clsx

Patches

• (types) Include bigint in ClassValue type: (#96): 3d960ab Accommodates recent @types/react changes to ReactNode. Thank you @ViliamKopecky~!

Chores

• Add licenses.dev badge: 684509c This service recursively analyzes entire dependency graphs to ensure that a package (or your project) is using permissive licenses. For example, here's a results table for polka@next and a larger astro example.

---

Full Changelog: https://github.com/lukeed/clsx/compare/v2.1.0...v2.1.1

Features

• Add new clsx/lite submodule for string-only usage: 1a49142

  This is a 140b version of clsx that is ideal for Tailwind and/or React contexts, which typically follow this clsx usage pattern:

  clsx('foo bar', props.maybe && 'conditional classes', props.className);
  

  Important: This clsx/lite module ignores all non-string arguments and is therefore not a 1:1 replacement for clsx itself!

  import { clsx } from 'clsx';
  import { clsx as lite } from 'clsx/lite';
  
  // strings-only usage is identical
  clsx('foo', null, 'bar', true && 'baz'); //=> "foo bar baz"
  lite('foo', null, 'bar', true && 'baz'); //=> "foo bar baz"
  
  // clsx/lite ignores all non-strings
  clsx('foo', { a: true, b: false, c: true }); //=> "foo a c"
  lite('foo', { a: true, b: false, c: true }); //=> "foo"
  

---

Full Changelog: https://github.com/lukeed/clsx/compare/v2.0.1...v2.1.0

Patches

• (perf) Cache arguments.length & array.length for 6% performance gain (#26): deff09b Adds 5 bytes (+2%) for ~3% avg performance gain Thank you @jalalazimi

Chores

• Update module size: bf64e71
• Update benchmark results: 855eec2, 6e3b2b9,
• Replace nyc with c8 in CI: 6e2468e
• Update Node CI matrix: 308a238
• Fix readme typos (#76, #82): 42354d3, 4c9a55d Thank you @andipaetzold and @acusti

---

Full Changelog: https://github.com/lukeed/clsx/compare/v2.0.0...v2.0.1

Breaking

• Add "exports" map for native ESM support (#57): 3ec8e9f, 0949868 Also supports TypeScript's node16/nodenext module resolution Maintains CommonJS support (with fixed TS definitions) Thank you @remcohaszing~!

Chores

• Add Tailwind Support section to README (#65, #68): 496db1d, 4a4eadd Thank you @kevinlowe0x3F7 & @kevinlowe0x3F7
• Add tests for numbers & variadic number arguments: c520353
• Update package module size: 03e1cf9

---

Full Changelog: https://github.com/lukeed/clsx/compare/v1.2.1...v2.0.0

Patches

• Ensure CommonJS and UMD entrypoints have the named clsx export too

Chores

• Build CJS & UMD files manually (#50): 3712966, 2114f5b

---

Full Changelog: https://github.com/lukeed/clsx/compare/v1.2.0...v1.2.1

Features

• Add named clsx export alias (#43, #44): 56ab81f Thank you @danikaze~! This is purely an alias for the default export so that TypeScript users can avoid the esModuleInterop setting. In other words, the follow import statements are effectively identical, but the latter is preferred by TypeScript:

  import clsx from 'clsx';
  // or
  import { clsx } from 'clsx';
  

  Important: Just to reiterate, both still work!

Chores

• Migrate from tape to uvu for tests: 1c36d10
• Add Node 14 and Node 16 to CI test matrix: bc4f827
• Update "string variadic" tests (#30): 74cefa6 Thank you @gingerrific~!
• Enable GitHub Sponsors button: 400b848

---

Full Changelog: https://github.com/lukeed/clsx/compare/v1.1.1...v1.2.0

Note: This is a performance-related patch only!Across all benchmarks, this version of clsx is ~1M ops/sec faster than clsx@1.1.0....It also happens to be 1 byte (gzip) smaller 😅🎉

Patches

• fix: Remove needless spacer on string/number condition: ff11464
• fix: Remove unnecessary recursive caller for object keys: f43dd23
• perf: Guard all toVal calls with truthy assertions: 4fa8811, 019ec02
• perf: Reorder typeof checks for common case: 08a5a7f

Chores

• Run CI via GitHub Actions: 64af363, 883363b, 866620d, 768ffab, 5e92fdf
• Report code coverage within CI script: 0fb6d98
• Use esm hook with tests: 0287e40
• Update module size (-1 byte): 7e8c75e
• Update benchmark results: 75053c0
• Lock all dev-deps: 3c6aa3f
• Update badges: fd1533f

Benchmarks

Run on Node.js v10.13.0.You may find updated browser benchmarks here.

The snippet below is comparing clsx@1.1.1 (this version) to the previous version, and to classnames for ballpark comparison. All candidates are functionally identical!

# Strings
  classnames    x  3,992,284 ops/sec ±1.64% (94 runs sampled)
  clsx@1.1.0    x 11,253,372 ops/sec ±0.35% (96 runs sampled)
  clsx@1.1.1    x 12,784,134 ops/sec ±0.42% (97 runs sampled)

# Objects
  classnames    x 3,772,978 ops/sec ±0.46% (96 runs sampled)
  clsx@1.1.0    x 7,288,178 ops/sec ±0.31% (96 runs sampled)
  clsx@1.1.1    x 9,412,010 ops/sec ±0.42% (95 runs sampled)

# Arrays
  classnames    x 1,665,275 ops/sec ±1.83% (93 runs sampled)
  clsx@1.1.0    x 8,340,174 ops/sec ±0.53% (96 runs sampled)
  clsx@1.1.1    x 9,141,916 ops/sec ±0.42% (95 runs sampled)

# Nested Arrays
  classnames    x 1,164,706 ops/sec ±1.60% (95 runs sampled)
  clsx@1.1.0    x 6,284,485 ops/sec ±0.58% (90 runs sampled)
  clsx@1.1.1    x 7,165,151 ops/sec ±0.47% (91 runs sampled)

# Nested Arrays w/ Objects
  classnames    x 1,597,180 ops/sec ±1.49% (93 runs sampled)
  clsx@1.1.0    x 6,345,248 ops/sec ±0.21% (95 runs sampled)
  clsx@1.1.1    x 7,651,411 ops/sec ±0.56% (95 runs sampled)

# Mixed
  classnames    x 2,129,199 ops/sec ±1.46% (94 runs sampled)
  clsx@1.1.0    x 6,557,515 ops/sec ±0.73% (91 runs sampled)
  clsx@1.1.1    x 8,119,210 ops/sec ±0.42% (93 runs sampled)

# Mixed (Bad Data)
  classnames    x 1,166,577 ops/sec ±0.84% (94 runs sampled)
  clsx@1.1.0    x 2,018,654 ops/sec ±0.15% (98 runs sampled)
  clsx@1.1.1    x 2,238,939 ops/sec ±0.34% (95 runs sampled)

Minor / Patch

• Use Array.isArray for proper Array type-checking (#17): 0d6023a

  This prevents an object with a truthy push key from being treated as an Array. While (likely) rare, this produced an incorrect result.

  Important: Because of this change, Internet Explorer 8 (and older) are no longer supported.If you happen to require support for these browsers, please continue to use the 1.0.x version range.

Chores

• Updated benchmark results: 3cfb6c3

  Because Array.isArray is highly optimized in all modern JS engines, we received a bit of a speed boost!

  Below is clsx@1.0.4 versus this release, running in Node.js 10.13.0

  # Strings
    clsx (prev)  x 10,336,101 ops/sec ±1.45% (90 runs sampled)
    clsx         x 11,191,101 ops/sec ±0.51% (97 runs sampled)
  
  # Objects
    clsx (prev)  x 7,195,049 ops/sec ±0.35% (92 runs sampled)
    clsx         x 7,133,215 ops/sec ±0.31% (93 runs sampled)
  
  # Arrays
    clsx (prev)  x 8,265,223 ops/sec ±0.32% (94 runs sampled)
    clsx         x 8,346,704 ops/sec ±0.29% (98 runs sampled)
  
  # Nested Arrays
    clsx (prev)  x 6,244,513 ops/sec ±1.41% (94 runs sampled)
    clsx         x 6,324,445 ops/sec ±0.38% (95 runs sampled)
  
  # Nested Arrays w/ Objects
    clsx (prev)  x 5,157,630 ops/sec ±0.34% (92 runs sampled)
    clsx         x 6,349,877 ops/sec ±0.26% (97 runs sampled)
  
  # Mixed
    clsx (prev)  x 5,751,180 ops/sec ±0.38% (95 runs sampled)
    clsx         x 6,722,160 ops/sec ±0.36% (98 runs sampled)
  
  # Mixed (Bad Data)
    clsx (prev)  x 1,935,973 ops/sec ±0.19% (97 runs sampled)
    clsx         x 2,053,230 ops/sec ±0.27% (95 runs sampled)
  

Patches

• Ignore function arguments: 4aff569

• Separate Array and Object loops: 91f5e68

  While this did add ~20 bytes to the package (which is a 10% size increase), it made any Array interactions 200% to 600% faster! 🎉 Below is a performance comparison between clsx@1.0.3 and clsx@1.0.4 running on Node v10.13.0:

  # Strings
    clsx (prev)  x 10,661,038 ops/sec ±0.46% (92 runs sampled)
    clsx         x 11,066,632 ops/sec ±0.23% (96 runs sampled)
  
  # Objects
    clsx (prev)  x 6,934,269 ops/sec ±0.23% (94 runs sampled)
    clsx         x 7,147,168 ops/sec ±0.30% (95 runs sampled)
  
  # Arrays
    clsx (prev)  x 1,681,568 ops/sec ±0.32% (97 runs sampled)
    clsx         x 7,922,225 ops/sec ±0.18% (95 runs sampled)
  
  # Nested Arrays
    clsx (prev)  x   940,741 ops/sec ±0.33% (96 runs sampled)
    clsx         x 6,132,222 ops/sec ±0.49% (95 runs sampled)
  
  # Nested Arrays w/ Objects
    clsx (prev)  x 1,428,975 ops/sec ±0.15% (95 runs sampled)
    clsx         x 5,156,844 ops/sec ±0.22% (95 runs sampled)
  
  # Mixed
    clsx (prev)  x 2,198,907 ops/sec ±0.34% (98 runs sampled)
    clsx         x 5,577,715 ops/sec ±0.20% (93 runs sampled)
  
  # Mixed (Bad Data)
    clsx (prev)  x 1,038,752 ops/sec ±0.32% (94 runs sampled)
    clsx         x 1,887,010 ops/sec ±0.14% (96 runs sampled)
  

Chores

• Include Node.js benchmarks: b7c3973
• Update benchmark results: 8d5a2f7, e85c5be
• Fix README typos: ead41a3, 4a4413d
• Update size: a0fd98a

Patches

• Changed "module" entry file to .js extension (#7): 29b2fce

  Unfortunately, webpack changed how they treat .mjs files somewhat recently. This is a huge step backwards and hopefully they revert this change soon. See https://github.com/dmtrKovalenko/material-ui-pickers/issues/894