Ability to generate test operations for original values in the first object (continuation)

warpech commented 5 years ago

Continuation of https://github.com/Starcounter-Jack/JSON-Patch/pull/226

Changes that I made compared to the original PR:

was: extra boolean argument on "observe" and extra object argument on "generate" and - "compare"
is: extra boolean argument "generate" and "compare"
changed the word inversible to invertible. Please correct me if I'm wrong
added way more tests for test operation generation

Pending questions:

https://github.com/immerjs/immer also registers inverse patches, but they do it in a separate patch, not within the same patch. Is the solution in this PR more or less useful?

Benchmark results:

Before this PR (bbd3002ac0ec296b8289590bdb5fd576dca17e0b, `npm run build && npm run bench-duplex`)

add operation
 Ops/sec: 3021278 ±0.33% Ran 154893 times in 0.051 seconds.
remove operation
 Ops/sec: 2467208 ±0.45% Ran 127226 times in 0.052 seconds.
replace operation
 Ops/sec: 2854277 ±0.28% Ran 146362 times in 0.051 seconds.
move operation
 Ops/sec: 271583 ±0.37% Ran 15241 times in 0.056 seconds.
copy operation
 Ops/sec: 214417 ±0.45% Ran 18773 times in 0.088 seconds.
test operation
 Ops/sec: 1316006 ±0.31% Ran 67352 times in 0.051 seconds.
===================
generate operation
 Ops/sec: 1680155 ±0.32% Ran 87099 times in 0.052 seconds.
generate operation and re-apply
 Ops/sec: 1483583 ±0.20% Ran 76006 times in 0.051 seconds.
compare operation
 Ops/sec: 427630 ±0.36% Ran 22101 times in 0.052 seconds.
compare operation same but deep objects
 Ops/sec: 122387928 ±0.31% Ran 6230468 times in 0.051 seconds.

After this PR

add operation
 Ops/sec: 3033581 ±0.26% Ran 155154 times in 0.051 seconds.
remove operation
 Ops/sec: 2513818 ±0.33% Ran 129460 times in 0.051 seconds.
replace operation
 Ops/sec: 2804656 ±0.31% Ran 144192 times in 0.051 seconds.
move operation
 Ops/sec: 271968 ±0.38% Ran 15132 times in 0.056 seconds.
copy operation
 Ops/sec: 215846 ±0.35% Ran 19086 times in 0.088 seconds.
test operation
 Ops/sec: 1311638 ±0.33% Ran 67526 times in 0.051 seconds.
==================
generate operation
 Ops/sec: 1705131 ±0.32% Ran 87963 times in 0.052 seconds.
generate operation and re-apply
 Ops/sec: 1452317 ±1.99% Ran 75998 times in 0.052 seconds.
compare operation
 Ops/sec: 422706 ±0.52% Ran 21843 times in 0.052 seconds.
compare operation same but deep objects
 Ops/sec: 115692461 ±0.71% Ran 6021898 times in 0.052 seconds.
================== new benchmarks start here: ==================
generate operation, invertible = true
 Ops/sec: 1607667 ±0.27% Ran 82412 times in 0.051 seconds.
generate operation and re-apply, invertible = true
 Ops/sec: 1401189 ±0.42% Ran 73320 times in 0.052 seconds.
compare operation, invertible = true
 Ops/sec: 237997 ±0.27% Ran 12237 times in 0.051 seconds.
compare operation same but deep objects, invertible = true
 Ops/sec: 116096872 ±0.50% Ran 5978518 times in 0.051 seconds.

warpech commented 5 years ago

cc @pytlesk4

warpech commented 5 years ago

@tomalec please re-review.

I have achieved performing the same test using invertible set as true and then false by creating a new helper functions: variantIt, insertIf and usage of spread operator. I think this is quite neat, similar to NUnit fixtures. I wonder if this is readable, though. Please let me know.

tomalec commented 5 years ago

I wonder if this is readable, though. Please let me know.

At least to me it is :)

pytlesk4 commented 5 years ago

@warpech thanks for taking this over the finish line!

tomalec commented 5 years ago

@pytlesk4 once again thanks for starting a great PR and sorry for being picky :)

warpech commented 5 years ago

Released today as 2.2.0

pytlesk4 commented 5 years ago

@warpech can this get published to NPM?

warpech commented 5 years ago

Done, thanks for pinging me!

Starcounter-Jack / JSON-Patch