add skipStringOptimizations options to toString

[Klaasvaak]

I'm optimizing the performance of a project I'm working on. It contains large paths (for example with a length of 160k). When skipping the string optimizations in toString I can see performance improvements around 7-11x. When having multiple of these large paths this adds up.

I made a simple example here: https://codepen.io/klaasvaak/pen/dyVjPje?editors=1011

Let me know if you have concerns about skipping these string optimizations.

[puzrin]

Are you sure this option required for other users? For example, you could just override prototype method (or do it in inherited class)

[Klaasvaak]

@puzrin Very good point. I will do this instead.

[puzrin]

No problem, happy to help.

BTW

• It's better to use benchmark.js, because single pass may be not enough to "warm up" JIT inline caches. See https://github.com/fontello/svgpath/tree/master/benchmark.
• You could try to optimize existing .toString() instead. For example:
  • use String.test(regexp) before String.replace(regexp).
  • use more clever join, instead of join everything via space and then removing those back

    svgpath/lib/svgpath.js

    Lines 201 to 207 in 70134e4

    	return elements.join(' ')
    	// Optimizations: remove spaces around commands & before `-`
    	//
    	// We could also remove leading zeros for `0.5`-like values,
    	// but their count is too small to spend time for.
    	.replace(/ ?([achlmqrstvz]) ?/gi, '$1')
    	.replace(/ \-/g, '-')

    . Existing code is from old time, when i tried to make all steps atomic, because was not certain about desired final sequence.

I did not experimented with this part of code too much. But after your example, i think there is room for improvements.

FYI. If you plan to dive into perfomance area, i strongly recommend to read this blog https://mrale.ph/. Because v8 JIT logic is completely different than anything "possible to imagine by normal human" :).

[Klaasvaak]

@puzrin Thank you for sharing of this, this is very useful for me.

[puzrin]

See v2.5.0.

./benchmark/benchmark.js 
.from("big.txt")

  > 2.4.1 x 10.33 ops/sec ±2.97% (30 runs sampled)
  > current x 10.33 ops/sec ±3.71% (30 runs sampled)

.from("one_path.txt")

  > 2.4.1 x 7,588 ops/sec ±2.90% (92 runs sampled)
  > current x 7,719 ops/sec ±1.42% (94 runs sampled)

.from(long).toString()

  > 2.4.1 x 3.17 ops/sec ±1.60% (12 runs sampled)
  > current x 103 ops/sec ±0.91% (74 runs sampled) <============

Enjoy!

[Klaasvaak]

@puzrin This is great, thanks! I can confirm using the newest version .toString() in my project speeds things up as well :)