuild's equivalent
jsx
setting override the one intsconfig.json
. However, some projects need to legitimately have different files within the same build use different transforms (i.e.react
vs.react-jsx
) and having esbuild's globaljsx
setting overridetsconfig.json
would prevent this from working. This release ignores"jsx": "preserve"
but still allows otherjsx
values intsconfig.json
files to override esbuild's globaljsx
setting to keep the ability for multiple files within the same build to use different transforms. -
useDefineForClassFields
behavior has changed (#2584, #2993)Class fields in TypeScript look like this (
x
is a class field):class Foo { x = 123 }
TypeScript has legacy behavior that uses assignment semantics instead of define semantics for class fields when
useDefineForClassFields
is enabled (in which case class fields in TypeScript behave differently than they do in JavaScript, which is arguably "wrong").This legacy behavior exists because TypeScript added class fields to TypeScript before they were added to JavaScript. The TypeScript team decided to go with assignment semantics and shipped their implementation. Much later on TC39 decided to go with define semantics for class fields in JavaScript instead. This behaves differently if the base class has a setter with the same name:
class Base { set x(_) { console.log('x:', _) } } // useDefineForClassFields: false class AssignSemantics extends Base { constructor() { super() this.x = 123 } } // useDefineForClassFields: true class DefineSemantics extends Base { constructor() { super() Object.defineProperty(this, 'x', { value: 123 }) } } console.log( new AssignSemantics().x, // Calls the setter new DefineSemantics().x // Doesn't call the setter )
When you run
tsc
, the value ofuseDefineForClassFields
defaults tofalse
when it's not specified and thetarget
intsconfig.json
is present but earlier thanES2022
. This sort of makes sense because the class field language feature was added in ES2022, so before ES2022 class fields didn't exist (and thus TypeScript's legacy behavior is active). However, TypeScript'starget
setting currently defaults toES3
which unfortunately means that theuseDefineForClassFields
setting currently defaults to false (i.e. to "wrong"). In other words if you runtsc
with all default settings, class fields will behave incorrectly.Previously esbuild tried to do what
tsc
did. That meant esbuild's version ofuseDefineForClassFields
wasfalse
by default, and was alsofalse
if esbuild's--target=
was present but earlier thanes2022
. However, TypeScript's legacy class field behavior is becoming increasingly irrelevant and people who expect class fields in TypeScript to work like they do in JavaScript are confused when they use esbuild with default settings. It's also confusing that the behavior of class fields would change if you changed the language target (even though that's exactly how TypeScript works).So with this release, esbuild will now only use the information in
tsconfig.json
to determine whetheruseDefineForClassFields
is true or not. SpecificallyuseDefineForClassFields
will be respected if present, otherwise it will befalse
iftarget
is present intsconfig.json
and isES2021
or earlier, otherwise it will betrue
. Targets passed to esbuild's--target=
setting will no longer affectuseDefineForClassFields
.Note that this means different directories in your build can have different values for this setting since esbuild allows different directories to have different
tsconfig.json
files within the same build. This should let you migrate your code one directory at a time without esbuild's--target=
setting affecting the semantics of your code. -
Add support for
verbatimModuleSyntax
from TypeScript 5.0TypeScript 5.0 added a new option called
verbatimModuleSyntax
that deprecates and replaces two older options,preserveValueImports
andimportsNotUsedAsValues
. SettingverbatimModuleSyntax
to true intsconfig.json
tells esbuild to not drop unused import statements. Specifically esbuild now treats"verbatimModuleSyntax": true
as if you had specified both"preserveValueImports": true
and"importsNotUsedAsValues": "preserve"
. -
Add multiple inheritance for
tsconfig.json
from TypeScript 5.0TypeScript 5.0 now allows multiple inheritance for
tsconfig.json
files. You can now pass an array of filenames via theextends
parameter and yourtsconfig.json
will start off containing properties from all of those configuration files, in order. This release of esbuild adds support for this new TypeScript feature. -
Remove support for
moduleSuffixes
(#2395)The community has requested that esbuild remove support for TypeScript's
moduleSuffixes
feature, so it has been removed in this release. Instead you can use esbuild's--resolve-extensions=
feature to select which module suffix you want to build with. -
Apply
--tsconfig=
overrides tostdin
and virtual files (#385, #2543)When you override esbuild's automatic
tsconfig.json
file detection with--tsconfig=
to pass a specifictsconfig.json
file, esbuild previously didn't apply these settings to source code passed via thestdin
API option or to TypeScript files from plugins that weren't in thefile
namespace. This release changes esbuild's behavior so that settings fromtsconfig.json
also apply to these source code files as well. -
Support
--tsconfig-raw=
in build API calls (#943, #2440)Previously if you wanted to override esbuild's automatic
tsconfig.json
file detection, you had to create a newtsconfig.json
file and pass the file name to esbuild via the--tsconfig=
flag. With this release, you can now optionally use--tsconfig-raw=
instead to pass the contents oftsconfig.json
to esbuild directly instead of passing the file name. For example, you can now use--tsconfig-raw={"compilerOptions":{"experimentalDecorators":true}}
to enable TypeScript experimental decorators directly using a command-line flag (assuming you escape the quotes correctly using your current shell's quoting rules). The--tsconfig-raw=
flag previously only worked with transform API calls but with this release, it now works with build API calls too. -
Ignore all
tsconfig.json
files innode_modules
(#276, #2386)This changes esbuild's behavior that applies
tsconfig.json
to all files in the subtree of the directory containingtsconfig.json
. In version 0.12.7, esbuild started ignoringtsconfig.json
files insidenode_modules
folders. The rationale is that people typically do this by mistake and that doing this intentionally is a rare use case that doesn't need to be supported. However, this change only applied to certain syntax-specific settings (e.g.jsxFactory
) but did not apply to path resolution settings (e.g.paths
). With this release, esbuild will now ignore alltsconfig.json
files innode_modules
instead of only ignoring certain settings. -
Ignore
tsconfig.json
when resolving paths withinnode_modules
(#2481)Previously fields in
tsconfig.json
related to path resolution (e.g.paths
) were respected for all files in the subtree containing thattsconfig.json
file, even within a nestednode_modules
subdirectory. This meant that a project'spaths
settings could potentially affect any bundled packages. With this release, esbuild will no longer usetsconfig.json
settings during path resolution inside nestednode_modules
subdirectories. -
Prefer
.js
over.ts
withinnode_modules
(#3019)The default list of implicit extensions that esbuild will try appending to import paths contains
.ts
before.js
. This makes it possible to bundle TypeScript projects that reference other files in the project using extension-less imports (e.g../some-file
to load./some-file.ts
instead of./some-file.js
). However, this behavior is undesirable withinnode_modules
directories. Some package authors publish both their original TypeScript code and their compiled JavaScript code side-by-side. In these cases, esbuild should arguably be using the compiled JavaScript files instead of the original TypeScript files because the TypeScript compilation settings for files within the package should be determined by the package author, not the user of esbuild. So with this release, esbuild will now prefer implicit.js
extensions over.ts
when searching for import paths withinnode_modules
.
These changes are intended to improve esbuild's compatibility with
tsc
and reduce the number of unfortunate behaviors regardingtsconfig.json
and esbuild. -
-
Add a workaround for bugs in Safari 16.2 and earlier (#3072)
Safari's JavaScript parser had a bug (which has now been fixed) where at least something about unary/binary operators nested inside default arguments nested inside either a function or class expression was incorrectly considered a syntax error if that expression was the target of a property assignment. Here are some examples that trigger this Safari bug:
❱ x(function (y = -1) {}.z = 2) SyntaxError: Left hand side of operator '=' must be a reference. ❱ x(class { f(y = -1) {} }.z = 2) SyntaxError: Left hand side of operator '=' must be a reference.
It's not clear what the exact conditions are that trigger this bug. However, a workaround for this bug appears to be to post-process your JavaScript to wrap any in function and class declarations that are the direct target of a property access expression in parentheses. That's the workaround that UglifyJS applies for this issue: mishoo/UglifyJS#2056. So that's what esbuild now does starting with this release:
// Original code x(function (y = -1) {}.z = 2, class { f(y = -1) {} }.z = 2) // Old output (with --minify --target=safari16.2) x(function(c=-1){}.z=2,class{f(c=-1){}}.z=2); // New output (with --minify --target=safari16.2) x((function(c=-1){}).z=2,(class{f(c=-1){}}).z=2);
This fix is not enabled by default. It's only enabled when
--target=
contains Safari 16.2 or earlier, such as with--target=safari16.2
. You can also explicitly enable or disable this specific transform (calledfunction-or-class-property-access
) with--supported:function-or-class-property-access=false
. -
Fix esbuild's TypeScript type declarations to forbid unknown properties (#3089)
Version 0.17.0 of esbuild introduced a specific form of function overloads in the TypeScript type definitions for esbuild's API calls that looks like this:
interface TransformOptions { legalComments?: 'none' | 'inline' | 'eof' | 'external' } interface TransformResult<ProvidedOptions extends TransformOptions = TransformOptions> { legalComments: string | (ProvidedOptions['legalComments'] extends 'external' ? never : undefined) } declare function transformSync<ProvidedOptions extends TransformOptions>(input: string, options?: ProvidedOptions): TransformResult<ProvidedOptions> declare function transformSync(input: string, options?: TransformOptions): TransformResult
This more accurately reflects how esbuild's JavaScript API behaves. The result object returned by
transformSync
only has thelegalComments
property if you passlegalComments: 'external'
:// These have type "string | undefined" transformSync('').legalComments transformSync('', { legalComments: 'eof' }).legalComments // This has type "string" transformSync('', { legalComments: 'external' }).legalComments
However, this form of function overloads unfortunately allows typos (e.g.
egalComments
) to pass the type checker without generating an error as TypeScript allows all objects with unknown properties to extendTransformOptions
. These typos result in esbuild's API throwing an error at run-time.To prevent typos during type checking, esbuild's TypeScript type definitions will now use a different form that looks like this:
type SameShape<Out, In extends Out> = In & { [Key in Exclude<keyof In, keyof Out>]: never } interface TransformOptions { legalComments?: 'none' | 'inline' | 'eof' | 'external' } interface TransformResult<ProvidedOptions extends TransformOptions = TransformOptions> { legalComments: string | (ProvidedOptions['legalComments'] extends 'external' ? never : undefined) } declare function transformSync<T extends TransformOptions>(input: string, options?: SameShape<TransformOptions, T>): TransformResult<T>
This change should hopefully not affect correct code. It should hopefully introduce type errors only for incorrect code.
-
Fix CSS nesting transform for pseudo-elements (#3119)
This release fixes esbuild's CSS nesting transform for pseudo-elements (e.g.
::before
and::after
). The CSS nesting specification says that the nesting selector does not work with pseudo-elements. This can be seen in the example below: esbuild does not carry the parent pseudo-element::before
through the nesting selector&
. However, that doesn't apply to pseudo-elements that are within the same selector. Previously esbuild had a bug where it considered pseudo-elements in both locations as invalid. This release changes esbuild to only consider those from the parent selector invalid, which should align with the specification:/* Original code */ a, b::before { &.c, &::after { content: 'd'; } } /* Old output (with --target=chrome90) */ a:is(.c, ::after) { content: "d"; } /* New output (with --target=chrome90) */ a.c, a::after { content: "d"; }
-
Forbid
&
before a type selector in nested CSSThe people behind the work-in-progress CSS nesting specification have very recently decided to forbid nested CSS that looks like
&div
. You will have to use eitherdiv&
or&:is(div)
instead. This release of esbuild has been updated to take this new change into consideration. Doing this now generates a warning. The suggested fix is slightly different depending on where in the overall selector it happened:▲ [WARNING] Cannot use type selector "input" directly after nesting selector "&" [css-syntax-error] example.css:2:3: 2 │ &input { │ ~~~~~ ╵ :is(input) CSS nesting syntax does not allow the "&" selector to come before a type selector. You can wrap this selector in ":is()" as a workaround. This restriction exists to avoid problems with SASS nesting, where the same syntax means something very different that has no equivalent in real CSS (appending a suffix to the parent selector). ▲ [WARNING] Cannot use type selector "input" directly after nesting selector "&" [css-syntax-error] example.css:6:8: 6 │ .form &input { │ ~~~~~~ ╵ input& CSS nesting syntax does not allow the "&" selector to come before a type selector. You can move the "&" to the end of this selector as a workaround. This restriction exists to avoid problems with SASS nesting, where the same syntax means something very different that has no equivalent in real CSS (appending a suffix to the parent selector).
v0.17.19
-
Fix CSS transform bugs with nested selectors that start with a combinator (#3096)
This release fixes several bugs regarding transforming nested CSS into non-nested CSS for older browsers. The bugs were due to lack of test coverage for nested selectors with more than one compound selector where they all start with the same combinator. Here's what some problematic cases look like before and after these fixes:
/* Original code */ .foo { > &a, > &b { color: red; } } .bar { > &a, + &b { color: green; } } /* Old output (with --target=chrome90) */ .foo :is(> .fooa, > .foob) { color: red; } .bar :is(> .bara, + .barb) { color: green; } /* New output (with --target=chrome90) */ .foo > :is(a.foo, b.foo) { color: red; } .bar > a.bar, .bar + b.bar { color: green; }
-
Fix bug with TypeScript parsing of instantiation expressions followed by
=
(#3111)This release fixes esbuild's TypeScript-to-JavaScript conversion code in the case where a potential instantiation expression is followed immediately by a
=
token (such that the trailing>
becomes a>=
token). Previously esbuild considered that to still be an instantiation expression, but the official TypeScript compiler considered it to be a>=
operator instead. This release changes esbuild's interpretation to match TypeScript. This edge case currently appears to be problematic for other TypeScript-to-JavaScript converters as well:Original code TypeScript esbuild 0.17.18 esbuild 0.17.19 Sucrase Babel x<y>=a<b<c>>()
x<y>=a();
x=a();
x<y>=a();
x=a()
Invalid left-hand side in assignment expression -
Avoid removing unrecognized directives from the directive prologue when minifying (#3115)
The directive prologue in JavaScript is a sequence of top-level string expressions that come before your code. The only directives that JavaScript engines currently recognize are
use strict
and sometimesuse asm
. However, the people behind React have made up their own directive for their own custom dialect of JavaScript. Previously esbuild only preserved theuse strict
directive when minifying, although you could still write React JavaScript with esbuild using something like--banner:js="'your directive here';"
. With this release, you can now put arbitrary directives in the entry point and esbuild will preserve them in its minified output:// Original code 'use wtf'; console.log(123) // Old output (with --minify) console.log(123); // New output (with --minify) "use wtf";console.log(123);
Note that this means esbuild will no longer remove certain stray top-level strings when minifying. This behavior is an intentional change because these stray top-level strings are actually part of the directive prologue, and could potentially have semantics assigned to them (as was the case with React).
-
Improved minification of binary shift operators
With this release, esbuild's minifier will now evaluate the
<<
and>>>
operators if the resulting code would be shorter:// Original code console.log(10 << 10, 10 << 20, -123 >>> 5, -123 >>> 10); // Old output (with --minify) console.log(10<<10,10<<20,-123>>>5,-123>>>10); // New output (with --minify) console.log(10240,10<<20,-123>>>5,4194303);
v0.17.18
-
Fix non-default JSON import error with
export {} from
(#3070)This release fixes a bug where esbuild incorrectly identified statements of the form
export { default as x } from "y" assert { type: "json" }
as a non-default import. The bug did not affect code of the formimport { default as x } from ...
(only code that used theexport
keyword). -
Fix a crash with an invalid subpath import (#3067)
Previously esbuild could crash when attempting to generate a friendly error message for an invalid subpath import (i.e. an import starting with
#
). This happened because esbuild originally only supported theexports
field and the code for that error message was not updated when esbuild later added support for theimports
field. This crash has been fixed.
v0.17.17
-
Fix CSS nesting transform for top-level
&
(#3052)Previously esbuild could crash with a stack overflow when lowering CSS nesting rules with a top-level
&
, such as in the code below. This happened because esbuild's CSS nesting transform didn't handle top-level&
, causing esbuild to inline the top-level selector into itself. This release handles top-level&
by replacing it with the:scope
pseudo-class:/* Original code */ &, a { .b { color: red; } } /* New output (with --target=chrome90) */ :is(:scope, a) .b { color: red; }
-
Support
exports
inpackage.json
forextends
intsconfig.json
(#3058)TypeScript 5.0 added the ability to use
extends
intsconfig.json
to reference a path in a package whosepackage.json
file contains anexports
map that points to the correct location. This doesn't automatically work in esbuild becausetsconfig.json
affects esbuild's path resolution, so esbuild's normal path resolution logic doesn't apply.This release adds support for doing this by adding some additional code that attempts to resolve the
extends
path using theexports
field. The behavior should be similar enough to esbuild's main path resolution logic to work as expected.Note that esbuild always treats this
extends
import as arequire()
import since that's what TypeScript appears to do. Specifically therequire
condition will be active and theimport
condition will be inactive. -
Fix watch mode with
NODE_PATH
(#3062)Node has a rarely-used feature where you can extend the set of directories that node searches for packages using the
NODE_PATH
environment variable. While esbuild supports this too, previously a bug prevented esbuild's watch mode from picking up changes to imported files that were contained directly in aNODE_PATH
directory. You're supposed to useNODE_PATH
for packages, but some people abuse this feature by putting files in that directory instead (e.g.node_modules/some-file.js
instead ofnode_modules/some-pkg/some-file.js
). The watch mode bug happens when you do this because esbuild first tries to readsome-file.js
as a directory and then as a file. Watch mode was incorrectly waiting forsome-file.js
to become a valid directory. This release fixes this edge case bug by changing watch mode to watchsome-file.js
as a file when this happens.
v0.17.16
-
Fix CSS nesting transform for triple-nested rules that start with a combinator (#3046)
This release fixes a bug with esbuild where triple-nested CSS rules that start with a combinator were not transformed correctly for older browsers. Here's an example of such a case before and after this bug fix:
/* Original input */ .a { color: red; > .b { color: green; > .c { color: blue; } } } /* Old output (with --target=chrome90) */ .a { color: red; } .a > .b { color: green; } .a .b > .c { color: blue; } /* New output (with --target=chrome90) */ .a { color: red; } .a > .b { color: green; } .a > .b > .c { color: blue; }
-
Support
--inject
with a file loaded using thecopy
loader (#3041)This release now allows you to use
--inject
with a file that is loaded using thecopy
loader. Thecopy
loader copies the imported file to the output directory verbatim and rewrites the path in theimport
statement to point to the copied output file. When used with--inject
, this means the injected file will be copied to the output directory as-is and a bareimport
statement for that file will be inserted in any non-copy output files that esbuild generates.Note that since esbuild doesn't parse the contents of copied files, esbuild will not expose any of the export names as usable imports when you do this (in the way that esbuild's
--inject
feature is typically used). However, any side-effects that the injected file has will still occur.
v0.17.15
-
Allow keywords as type parameter names in mapped types (#3033)
TypeScript allows type keywords to be used as parameter names in mapped types. Previously esbuild incorrectly treated this as an error. Code that does this is now supported:
type Foo = 'a' | 'b' | 'c' type A = { [keyof in Foo]: number } type B = { [infer in Foo]: number } type C = { [readonly in Foo]: number }
-
Add annotations for re-exported modules in node (#2486, #3029)
Node lets you import named imports from a CommonJS module using ESM import syntax. However, the allowed names aren't derived from the properties of the CommonJS module. Instead they are derived from an arbitrary syntax-only analysis of the CommonJS module's JavaScript AST.
To accommodate node doing this, esbuild's ESM-to-CommonJS conversion adds a special non-executable "annotation" for node that describes the exports that node should expose in this scenario. It takes the form
0 && (module.exports = { ... })
and comes at the end of the file (0 && expr
meansexpr
is never evaluated).Previously esbuild didn't do this for modules re-exported using the
export * from
syntax. Annotations for these re-exports will now be added starting with this release:// Original input export { foo } from './foo' export * from './bar' // Old output (with --format=cjs --platform=node) ... 0 && (module.exports = { foo }); // New output (with --format=cjs --platform=node) ... 0 && (module.exports = { foo, ...require("./bar") });
Note that you need to specify both
--format=cjs
and--platform=node
to get these node-specific annotations. -
Avoid printing an unnecessary space in between a number and a
.
(#3026)JavaScript typically requires a space in between a number token and a
.
token to avoid the.
being interpreted as a decimal point instead of a member expression. However, this space is not required if the number token itself contains a decimal point, an exponent, or uses a base other than 10. This release of esbuild now avoids printing the unnecessary space in these cases:// Original input foo(1000 .x, 0 .x, 0.1 .x, 0.0001 .x, 0xFFFF_0000_FFFF_0000 .x) // Old output (with --minify) foo(1e3 .x,0 .x,.1 .x,1e-4 .x,0xffff0000ffff0000 .x); // New output (with --minify) foo(1e3.x,0 .x,.1.x,1e-4.x,0xffff0000ffff0000.x);
-
Fix server-sent events with live reload when writing to the file system root (#3027)
This release fixes a bug where esbuild previously failed to emit server-sent events for live reload when
outdir
was the file system root, such as/
. This happened because/
is the only path on Unix that cannot have a trailing slash trimmed from it, which was fixed by improved path handling.
v0.17.14
-
Allow the TypeScript 5.0
const
modifier in object type declarations (#3021)The new TypeScript 5.0
const
modifier was added to esbuild in version 0.17.5, and works with classes, functions, and arrow expressions. However, support for it wasn't added to object type declarations (e.g. interfaces) due to an oversight. This release adds support for these cases, so the following TypeScript 5.0 code can now be built with esbuild:interface Foo { <const T>(): T } type Bar = { new <const T>(): T }
-
Implement preliminary lowering for CSS nesting (#1945)
Chrome has implemented the new CSS nesting specification in version 112, which is currently in beta but will become stable very soon. So CSS nesting is now a part of the web platform!
This release of esbuild can now transform nested CSS syntax into non-nested CSS syntax for older browsers. The transformation relies on the
:is()
pseudo-class in many cases, so the transformation is only guaranteed to work when targeting browsers that support:is()
(e.g. Chrome 88+). You'll need to set esbuild'starget
to the browsers you intend to support to tell esbuild to do this transformation. You will get a warning if you use CSS nesting syntax with atarget
which includes older browsers that don't support:is()
.The lowering transformation looks like this:
/* Original input */ a.btn { color: #​333; &:hover { color: #​444 } &:active { color: #​555 } } /* New output (with --target=chrome88) */ a.btn { color: #​333; } a.btn:hover { color: #​444; } a.btn:active { color: #​555; }
More complex cases may generate the
:is()
pseudo-class:/* Original input */ div, p { .warning, .error { padding: 20px; } } /* New output (with --target=chrome88) */ :is(div, p) :is(.warning, .error) { padding: 20px; }
In addition, esbuild now has a special warning message for nested style rules that start with an identifier. This isn't allowed in CSS because the syntax would be ambiguous with the existing declaration syntax. The new warning message looks like this:
▲ [WARNING] A nested style rule cannot start with "p" because it looks like the start of a declaration [css-syntax-error] <stdin>:1:7: 1 │ main { p { margin: auto } } │ ^ ╵ :is(p) To start a nested style rule with an identifier, you need to wrap the identifier in ":is(...)" to prevent the rule from being parsed as a declaration.
Keep in mind that the transformation in this release is a preliminary implementation. CSS has many features that interact in complex ways, and there may be some edge cases that don't work correctly yet.
-
Minification now removes unnecessary
&
CSS nesting selectorsThis release introduces the following CSS minification optimizations:
/* Original input */ a { font-weight: bold; & { color: blue; } & :hover { text-decoration: underline; } } /* Old output (with --minify) */ a{font-weight:700;&{color:#​00f}& :hover{text-decoration:underline}} /* New output (with --minify) */ a{font-weight:700;:hover{text-decoration:underline}color:#​00f}
-
Minification now removes duplicates from CSS selector lists
This release introduces the following CSS minification optimization:
/* Original input */ div, div { color: red } /* Old output (with --minify) */ div,div{color:red} /* New output (with --minify) */ div{color:red}
v0.17.13
-
Work around an issue with
NODE_PATH
and Go's WebAssembly internals (#3001)Go's WebAssembly implementation returns
EINVAL
instead ofENOTDIR
when using thereaddir
syscall on a file. This messes up esbuild's implementation of node's module resolution algorithm since encounteringENOTDIR
causes esbuild to continue its search (since it's a normal condition) while other encountering other errors causes esbuild to fail with an I/O error (since it's an unexpected condition). You can encounter this issue in practice if you use node's legacyNODE_PATH
feature to tell esbuild to resolve node modules in a custom directory that was not installed by npm. This release works around this problem by convertingEINVAL
intoENOTDIR
for thereaddir
syscall. -
Fix a minification bug with CSS
@layer
rules that have parsing errors (#3016)CSS at-rules require either a
{}
block or a semicolon at the end. Omitting both of these causes esbuild to treat the rule as an unknown at-rule. Previous releases of esbuild had a bug that incorrectly removed unknown at-rules without any children during minification if the at-rule token matched an at-rule that esbuild can handle. Specifically cssnano can generate@layer
rules with parsing errors, and empty@layer
rules cannot be removed because they have side effects (@layer
didn't exist when esbuild's CSS support was added, so esbuild wasn't written to handle this). This release changes esbuild to no longer discard@layer
rules with parsing errors when minifying (the rule@layer c
has a parsing error):/* Original input */ @​layer a { @​layer b { @​layer c } } /* Old output (with --minify) */ @​layer a.b; /* New output (with --minify) */ @​layer a.b.c;
-
Unterminated strings in CSS are no longer an error
The CSS specification provides rules for handling parsing errors. One of those rules is that user agents must close strings upon reaching the end of a line (i.e., before an unescaped line feed, carriage return or form feed character), but then drop the construct (declaration or rule) in which the string was found. For example:
p { color: green; font-family: 'Courier New Times color: red; color: green; }
...would be treated the same as:
p { color: green; color: green; }
...because the second declaration (from
font-family
to the semicolon aftercolor: red
) is invalid and is dropped.Previously using this CSS with esbuild failed to build due to a syntax error, even though the code can be interpreted by a browser. With this release, the code now produces a warning instead of an error, and esbuild prints the invalid CSS such that it stays invalid in the output:
/* esbuild's new non-minified output: */ p { color: green; font-family: 'Courier New Times color: red; color: green; }
/* esbuild's new minified output: */ p{font-family:'Courier New Times color: red;color:green}
v0.17.12
-
Fix a crash when parsing inline TypeScript decorators (#2991)
Previously esbuild's TypeScript parser crashed when parsing TypeScript decorators if the definition of the decorator was inlined into the decorator itself:
@​(function sealed(constructor: Function) { Object.seal(constructor); Object.seal(constructor.prototype); }) class Foo {}
This crash was not noticed earlier because this edge case did not have test coverage. The crash is fixed in this release.
v0.17.11
-
Fix the
alias
feature to always prefer the longest match (#2963)It's possible to configure conflicting aliases such as
--alias:a=b
and--alias:a/c=d
, which is ambiguous for the import patha/c/x
(since it could map to eitherb/c/x
ord/x
). Previously esbuild would pick the first matchingalias
, which would non-deterministically pick between one of the possible matches. This release fixes esbuild to always deterministically pick the longest possible match. -
Minify calls to some global primitive constructors (#2962)
With this release, esbuild's minifier now replaces calls to
Boolean
/Number
/String
/BigInt
with equivalent shorter code when relevant:// Original code console.log( Boolean(a ? (b | c) !== 0 : (c & d) !== 0), Number(e ? '1' : '2'), String(e ? '1' : '2'), BigInt(e ? 1n : 2n), ) // Old output (with --minify) console.log(Boolean(a?(b|c)!==0:(c&d)!==0),Number(e?"1":"2"),String(e?"1":"2"),BigInt(e?1n:2n)); // New output (with --minify) console.log(!!(a?b|c:c&d),+(e?"1":"2"),e?"1":"2",e?1n:2n);
-
Adjust some feature compatibility tables for node (#2940)
This release makes the following adjustments to esbuild's internal feature compatibility tables for node, which tell esbuild which versions of node are known to support all aspects of that feature:
-
class-private-brand-checks
: node v16.9+ => node v16.4+ (a decrease) -
hashbang
: node v12.0+ => node v12.5+ (an increase) -
optional-chain
: node v16.9+ => node v16.1+ (a decrease) -
template-literal
: node v4+ => node v10+ (an increase)
Each of these adjustments was identified by comparing against data from the
node-compat-table
package and was manually verified using old node executables downloaded from https://nodejs.org/download/release/. -
v0.17.10
-
Update esbuild's handling of CSS nesting to match the latest specification changes (#1945)
The syntax for the upcoming CSS nesting feature has recently changed. The
@nest
prefix that was previously required in some cases is now gone, and nested rules no longer have to start with&
(as long as they don't start with an identifier or function token).This release updates esbuild's pass-through handling of CSS nesting syntax to match the latest specification changes. So you can now use esbuild to bundle CSS containing nested rules and try them out in a browser that supports CSS nesting (which includes nightly builds of both Chrome and Safari).
However, I'm not implementing lowering of nested CSS to non-nested CSS for older browsers yet. While the syntax has been decided, the semantics are still in flux. In particular, there is still some debate about changing the fundamental way that CSS nesting works. For example, you might think that the following CSS is equivalent to a
.outer .inner button { ... }
rule:.inner button { .outer & { color: red; } }
But instead it's actually equivalent to a
.outer :is(.inner button) { ... }
rule which unintuitively also matches the following DOM structure:<div class="inner"> <div class="outer"> <button></button> </div> </div>
The
:is()
behavior is preferred by browser implementers because it's more memory-efficient, but the straightforward translation into a.outer .inner button { ... }
rule is preferred by developers used to the existing CSS preprocessing ecosystem (e.g. SASS). It seems premature to commit esbuild to specific semantics for this syntax at this time given the ongoing debate. -
Fix cross-file CSS rule deduplication involving
url()
tokens (#2936)Previously cross-file CSS rule deduplication didn't handle
url()
tokens correctly. These tokens contain references to import paths which may be internal (i.e. in the bundle) or external (i.e. not in the bundle). When comparing twourl()
tokens for equality, the underlying import paths should be compared instead of their references. This release of esbuild fixesurl()
token comparisons. One side effect is that@font-face
rules should now be deduplicated correctly across files:/* Original code */ @​import "data:text/css, \ @​import 'http://example.com/style.css'; \ @​font-face { src: url(http://example.com/font.ttf) }"; @​import "data:text/css, \ @​font-face { src: url(http://example.com/font.ttf) }"; /* Old output (with --bundle --minify) */ @​import"http://example.com/style.css";@​font-face{src:url(http://example.com/font.ttf)}@​font-face{src:url(http://example.com/font.ttf)} /* New output (with --bundle --minify) */ @​import"http://example.com/style.css";@​font-face{src:url(http://example.com/font.ttf)}
v0.17.9
-
Parse rest bindings in TypeScript types (#2937)
Previously esbuild was unable to parse the following valid TypeScript code:
let tuple: (...[e1, e2, ...es]: any) => any
This release includes support for parsing code like this.
-
Fix TypeScript code translation for certain computed
declare
class fields (#2914)In TypeScript, the key of a computed
declare
class field should only be preserved if there are no decorators for that field. Previously esbuild always preserved the key, but esbuild will now remove the key to match the output of the TypeScript compiler:// Original code declare function dec(a: any, b: any): any declare const removeMe: unique symbol declare const keepMe: unique symbol class X { declare [removeMe]: any @​dec declare [keepMe]: any } // Old output var _a; class X { } removeMe, _a = keepMe; __decorateClass([ dec ], X.prototype, _a, 2); // New output var _a; class X { } _a = keepMe; __decorateClass([ dec ], X.prototype, _a, 2);
-
Fix a crash with path resolution error generation (#2913)
In certain situations, a module containing an invalid import path could previously cause esbuild to crash when it attempts to generate a more helpful error message. This crash has been fixed.
v0.17.8
-
Fix a minification bug with non-ASCII identifiers (#2910)
This release fixes a bug with esbuild where non-ASCII identifiers followed by a keyword were incorrectly not separated by a space. This bug affected both the
in
andinstanceof
keywords. Here's an example of the fix:// Original code π in a // Old output (with --minify --charset=utf8) πin a; // New output (with --minify --charset=utf8) π in a;
-
Fix a regression with esbuild's WebAssembly API in version 0.17.6 (#2911)
Version 0.17.6 of esbuild updated the Go toolchain to version 1.20.0. This had the unfortunate side effect of increasing the amount of stack space that esbuild uses (presumably due to some changes to Go's WebAssembly implementation) which could cause esbuild's WebAssembly-based API to crash with a stack overflow in cases where it previously didn't crash. One such case is the package
grapheme-splitter
which contains code that looks like this:if ( (0x0300 <= code && code <= 0x036F) || (0x0483 <= code && code <= 0x0487) || (0x0488 <= code && code <= 0x0489) || (0x0591 <= code && code <= 0x05BD) || // ... many hundreds of lines later ... ) { return; }
This edge case involves a chain of binary operators that results in an AST over 400 nodes deep. Normally this wouldn't be a problem because Go has growable call stacks, so the call stack would just grow to be as large as needed. However, WebAssembly byte code deliberately doesn't expose the ability to manipulate the stack pointer, so Go's WebAssembly translation is forced to use the fixed-size WebAssembly call stack. So esbuild's WebAssembly implementation is vulnerable to stack overflow in cases like these.
It's not unreasonable for this to cause a stack overflow, and for esbuild's answer to this problem to be "don't write code like this." That's how many other AST-manipulation tools handle this problem. However, it's possible to implement AST traversal using iteration instead of recursion to work around limited call stack space. This version of esbuild implements this code transformation for esbuild's JavaScript parser and printer, so esbuild's WebAssembly implementation is now able to process the
grapheme-splitter
package (at least when compiled with Go 1.20.0 and run with node's WebAssembly implementation).
v0.17.7
-
Change esbuild's parsing of TypeScript instantiation expressions to match TypeScript 4.8+ (#2907)
This release updates esbuild's implementation of instantiation expression erasure to match microsoft/TypeScript#49353. The new rules are as follows (copied from TypeScript's MR description):
When a potential type argument list is followed by
- a line break,
- an
(
token, - a template literal string, or
- any token except
<
or>
that isn't the start of an expression,
we consider that construct to be a type argument list. Otherwise we consider the construct to be a
<
relational expression followed by a>
relational expression. -
Ignore
sideEffects: false
for imported CSS files (#1370, #1458, #2905)This release ignores the
sideEffects
annotation inpackage.json
for CSS files that are imported into JS files using esbuild'scss
loader. This means that these CSS files are no longer be tree-shaken.Importing CSS into JS causes esbuild to automatically create a CSS entry point next to the JS entry point containing the bundled CSS. Previously packages that specified some form of
"sideEffects": false
could potentially cause esbuild to consider one or more of the JS files on the import path to the CSS file to be side-effect free, which would result in esbuild removing that CSS file from the bundle. This was problematic because the removal of that CSS is outwardly observable, since all CSS is global, so it was incorrect for previous versions of esbuild to tree-shake CSS files imported into JS files. -
Add constant folding for certain additional equality cases (#2394, #2895)
This release adds constant folding for expressions similar to the following:
// Original input console.log( null === 'foo', null === undefined, null == undefined, false === 0, false == 0, 1 === true, 1 == true, ) // Old output console.log( null === "foo", null === void 0, null == void 0, false === 0, false == 0, 1 === true, 1 == true ); // New output console.log( false, false, true, false, true, false, true );
v0.17.6
-
Fix a CSS parser crash on invalid CSS (#2892)
Previously the following invalid CSS caused esbuild's parser to crash:
@​media screen
The crash was caused by trying to construct a helpful error message assuming that there was an opening
{
token, which is not the case here. This release fixes the crash. -
Inline TypeScript enums that are referenced before their declaration
Previously esbuild inlined enums within a TypeScript file from top to bottom, which meant that references to TypeScript enum members were only inlined within the same file if they came after the enum declaration. With this release, esbuild will now inline enums even when they are referenced before they are declared:
// Original input export const foo = () => Foo.FOO const enum Foo { FOO = 0 } // Old output (with --tree-shaking=true) export const foo = () => Foo.FOO; var Foo = /* @​__PURE__ */ ((Foo2) => { Foo2[Foo2["FOO"] = 0] = "FOO"; return Foo2; })(Foo || {}); // New output (with --tree-shaking=true) export const foo = () => 0 /* FOO */;
This makes esbuild's TypeScript output smaller and faster when processing code that does this. I noticed this issue when I ran the TypeScript compiler's source code through esbuild's bundler. Now that the TypeScript compiler is going to be bundled with esbuild in the upcoming TypeScript 5.0 release, improvements like this will also improve the TypeScript compiler itself!
-
Fix esbuild installation on Arch Linux (#2785, #2812, #2865)
Someone made an unofficial
esbuild
package for Linux that adds theESBUILD_BINARY_PATH=/usr/bin/esbuild
environment variable to the user's default environment. This breaks all npm installations of esbuild for users with this unofficial Linux package installed, which has affected many people. Most (all?) people who encounter this problem haven't even installed this unofficial package themselves; instead it was installed for them as a dependency of another Linux package. The problematic change to add theESBUILD_BINARY_PATH
environment variable was reverted in the latest version of this unofficial package. However, old versions of this unofficial package are still there and will be around forever. With this release,ESBUILD_BINARY_PATH
is now ignored by esbuild's install script when it's set to the value/usr/bin/esbuild
. This should unbreak using npm to installesbuild
in these problematic Linux environments.Note: The
ESBUILD_BINARY_PATH
variable is an undocumented way to override the location of esbuild's binary when esbuild's npm package is installed, which is necessary to substitute your own locally-built esbuild binary when debugging esbuild's npm package. It's only meant for very custom situations and should absolutely not be forced on others by default, especially without their knowledge. I may remove the code in esbuild's installer that readsESBUILD_BINARY_PATH
in the future to prevent these kinds of issues. It will unfortunately make debugging esbuild harder. IfESBUILD_BINARY_PATH
is ever removed, it will be done in a "breaking change" release.
v0.17.5
-
Parse
const
type parameters from TypeScript 5.0The TypeScript 5.0 beta announcement adds
const
type parameters to the language. You can now add theconst
modifier on a type parameter of a function, method, or class like this:type HasNames = { names: readonly string[] }; const getNamesExactly = <const T extends HasNames>(arg: T): T["names"] => arg.names; const names = getNamesExactly({ names: ["Alice", "Bob", "Eve"] });
The type of
names
in the above example isreadonly ["Alice", "Bob", "Eve"]
. Marking the type parameter asconst
behaves as if you had writtenas const
at every use instead. The above code is equivalent to the following TypeScript, which was the only option before TypeScript 5.0:type HasNames = { names: readonly string[] }; const getNamesExactly = <T extends HasNames>(arg: T): T["names"] => arg.names; const names = getNamesExactly({ names: ["Alice", "Bob", "Eve"] } as const);
You can read the announcement for more information.
-
Make parsing generic
async
arrow functions more strict in.tsx
filesPreviously esbuild's TypeScript parser incorrectly accepted the following code as valid:
let fn = async <T> () => {};
The official TypeScript parser rejects this code because it thinks it's the identifier
async
followed by a JSX element starting with<T>
. So with this release, esbuild will now reject this syntax in.tsx
files too. You'll now have to add a comma after the type parameter to get generic arrow functions like this to parse in.tsx
files:let fn = async <T,> () => {};
-
Allow the
in
andout
type parameter modifiers on class expressionsTypeScript 4.7 added the
in
andout
modifiers on the type parameters of classes, interfaces, and type aliases. However, while TypeScript supported them on both class expressions and class statements, previously esbuild only supported them on class statements due to an oversight. This release now allows these modifiers on class expressions too:declare let Foo: any; Foo = class <in T> { }; Foo = class <out T> { };
-
Update
enum
constant folding for TypeScript 5.0TypeScript 5.0 contains an updated definition of what it considers a constant expression:
An expression is considered a constant expression if it is
- a number or string literal,
- a unary
+
,-
, or~
applied to a numeric constant expression, - a binary
+
,-
,*
,/
,%
,**
,<<
,>>
,>>>
,|
,&
,^
applied to two numeric constant expressions, - a binary
+
applied to two constant expressions whereof at least one is a string, - a template expression where each substitution expression is a constant expression,
- a parenthesized constant expression,
- a dotted name (e.g.
x.y.z
) that references aconst
variable with a constant expression initializer and no type annotation, - a dotted name that references an enum member with an enum literal type, or
- a dotted name indexed by a string literal (e.g.
x.y["z"]
) that references an enum member with an enum literal type.
This impacts esbuild's implementation of TypeScript's
const enum
feature. With this release, esbuild will now attempt to follow these new rules. For example, you can now initialize anenum
member with a template literal expression that contains a numeric constant:// Original input const enum Example { COUNT = 100, ERROR = `Expected ${COUNT} items`, } console.log( Example.COUNT, Example.ERROR, ) // Old output (with --tree-shaking=true) var Example = /* @​__PURE__ */ ((Example2) => { Example2[Example2["COUNT"] = 100] = "COUNT"; Example2[Example2["ERROR"] = `Expected ${100 /* COUNT */} items`] = "ERROR"; return Example2; })(Example || {}); console.log( 100 /* COUNT */, Example.ERROR ); // New output (with --tree-shaking=true) console.log( 100 /* COUNT */, "Expected 100 items" /* ERROR */ );
These rules are not followed exactly due to esbuild's limitations. The rule about dotted references to
const
variables is not followed both because esbuild's enum processing is done in an isolated module setting and because doing so would potentially require esbuild to use a type system, which it doesn't have. For example:// The TypeScript compiler inlines this but esbuild doesn't: declare const x = 'foo' const enum Foo { X = x } console.log(Foo.X)
Also, the rule that requires converting numbers to a string currently only followed for 32-bit signed integers and non-finite numbers. This is done to avoid accidentally introducing a bug if esbuild's number-to-string operation doesn't exactly match the behavior of a real JavaScript VM. Currently esbuild's number-to-string constant folding is conservative for safety.
-
Forbid definite assignment assertion operators on class methods
In TypeScript, class methods can use the
?
optional property operator but not the!
definite assignment assertion operator (while class fields can use both):class Foo { // These are valid TypeScript a? b! x?() {} // This is invalid TypeScript y!() {} }
Previously esbuild incorrectly allowed the definite assignment assertion operator with class methods. This will no longer be allowed starting with this release.
v0.17.4
-
Implement HTTP
HEAD
requests in serve mode (#2851)Previously esbuild's serve mode only responded to HTTP
GET
requests. With this release, esbuild's serve mode will also respond to HTTPHEAD
requests, which are just like HTTPGET
requests except that the body of the response is omitted. -
Permit top-level await in dead code branches (#2853)
Adding top-level await to a file has a few consequences with esbuild:
- It causes esbuild to assume that the input module format is ESM, since top-level await is only syntactically valid in ESM. That prevents you from using
module
andexports
for exports and also enables strict mode, which disables certain syntax and changes how function hoisting works (among other things). - This will cause esbuild to fail the build if either top-level await isn't supported by your language target (e.g. it's not supported in ES2021) or if top-level await isn't supported by the chosen output format (e.g. it's not supported with CommonJS).
- Doing this will prevent you from using
require()
on this file or on any file that imports this file (even indirectly), since therequire()
function doesn't return a promise and so can't represent top-level await.
This release relaxes these rules slightly: rules 2 and 3 will now no longer apply when esbuild has identified the code branch as dead code, such as when it's behind an
if (false)
check. This should make it possible to use esbuild to convert code into different output formats that only uses top-level await conditionally. This release does not relax rule 1. Top-level await will still cause esbuild to unconditionally consider the input module format to be ESM, even when the top-levelawait
is in a dead code branch. This is necessary because whether the input format is ESM or not affects the whole file, not just the dead code branch. - It causes esbuild to assume that the input module format is ESM, since top-level await is only syntactically valid in ESM. That prevents you from using
-
Fix entry points where the entire file name is the extension (#2861)
Previously if you passed esbuild an entry point where the file extension is the entire file name, esbuild would use the parent directory name to derive the name of the output file. For example, if you passed esbuild a file
./src/.ts
then the output name would besrc.js
. This bug happened because esbuild first strips the file extension to get./src/
and then joins the path with the working directory to get the absolute path (e.g.join("/working/dir", "./src/")
gives/working/dir/src
). However, the join operation also canonicalizes the path which strips the trailing/
. Later esbuild uses the "base name" operation to extract the name of the output file. Since there is no trailing/
, esbuild returns"src"
as the base name instead of""
, which causes esbuild to incorrectly include the directory name in the output file name. This release fixes this bug by deferring the stripping of the file extension until after all path manipulations have been completed. So now the file./src/.ts
will generate an output file named.js
. -
Support replacing property access expressions with inject
At a high level, this change means the
inject
feature can now replace all of the same kinds of names as thedefine
feature. Soinject
is basically now a more powerful version ofdefine
, instead of previously only being able to do some of the things thatdefine
could do.Soem background is necessary to understand this change if you aren't already familiar with the
inject
feature. Theinject
feature lets you replace references to global variable with a shim. It works like this:- Put the shim in its own file
- Export the shim as the name of the global variable you intend to replace
- Pass the file to esbuild using the
inject
feature
For example, if you inject the following file using
--inject:./injected.js
:// injected.js let processShim = { cwd: () => '/' } export { processShim as process }
Then esbuild will replace all references to
process
with theprocessShim
variable, which will causeprocess.cwd()
to return'/'
. This feature is sort of abusing the ESM export alias syntax to specify the mapping of global variables to shims. But esbuild works this way because using this syntax for that purpose is convenient and terse.However, if you wanted to replace a property access expression, the process was more complicated and not as nice. You would have to:
- Put the shim in its own file
- Export the shim as some random name
- Pass the file to esbuild using the
inject
feature - Use esbuild's
define
feature to map the property access expression to the random name you made in step 2
For example, if you inject the following file using
--inject:./injected2.js --define:process.cwd=someRandomName
:// injected2.js let cwdShim = () => '/' export { cwdShim as someRandomName }
Then esbuild will replace all references to
process.cwd
with thecwdShim
variable, which will also causeprocess.cwd()
to return'/'
(but which this time will not mess with other references toprocess
, which might be desirable).With this release, using the inject feature to replace a property access expression is now as simple as using it to replace an identifier. You can now use JavaScript's "arbitrary module namespace identifier names" feature to specify the property access expression directly using a string literal. For example, if you inject the following file using
--inject:./injected3.js
:// injected3.js let cwdShim = () => '/' export { cwdShim as 'process.cwd' }
Then esbuild will now replace all references to
process.cwd
with thecwdShim
variable, which will also causeprocess.cwd()
to return'/'
(but which will also not mess with other references toprocess
).In addition to inserting a shim for a global variable that doesn't exist, another use case is replacing references to static methods on global objects with cached versions to both minify them better and to make access to them potentially faster. For example:
// Injected file let cachedMin = Math.min let cachedMax = Math.max export { cachedMin as 'Math.min', cachedMax as 'Math.max', } // Original input function clampRGB(r, g, b) { return { r: Math.max(0, Math.min(1, r)), g: Math.max(0, Math.min(1, g)), b: Math.max(0, Math.min(1, b)), } } // Old output (with --minify) function clampRGB(a,t,m){return{r:Math.max(0,Math.min(1,a)),g:Math.max(0,Math.min(1,t)),b:Math.max(0,Math.min(1,m))}} // New output (with --minify) var a=Math.min,t=Math.max;function clampRGB(h,M,m){return{r:t(0,a(1,h)),g:t(0,a(1,M)),b:t(0,a(1,m))}}
v0.17.3
-
Fix incorrect CSS minification for certain rules (#2838)
Certain rules such as
@media
could previously be minified incorrectly. Due to a typo in the duplicate rule checker, two known@
-rules that share the same hash code were incorrectly considered to be equal. This problem was made worse by the rule hashing code considering two unknown declarations (such as CSS variables) to have the same hash code, which also isn't optimal from a performance perspective. Both of these issues have been fixed:/* Original input */ @​media (prefers-color-scheme: dark) { body { --VAR-1: #​000; } } @​media (prefers-color-scheme: dark) { body { --VAR-2: #​000; } } /* Old output (with --minify) */ @​media (prefers-color-scheme: dark){body{--VAR-2: #​000}} /* New output (with --minify) */ @​media (prefers-color-scheme: dark){body{--VAR-1: #​000}}@​media (prefers-color-scheme: dark){body{--VAR-2: #​000}}
v0.17.2
-
Add
onDispose
to the plugin API (#2140, #2205)If your plugin wants to perform some cleanup after it's no longer going to be used, you can now use the
onDispose
API to register a callback for cleanup-related tasks. For example, if a plugin starts a long-running child process then it may want to terminate that process when the plugin is discarded. Previously there was no way to do this. Here's an example:let examplePlugin = { name: 'example', setup(build) { build.onDispose(() => { console.log('This plugin is no longer used') }) }, }
These
onDispose
callbacks will be called after everybuild()
call regardless of whether the build failed or not as well as after the firstdispose()
call on a given build context.
v0.17.1
-
Fix CSS transform bugs with nested selectors that start with a combinator (#3096)
This release fixes several bugs regarding transforming nested CSS into non-nested CSS for older browsers. The bugs were due to lack of test coverage for nested selectors with more than one compound selector where they all start with the same combinator. Here's what some problematic cases look like before and after these fixes:
/* Original code */ .foo { > &a, > &b { color: red; } } .bar { > &a, + &b { color: green; } } /* Old output (with --target=chrome90) */ .foo :is(> .fooa, > .foob) { color: red; } .bar :is(> .bara, + .barb) { color: green; } /* New output (with --target=chrome90) */ .foo > :is(a.foo, b.foo) { color: red; } .bar > a.bar, .bar + b.bar { color: green; }
-
Fix bug with TypeScript parsing of instantiation expressions followed by
=
(#3111)This release fixes esbuild's TypeScript-to-JavaScript conversion code in the case where a potential instantiation expression is followed immediately by a
=
token (such that the trailing>
becomes a>=
token). Previously esbuild considered that to still be an instantiation expression, but the official TypeScript compiler considered it to be a>=
operator instead. This release changes esbuild's interpretation to match TypeScript. This edge case currently appears to be problematic for other TypeScript-to-JavaScript converters as well:Original code TypeScript esbuild 0.17.18 esbuild 0.17.19 Sucrase Babel x<y>=a<b<c>>()
x<y>=a();
x=a();
x<y>=a();
x=a()
Invalid left-hand side in assignment expression -
Avoid removing unrecognized directives from the directive prologue when minifying (#3115)
The directive prologue in JavaScript is a sequence of top-level string expressions that come before your code. The only directives that JavaScript engines currently recognize are
use strict
and sometimesuse asm
. However, the people behind React have made up their own directive for their own custom dialect of JavaScript. Previously esbuild only preserved theuse strict
directive when minifying, although you could still write React JavaScript with esbuild using something like--banner:js="'your directive here';"
. With this release, you can now put arbitrary directives in the entry point and esbuild will preserve them in its minified output:// Original code 'use wtf'; console.log(123) // Old output (with --minify) console.log(123); // New output (with --minify) "use wtf";console.log(123);
Note that this means esbuild will no longer remove certain stray top-level strings when minifying. This behavior is an intentional change because these stray top-level strings are actually part of the directive prologue, and could potentially have semantics assigned to them (as was the case with React).
-
Improved minification of binary shift operators
With this release, esbuild's minifier will now evaluate the
<<
and>>>
operators if the resulting code would be shorter:// Original code console.log(10 << 10, 10 << 20, -123 >>> 5, -123 >>> 10); // Old output (with --minify) console.log(10<<10,10<<20,-123>>>5,-123>>>10); // New output (with --minify) console.log(10240,10<<20,-123>>>5,4194303);
Configuration
-
If you want to rebase/retry this MR, check this box
This MR has been generated by Renovate Bot.