Adding support for TypedArrays

[calumrussell]

I have added support for Uint8Array with a test showing that it works as expected. No additional types have been created, this works by simply treating a Uint8Array like a normal array. Afaik, you cannot create a nested array this way so the changes are quite limited.

Support for more types of TypedArray is easy to add but as I worked on this issue further I realised that there may be wider implications for this:

• Does this conflict with JSONSchema?
• TypedArrays appear to be strongly typed (for example, creating a Uint8Array with values longer than 8 bits will always be cast down to 8 bit values) so would it be possible to improve perf by removing type checks (I looked at this briefly but I didn't see how it could be done without broader changes).
• Which TypedArray should be supported?

Related to - #626 - I haven't verified the performance difference either, I am assuming this can only come from removing type checks?

Checklist

• run npm run test and npm run benchmark
• tests and/or benchmarks are included
• documentation is changed or added
• commit message and code follows the Developer's Certification of Origin
  and the Code of conduct

[Uzlopak]

ugh. It would be better to put the array in a "global" scope and make an indexOf by the obj.constructor.name than this imho.

I think this will slow down array significantly.

[calumrussell]

I am not familiar with perf implications here.

Would Array.includes also be worth testing?

[ivan-tymoshenko]

This would not work in places where we use Ajv. To test it put an array under anyOf, oneOf or if.

[github-actions]

PR:

short string............................................. x 13,595,841 ops/sec ±1.51% (184 runs sampled)
short string with double quote........................... x 9,215,276 ops/sec ±1.55% (183 runs sampled)
long string without double quotes........................ x 25,779 ops/sec ±1.50% (182 runs sampled)
long string.............................................. x 10,345 ops/sec ±1.46% (186 runs sampled)
number................................................... x 896,130,427 ops/sec ±1.17% (184 runs sampled)
integer.................................................. x 198,751,726 ops/sec ±1.43% (181 runs sampled)
formatted date-time...................................... x 2,369,994 ops/sec ±1.32% (184 runs sampled)
formatted date........................................... x 1,656,788 ops/sec ±1.38% (183 runs sampled)
formatted time........................................... x 1,638,795 ops/sec ±1.31% (184 runs sampled)
short array of numbers................................... x 53,333 ops/sec ±1.53% (178 runs sampled)
short array of integers.................................. x 47,163 ops/sec ±1.81% (184 runs sampled)
short array of short strings............................. x 13,496 ops/sec ±1.20% (181 runs sampled)
short array of long strings.............................. x 13,338 ops/sec ±1.33% (182 runs sampled)
short array of objects with properties of different types x 6,793 ops/sec ±1.56% (182 runs sampled)
object with number property.............................. x 896,784,352 ops/sec ±1.24% (185 runs sampled)
object with integer property............................. x 193,037,691 ops/sec ±1.14% (184 runs sampled)
object with short string property........................ x 13,774,587 ops/sec ±1.27% (182 runs sampled)
object with long string property......................... x 10,398 ops/sec ±1.32% (184 runs sampled)
object with properties of different types................ x 1,403,202 ops/sec ±1.60% (179 runs sampled)
simple object............................................ x 7,110,621 ops/sec ±1.33% (185 runs sampled)
simple object with required fields....................... x 7,041,745 ops/sec ±1.37% (183 runs sampled)
object with const string property........................ x 875,952,643 ops/sec ±1.30% (182 runs sampled)
object with const number property........................ x 897,199,115 ops/sec ±1.33% (186 runs sampled)
object with const bool property.......................... x 904,853,523 ops/sec ±1.23% (184 runs sampled)
object with const object property........................ x 886,073,478 ops/sec ±1.32% (182 runs sampled)
object with const null property.......................... x 902,604,508 ops/sec ±1.35% (183 runs sampled)

MASTER:

short string............................................. x 14,150,071 ops/sec ±1.28% (185 runs sampled)
short string with double quote........................... x 9,779,016 ops/sec ±1.25% (184 runs sampled)
long string without double quotes........................ x 26,082 ops/sec ±1.46% (180 runs sampled)
long string.............................................. x 10,256 ops/sec ±1.43% (182 runs sampled)
number................................................... x 892,098,738 ops/sec ±1.08% (184 runs sampled)
integer.................................................. x 202,345,214 ops/sec ±1.16% (187 runs sampled)
formatted date-time...................................... x 2,430,098 ops/sec ±0.89% (185 runs sampled)
formatted date........................................... x 1,664,159 ops/sec ±1.09% (185 runs sampled)
formatted time........................................... x 1,654,243 ops/sec ±1.28% (184 runs sampled)
short array of numbers................................... x 55,645 ops/sec ±1.60% (184 runs sampled)
short array of integers.................................. x 48,082 ops/sec ±1.40% (181 runs sampled)
short array of short strings............................. x 13,352 ops/sec ±1.41% (181 runs sampled)
short array of long strings.............................. x 12,081 ops/sec ±0.97% (181 runs sampled)
short array of objects with properties of different types x 6,302 ops/sec ±1.36% (181 runs sampled)
object with number property.............................. x 805,934,162 ops/sec ±1.33% (182 runs sampled)
object with integer property............................. x 189,319,369 ops/sec ±1.34% (185 runs sampled)
object with short string property........................ x 13,993,875 ops/sec ±1.24% (186 runs sampled)
object with long string property......................... x 10,526 ops/sec ±1.15% (185 runs sampled)
object with properties of different types................ x 1,390,259 ops/sec ±1.73% (181 runs sampled)
simple object............................................ x 7,115,977 ops/sec ±1.22% (184 runs sampled)
simple object with required fields....................... x 7,092,319 ops/sec ±1.42% (180 runs sampled)
object with const string property........................ x 887,052,086 ops/sec ±1.38% (181 runs sampled)
object with const number property........................ x 859,910,878 ops/sec ±1.50% (179 runs sampled)
object with const bool property.......................... x 863,841,739 ops/sec ±1.46% (181 runs sampled)
object with const object property........................ x 855,027,252 ops/sec ±1.42% (182 runs sampled)
object with const null property.......................... x 896,106,854 ops/sec ±1.22% (184 runs sampled)

[calumrussell]

This would not work in places where we use Ajv. To test it put an array under anyOf, oneOf or if.

It seems like there is no way to solve this?

[Tony133]

Closing this PR as it has been inactive for some time and may no longer be aligned with the current state of the codebase.

If the changes are still relevant, feel free to open a new PR with the necessary updates.