szilanor / stream Goto Github PK

The examples used for "Classical JS" seems to use code that is very slow and contains extensive typing and unneeded code.

As an example, here's a benchmark using the second to last code example (concerning readability):

Map x 636 ops/sec ±124.98% (5 runs sampled)
Stream x 30,502 ops/sec ±15.18% (37 runs sampled)
Object x 441,113 ops/sec ±19.31% (34 runs sampled)
Reduce x 1,285,775 ops/sec ±5.84% (43 runs sampled)
Object with For Loop x 443,450 ops/sec ±13.82% (43 runs sampled)
Fastest is Reduce

As it can be seen, the map from the example is very very slow. Using an object instead of map greatly improves performance and outperforms this library (as also indicated in #2 )

REPL with the benchmark

In terms of readability I'd like to suggest a more "cleaned up" comparison with less extensive typing and removing unneeded code so users can have a fair comparison between this library and "Classic JS". See following TS Playground as an example with different approaches to the problem from the example.

Example TS Playground

When using the provided benchmark, I noticed that I could not reproduce the stated result. In my tests, this library always was the slowest.

The results of my benchmark (I added a For Loop as comparison):

Classic JS x 11,375,579 ops/sec ±10.54% (48 runs sampled)
Stream API x 33,607 ops/sec ±25.26% (40 runs sampled)
For loop x 34,567,827 ops/sec ±91.45% (48 runs sampled)
Fastest is For loop

Online REPL

Did I maybe do something wrong in my setup? If not, I think it'd be good to either remove the benchmark or update them.