What Micro Optimization?
Removed defer from the simple Tag Cache lock & unlock and reduced execution time for even single field validation by almost 100ns from 254ns/op to 163ns/op... Not bad for such a simple modification and this effect ripples throughout struct validation.
New Benchmarks
$ go test -cpu=4 -bench=. -benchmem=true
PASS
BenchmarkFieldSuccess-4 10000000 163 ns/op 0 B/op 0 allocs/op
BenchmarkFieldFailure-4 2000000 673 ns/op 400 B/op 4 allocs/op
BenchmarkFieldDiveSuccess-4 500000 3019 ns/op 480 B/op 27 allocs/op
BenchmarkFieldDiveFailure-4 500000 3553 ns/op 880 B/op 31 allocs/op
BenchmarkFieldCustomTypeSuccess-4 5000000 347 ns/op 32 B/op 2 allocs/op
BenchmarkFieldCustomTypeFailure-4 2000000 645 ns/op 400 B/op 4 allocs/op
BenchmarkFieldOrTagSuccess-4 1000000 1177 ns/op 16 B/op 1 allocs/op
BenchmarkFieldOrTagFailure-4 1000000 1093 ns/op 432 B/op 6 allocs/op
BenchmarkStructLevelValidationSuccess-4 2000000 702 ns/op 160 B/op 6 allocs/op
BenchmarkStructLevelValidationFailure-4 1000000 1279 ns/op 592 B/op 11 allocs/op
BenchmarkStructSimpleCustomTypeSuccess-4 1000000 1010 ns/op 80 B/op 5 allocs/op
BenchmarkStructSimpleCustomTypeFailure-4 1000000 1544 ns/op 624 B/op 11 allocs/op
BenchmarkStructPartialSuccess-4 1000000 1249 ns/op 400 B/op 11 allocs/op
BenchmarkStructPartialFailure-4 1000000 1797 ns/op 816 B/op 16 allocs/op
BenchmarkStructExceptSuccess-4 2000000 927 ns/op 368 B/op 9 allocs/op
BenchmarkStructExceptFailure-4 1000000 1259 ns/op 400 B/op 11 allocs/op
BenchmarkStructSimpleCrossFieldSuccess-4 1000000 1076 ns/op 128 B/op 6 allocs/op
BenchmarkStructSimpleCrossFieldFailure-4 1000000 1623 ns/op 560 B/op 11 allocs/op
BenchmarkStructSimpleCrossStructCrossFieldSuccess-4 1000000 1582 ns/op 176 B/op 9 allocs/op
BenchmarkStructSimpleCrossStructCrossFieldFailure-4 1000000 2139 ns/op 608 B/op 14 allocs/op
BenchmarkStructSimpleSuccess-4 1000000 1040 ns/op 48 B/op 3 allocs/op
BenchmarkStructSimpleFailure-4 1000000 1683 ns/op 624 B/op 11 allocs/op
BenchmarkStructSimpleSuccessParallel-4 5000000 356 ns/op 48 B/op 3 allocs/op
BenchmarkStructSimpleFailureParallel-4 2000000 831 ns/op 624 B/op 11 allocs/op
BenchmarkStructComplexSuccess-4 200000 6738 ns/op 512 B/op 30 allocs/op
BenchmarkStructComplexFailure-4 200000 11387 ns/op 3415 B/op 72 allocs/op
BenchmarkStructComplexSuccessParallel-4 500000 2330 ns/op 512 B/op 30 allocs/op
BenchmarkStructComplexFailureParallel-4 300000 4857 ns/op 3416 B/op 72 allocs/opOld Benchmarks
$ go test -cpu=4 -bench=. -benchmem=true
PASS
BenchmarkFieldSuccess-4 5000000 254 ns/op 0 B/op 0 allocs/op
BenchmarkFieldFailure-4 2000000 779 ns/op 400 B/op 4 allocs/op
BenchmarkFieldDiveSuccess-4 500000 3451 ns/op 480 B/op 27 allocs/op
BenchmarkFieldDiveFailure-4 300000 3954 ns/op 880 B/op 31 allocs/op
BenchmarkFieldCustomTypeSuccess-4 3000000 451 ns/op 32 B/op 2 allocs/op
BenchmarkFieldCustomTypeFailure-4 2000000 751 ns/op 400 B/op 4 allocs/op
BenchmarkFieldOrTagSuccess-4 1000000 1312 ns/op 16 B/op 1 allocs/op
BenchmarkFieldOrTagFailure-4 1000000 1206 ns/op 432 B/op 6 allocs/op
BenchmarkStructLevelValidationSuccess-4 2000000 829 ns/op 160 B/op 6 allocs/op
BenchmarkStructLevelValidationFailure-4 1000000 1403 ns/op 592 B/op 11 allocs/op
BenchmarkStructSimpleCustomTypeSuccess-4 1000000 1238 ns/op 80 B/op 5 allocs/op
BenchmarkStructSimpleCustomTypeFailure-4 1000000 1771 ns/op 624 B/op 11 allocs/op
BenchmarkStructPartialSuccess-4 1000000 1399 ns/op 400 B/op 11 allocs/op
BenchmarkStructPartialFailure-4 1000000 1926 ns/op 816 B/op 16 allocs/op
BenchmarkStructExceptSuccess-4 2000000 930 ns/op 368 B/op 9 allocs/op
BenchmarkStructExceptFailure-4 1000000 1381 ns/op 400 B/op 11 allocs/op
BenchmarkStructSimpleCrossFieldSuccess-4 1000000 1253 ns/op 128 B/op 6 allocs/op
BenchmarkStructSimpleCrossFieldFailure-4 1000000 1849 ns/op 560 B/op 11 allocs/op
BenchmarkStructSimpleCrossStructCrossFieldSuccess-4 1000000 1902 ns/op 176 B/op 9 allocs/op
BenchmarkStructSimpleCrossStructCrossFieldFailure-4 500000 2453 ns/op 608 B/op 14 allocs/op
BenchmarkStructSimpleSuccess-4 1000000 1182 ns/op 48 B/op 3 allocs/op
BenchmarkStructSimpleFailure-4 1000000 1842 ns/op 624 B/op 11 allocs/op
BenchmarkStructSimpleSuccessParallel-4 5000000 342 ns/op 48 B/op 3 allocs/op
BenchmarkStructSimpleFailureParallel-4 2000000 665 ns/op 624 B/op 11 allocs/op
BenchmarkStructComplexSuccess-4 200000 7924 ns/op 512 B/op 30 allocs/op
BenchmarkStructComplexFailure-4 100000 12605 ns/op 3415 B/op 72 allocs/op
BenchmarkStructComplexSuccessParallel-4 1000000 2421 ns/op 512 B/op 30 allocs/op
BenchmarkStructComplexFailureParallel-4 300000 4234 ns/op 3416 B/op 72 allocs/opP.S.
I've become a little obsessed with efficiency and allocations as of late so don't be surprised to see more improvements keep coming! and who knows maybe there's even an attempt at a zero allocation validator in the works... 😉 😉