Add test verifying empty data appends have no effect on hash results

[Copilot]

Description

Adds AppendingEmptyHasNoEffect() test to NonCryptoHashTestDriver base class, verifying that interleaving empty span appends with actual data produces identical hashes to appending data alone.

Changes

• Added [Fact] test method to NonCryptoHashTestDriver.cs
• Test pattern: reference hash with [1,2,3] vs. hash with Empty + [1,2,3] + Empty
• Automatically applies to all implementations: CRC32, CRC64, XxHash32, XxHash64 (including seeded variants), and future additions

[Fact]
public void AppendingEmptyHasNoEffect()
{
    NonCryptographicHashAlgorithm reference = CreateInstance();
    reference.Append(new byte[] { 1, 2, 3 });
    byte[] expected = reference.GetCurrentHash();

    NonCryptographicHashAlgorithm hash = CreateInstance();
    hash.Append(ReadOnlySpan<byte>.Empty);
    hash.Append(new byte[] { 1, 2, 3 });
    hash.Append(ReadOnlySpan<byte>.Empty);
    byte[] actual = hash.GetCurrentHash();

    Assert.Equal(expected, actual);
}

Testing

All 1061 tests pass, including 8 instances of the new test across hash implementations.

Original prompt

Summary

Add a new test to the NonCryptoHashTestDriver base class that verifies appending empty data (empty spans/arrays) has no observable effect on the hash result.

Background

This was requested by @vcsjones in PR #123601 review comment: #123601 (comment)

I think we should add a test to the base test fixtures that basically do:

public void Test1()
{
    Hash reference = new();
    reference.Append([1, 2, 3]);
    uint expected = reference.GetCurrentHashAsUInt32();

    Hash h = new();
    h.Append([]);
    h.Append([1, 2, 3]);
    h.Append([]);
    uint actual = h.GetCurrentHashAsUInt32();
    
    Assert.Equal(expected, actual);
}

I am not aware of any hash function implementations where appending empty has an observable effect.

Implementation

Add a new [Fact] test method to src/libraries/System.IO.Hashing/tests/NonCryptoHashTestDriver.cs that:

1. Creates a reference hash instance and appends some test data (e.g., [1, 2, 3])
2. Gets the expected hash value
3. Creates another hash instance
4. Appends empty data, then the same test data, then empty data again
5. Gets the actual hash value
6. Asserts they are equal

Example implementation:

[Fact]
public void AppendingEmptyHasNoEffect()
{
    NonCryptographicHashAlgorithm reference = CreateInstance();
    reference.Append(new byte[] { 1, 2, 3 });
    byte[] expected = reference.GetCurrentHash();

    NonCryptographicHashAlgorithm hash = CreateInstance();
    hash.Append(ReadOnlySpan<byte>.Empty);
    hash.Append(new byte[] { 1, 2, 3 });
    hash.Append(ReadOnlySpan<byte>.Empty);
    byte[] actual = hash.GetCurrentHash();

    Assert.Equal(expected, actual);
}

This test will automatically run for all NonCryptographicHashAlgorithm implementations (CRC32, CRC64, XxHash32, XxHash64, XxHash3, XxHash128, and the new Adler32), ensuring consistent behavior across the board.

This pull request was created from Copilot chat.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-io
See info in area-owners.md if you want to be subscribed.

[Copilot]

Pull request overview

This PR adds a new test to verify that appending empty data (empty spans) has no observable effect on hash computation results across all non-cryptographic hash algorithm implementations.

Changes:

• Added AppendingEmptyHasNoEffect() test method to NonCryptoHashTestDriver base class
• Test validates that interleaving empty span appends with actual data produces identical hashes to appending data alone
• Automatically applies to all implementations: CRC32, CRC64, XxHash32, XxHash64 (including seeded variants)

[stephentoub]

/ba-g "Failed to generate "CoreCLR component" build project!" unrelated to this test change