Benchmarking two different approaches for collection validation

Vitor R. F. Ribeiro — Sun, 09 May 2021 14:36:51 +0000

I recently came up in a situation at work where I needed to validate a collection of permission's ids (providedIds) based on a valid collection of permission's ids (validIds). If I find at least one invalid Id, the whole collection fails the validation.

So I used LINQ Except extension to check what exists on my providedIds that doens't exist in my validIds. If there are any, my providedIds are not valid.

public bool ExceptApproach(IEnumerable<int> permissionIds, IEnumerable<int> validPermissionsIds)
{
    var invalidPermissions = permissionIds.Except(validPermissionsIds);

    return !invalidPermissions.Any();
}

But then it occurred to me that all I need to know is if there are a single invalid id in my collection. So I tryed a different approach:

public bool AnyApproach(IEnumerable<int> permissionIds, IEnumerable<int> validPermissionsIds)
{
    var atLeastOneInvalidPermission = permissionIds.Any(id => !validPermissionsIds.Contains(id));

    return !atLeastOneInvalidPermission;
}

But, is it really better? To answear that question I searched a way to benchmark it and decided to use: benchmarkdotnet. And I came up with this:


BenchmarkDotNet=v0.12.1, OS=Windows 10.0.19041.928 (2004/?/20H1)
Intel Core i5-9400F CPU 2.90GHz (Coffee Lake), 1 CPU, 6 logical and 6 physical cores
.NET Core SDK=5.0.101
  [Host]     : .NET Core 5.0.1 (CoreCLR 5.0.120.57516, CoreFX 5.0.120.57516), X64 RyuJIT
  DefaultJob : .NET Core 5.0.1 (CoreCLR 5.0.120.57516, CoreFX 5.0.120.57516), X64 RyuJIT

Method	permissionIds	validPermissionsIds	Mean	Error	StdDev
ExceptApproach	[1,2,3]	[1,2,3]	199.40 ns	1.181 ns	1.104 ns
ExceptApproach	[1,2,3]	[2,3,4]	162.11 ns	2.729 ns	2.553 ns
ExceptApproach	Range(1, 100)	Range(1, 100)	3,787.92 ns	59.032 ns	55.219 ns
ExceptApproach	Range(1, 100)	Range(2, 101)	2,242.24 ns	7.806 ns	6.920 ns
AnyApproach	[1,2,3]	[1,2,3]	83.33 ns	0.294 ns	0.260 ns
AnyApproach	[1,2,3]	[2,3,4]	41.22 ns	0.479 ns	0.425 ns
AnyApproach	Range(1, 100)	Range(1, 100)	2,398.31 ns	22.912 ns	21.432 ns
AnyApproach	Range(1, 100)	Range(2, 101)	57.68 ns	0.770 ns	0.721 ns

The AnyApproach is better than the ExceptApproach for small collections. When the collection
[1,2,3] is valid, the AnyApproach was about 50% faster. And when the collection is invalid [2,3,4]
the AnyApproach was way better (because the first element was already invalid)

For larger collections, the ExceptApproach and AnyApproach were more close: 3,787.92 ns x 2,398.31 ns and yet, the AnyApproach was faster (for a valid collection). For a invalid collection, the AnyApproach was significantly faster (again, because the first element was already invalid)

In conclusion, I think using the AnyApproach is better than the ExceptApproach. I know I should have included more cenarios like: "what if only the last permission Id is invalid?" or "what if only the element at the center of the collection is invalid?", but for the situation I have at work, this is enough to go on.

*For those wandering, I edited the columns permissionIds and validPermissionsIds because I didn't find a way to include custom names for the values I used. Here's the full code.

DEV Community: Vitor R. F. Ribeiro

Benchmarking two different approaches for collection validation