Beware of inaccurate benchmark results about classnames

[dcousens]

Welcome to hot spotting 👍

[victor-homyakov]

Some results from real browsers (again, with unchanged local version), if anyone is interested

Chrome 53 (Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/53.0.2785.116 Safari/537.36):

* local#strings x 3,457,677 ops/sec ±1.44% (98 runs sampled)
*   npm#strings x 3,531,513 ops/sec ±0.21% (97 runs sampled)
* local/dedupe#strings x 791,514 ops/sec ±1.01% (98 runs sampled)
*   npm/dedupe#strings x 779,305 ops/sec ±0.40% (95 runs sampled)

> Fastest is npm#strings

* local#object x 2,379,320 ops/sec ±0.23% (99 runs sampled)
*   npm#object x 3,400,455 ops/sec ±0.44% (97 runs sampled)
* local/dedupe#object x 1,942,161 ops/sec ±0.28% (97 runs sampled)
*   npm/dedupe#object x 1,918,508 ops/sec ±0.40% (96 runs sampled)

> Fastest is npm#object

* local#strings, object x 2,347,997 ops/sec ±0.64% (99 runs sampled)
*   npm#strings, object x 3,093,171 ops/sec ±0.13% (96 runs sampled)
* local/dedupe#strings, object x 872,712 ops/sec ±0.23% (101 runs sampled)
*   npm/dedupe#strings, object x 804,977 ops/sec ±1.34% (93 runs sampled)

> Fastest is npm#strings, object

* local#mix x 1,418,466 ops/sec ±0.38% (96 runs sampled)
*   npm#mix x 2,194,682 ops/sec ±0.40% (97 runs sampled)
* local/dedupe#mix x 452,717 ops/sec ±0.19% (99 runs sampled)
*   npm/dedupe#mix x 448,950 ops/sec ±0.17% (98 runs sampled)

> Fastest is npm#mix

* local#arrays x 663,922 ops/sec ±0.15% (102 runs sampled)
*   npm#arrays x 766,503 ops/sec ±0.16% (102 runs sampled)
* local/dedupe#arrays x 437,192 ops/sec ±1.09% (97 runs sampled)
*   npm/dedupe#arrays x 447,156 ops/sec ±0.25% (98 runs sampled)

> Fastest is npm#arrays

Firefox 47 (Mozilla/5.0 (Windows NT 6.1; WOW64; rv:47.0) Gecko/20100101 Firefox/47.0):

* local#strings x 2,176,074 ops/sec ±1.00% (94 runs sampled)
*   npm#strings x 2,177,106 ops/sec ±1.17% (94 runs sampled)
* local/dedupe#strings x 391,829 ops/sec ±3.08% (93 runs sampled)
*   npm/dedupe#strings x 399,194 ops/sec ±1.15% (89 runs sampled)

> Fastest is local#strings |   npm#strings

* local#object x 1,763,326 ops/sec ±0.95% (98 runs sampled)
*   npm#object x 1,675,736 ops/sec ±1.51% (85 runs sampled)
* local/dedupe#object x 1,369,897 ops/sec ±0.70% (94 runs sampled)
*   npm/dedupe#object x 1,378,539 ops/sec ±0.59% (95 runs sampled)

> Fastest is local#object

* local#strings, object x 1,318,376 ops/sec ±1.09% (94 runs sampled)
*   npm#strings, object x 1,337,806 ops/sec ±0.99% (91 runs sampled)
* local/dedupe#strings, object x 431,453 ops/sec ±0.63% (93 runs sampled)
*   npm/dedupe#strings, object x 455,738 ops/sec ±0.43% (96 runs sampled)

> Fastest is npm#strings, object

* local#mix x 648,665 ops/sec ±4.03% (92 runs sampled)
*   npm#mix x 667,840 ops/sec ±0.93% (90 runs sampled)
* local/dedupe#mix x 177,309 ops/sec ±2.00% (86 runs sampled)
*   npm/dedupe#mix x 174,356 ops/sec ±2.72% (83 runs sampled)

> Fastest is npm#mix

* local#arrays x 148,257 ops/sec ±2.45% (86 runs sampled)
*   npm#arrays x 150,536 ops/sec ±1.07% (93 runs sampled)
* local/dedupe#arrays x 190,409 ops/sec ±0.27% (96 runs sampled)
*   npm/dedupe#arrays x 179,243 ops/sec ±1.00% (96 runs sampled)

> Fastest is local/dedupe#arrays

IE11 (Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; SLCC2; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30729; Media Center PC 6.0; .NET4.0C; .NET4.0E; rv:11.0) like Gecko):

* local#strings x 1,609,608 ops/sec ±0.95% (85 runs sampled)
* npm#strings x 1,605,186 ops/sec ±0.99% (95 runs sampled)
* local/dedupe#strings x 286,420 ops/sec ±1.17% (93 runs sampled)
* npm/dedupe#strings x 288,187 ops/sec ±1.36% (93 runs sampled)

> Fastest is local#strings | npm#strings

* local#object x 886,217 ops/sec ±1.14% (95 runs sampled)
* npm#object x 892,052 ops/sec ±1.12% (96 runs sampled)
* local/dedupe#object x 346,228 ops/sec ±1.16% (98 runs sampled)
* npm/dedupe#object x 346,862 ops/sec ±1.05% (97 runs sampled)

> Fastest is npm#object | local#object

* local#strings, object x 920,420 ops/sec ±1.49% (95 runs sampled)
* npm#strings, object x 903,851 ops/sec ±1.92% (90 runs sampled)
* local/dedupe#strings, object x 230,905 ops/sec ±1.73% (94 runs sampled)
* npm/dedupe#strings, object x 235,965 ops/sec ±1.47% (97 runs sampled)

> Fastest is local#strings, object

* local#mix x 523,688 ops/sec ±1.57% (95 runs sampled)
* npm#mix x 524,289 ops/sec ±1.47% (95 runs sampled)
* local/dedupe#mix x 157,679 ops/sec ±1.82% (93 runs sampled)
* npm/dedupe#mix x 158,168 ops/sec ±1.72% (94 runs sampled)

> Fastest is npm#mix | local#mix

* local#arrays x 265,452 ops/sec ±2.35% (91 runs sampled)
* npm#arrays x 267,699 ops/sec ±2.47% (91 runs sampled)
* local/dedupe#arrays x 124,632 ops/sec ±2.00% (94 runs sampled)
* npm/dedupe#arrays x 118,793 ops/sec ±3.27% (88 runs sampled)

> Fastest is npm#arrays | local#arrays

Again, results for the same code loaded from different sources differ very significantly.

[victor-homyakov]

In the attachment is the HTML page - maybe you'll consider it worth adding to the repository. In that case benchmark scripts should be refactored to reuse common parts both on page and in node script.

benchmarks

[dcousens]

@victor-homyakov you're welcome to make a PR 👍

[victor-homyakov]

PR #100

[dcousens]

Closing in favour of #100