v2.6.1 breaks existing test - Assert.Equal() Failure: Collections differ

[tig]

I have not been able to create a simpler repro yet but will if it's needed.

We have the following test:

Assert.Equal (TextModel.StringToRuneCells ("Threally really first line."), tm.GetLine (0));

With v2.5.3 this test passed.

With xUnit v2.6.1 we get:

Xunit.Sdk.EqualException: 'Assert.Equal() Failure: Collections differ
           ↓ (pos 0)
Expected: [U+0054 'T'; null, U+0068 'h'; null, U+0072 'r'; null, U+0065 'e'; null, U+0061 'a'; null, ···]
Actual:   [U+0054 'T'; null, U+0068 'h'; null, U+0072 'r'; null, U+0065 'e'; null, U+0061 'a'; null, ···]
           ↑ (pos 0)'

Breaking on this Assert and viewing the two objects in Immediate Window, they are the same:

tm.GetLine(0)
Count = 27
    [0]: "U+0054 'T'; null"
    [1]: "U+0068 'h'; null"
    [2]: "U+0072 'r'; null"
    [3]: "U+0065 'e'; null"
    [4]: "U+0061 'a'; null"
    [5]: "U+006C 'l'; null"
    [6]: "U+006C 'l'; null"
    [7]: "U+0079 'y'; null"
    [8]: "U+0020 ' '; null"
    [9]: "U+0072 'r'; null"
    [10]: "U+0065 'e'; null"
    [11]: "U+0061 'a'; null"
    [12]: "U+006C 'l'; null"
    [13]: "U+006C 'l'; null"
    [14]: "U+0079 'y'; null"
    [15]: "U+0020 ' '; null"
    [16]: "U+0066 'f'; null"
    [17]: "U+0069 'i'; null"
    [18]: "U+0072 'r'; null"
    [19]: "U+0073 's'; null"
    [20]: "U+0074 't'; null"
    [21]: "U+0020 ' '; null"
    [22]: "U+006C 'l'; null"
    [23]: "U+0069 'i'; null"
    [24]: "U+006E 'n'; null"
    [25]: "U+0065 'e'; null"
    [26]: "U+002E '.'; null"
TextModel.StringToRuneCells ("Threally really first line.")
Count = 27
    [0]: "U+0054 'T'; null"
    [1]: "U+0068 'h'; null"
    [2]: "U+0072 'r'; null"
    [3]: "U+0065 'e'; null"
    [4]: "U+0061 'a'; null"
    [5]: "U+006C 'l'; null"
    [6]: "U+006C 'l'; null"
    [7]: "U+0079 'y'; null"
    [8]: "U+0020 ' '; null"
    [9]: "U+0072 'r'; null"
    [10]: "U+0065 'e'; null"
    [11]: "U+0061 'a'; null"
    [12]: "U+006C 'l'; null"
    [13]: "U+006C 'l'; null"
    [14]: "U+0079 'y'; null"
    [15]: "U+0020 ' '; null"
    [16]: "U+0066 'f'; null"
    [17]: "U+0069 'i'; null"
    [18]: "U+0072 'r'; null"
    [19]: "U+0073 's'; null"
    [20]: "U+0074 't'; null"
    [21]: "U+0020 ' '; null"
    [22]: "U+006C 'l'; null"
    [23]: "U+0069 'i'; null"
    [24]: "U+006E 'n'; null"
    [25]: "U+0065 'e'; null"
    [26]: "U+002E '.'; null"

Are there any known issues in xUnit > v2.5.3 that might be causing this?

[tig]

Relevant code:

	/// <summary>
	/// Represents a single row/column within the <see cref="TextView"/>. Includes the glyph and the foreground/background colors.
	/// </summary>
	[DebuggerDisplay ("{DebuggerDisplay}")]
	public class RuneCell : IEquatable<RuneCell> {
		/// <summary>
		/// The glyph to draw.
		/// </summary>
		[JsonConverter (typeof (RuneJsonConverter))]
		public Rune Rune { get; set; }

		/// <summary>
		/// The <see cref="Terminal.Gui.ColorScheme"/> color sets to draw the glyph with.
		/// </summary>
		[JsonConverter (typeof (ColorSchemeJsonConverter))]
		public ColorScheme? ColorScheme { get; set; }

		/// <summary>Indicates whether the current object is equal to another object of the same type.</summary>
		/// <param name="other">An object to compare with this object.</param>
		/// <returns>
		///   <see langword="true" /> if the current object is equal to the <paramref name="other" /> parameter; 
		///   otherwise, <see langword="false" />.</returns>
		public bool Equals (RuneCell? other)
		{
			return other != null &&
			    Rune.Equals (other.Rune) &&
			    ColorScheme == other.ColorScheme;
		}

		/// <summary>Returns a string that represents the current object.</summary>
		/// <returns>A string that represents the current object.</returns>
		public override string ToString ()
		{
			string colorSchemeStr = ColorSchemeDebuggerDisplay ();
			return DebuggerDisplay;
		}

		private string ColorSchemeDebuggerDisplay ()
		{
			var colorSchemeStr = "null";
			if (ColorScheme != null) {
				colorSchemeStr = $"Normal: {ColorScheme.Normal.Foreground},{ColorScheme.Normal.Background}; " +
				    $"Focus: {ColorScheme.Focus.Foreground},{ColorScheme.Focus.Background}; " +
				    $"HotNormal: {ColorScheme.HotNormal.Foreground},{ColorScheme.HotNormal.Background}; " +
				    $"HotFocus: {ColorScheme.HotFocus.Foreground},{ColorScheme.HotFocus.Background}; " +
				    $"Disabled: {ColorScheme.Disabled.Foreground},{ColorScheme.Disabled.Background}";
			}

			return colorSchemeStr;
		}

		private string DebuggerDisplay {
			get {
				string colorSchemeStr = ColorSchemeDebuggerDisplay ();
				return $"U+{Rune.Value:X4} '{Rune.ToString ()}'; {colorSchemeStr}";
			}
		}
	}

// Turns the string into cells, this does not split the 
// contents on a newline if it is present.
internal static List<RuneCell> StringToRuneCells (string str, ColorScheme? colorScheme = null)
{
	List<RuneCell> cells = new List<RuneCell> ();
	foreach (var rune in str.ToRunes ()) {
		cells.Add (new RuneCell { Rune = rune, ColorScheme = colorScheme });
	}
	return cells;
}

/// <summary>
/// Returns the specified line as a List of Rune
/// </summary>
/// <returns>The line.</returns>
/// <param name="line">Line number to retrieve.</param>
public List<RuneCell> GetLine (int line)
{
	if (_lines.Count > 0) {
		if (line < Count) {
			return _lines [line];
		} else {
			return _lines [Count - 1];
		}
	} else {
		_lines.Add (new List<RuneCell> ());
		return _lines [0];
	}
}

[bradwilson]

@tig I will need a full repro to track this down.

[tig]

@tig I will need a full repro to track this down.

With xUnit 2.5.3 TestObj.Equals gets called.

With xUnit 2.6.1 it does not.

		public class TestObj : IEquatable<TestObj> {
			public char Char { get; set; }

			public bool Equals (TestObj other)
			{
				return other?.Char == Char;
			}
		}


		[Fact]
		public void Repo_xUnit_2821 ()
		{
			List<TestObj> list1 = new List<TestObj> ();
			list1.Add (new TestObj { Char = 'a' });

			List<TestObj> list2 = new List<TestObj> ();
			list2.Add (new TestObj { Char = 'a' });

			Assert.Equal(list1, list2);
		}

[bradwilson]

This looks like a pretty strange edge case that came about because of performance optimizations. We were sort of accidentally getting the right result for the wrong reason previously. Working on a fix now.

[bradwilson]

Available in v2: 2.6.2-pre.15
Available in v3: 0.1.1-pre.325

[tig]

2.6.2-pre

Will you push this to nuget?

[bradwilson]

Not until a final RTM. You can use CI builds from the feedz.io feed, instructions here: https://xunit.net/docs/using-ci-builds

[gioce90]

Hi, is this fixed with 2.6.2 or with a future release? Because in 2.6.2 I think there is still the same (or similar) regression

[bradwilson]

@gioce90 Yes, the bug specified here was fixed in 2.6.2.

The new bug you opened is a duplicate of a bug that is already fixed and will be shipped in 2.6.3.

[gioce90]

Okay thank you