repo.Commits.QueryBy(<path>) doesn't return entire history of changing commits

Empieza reproduciendo el repositorio de ejemplo y ejecutando el código repo.Commits.QueryBy(filePath) mostrado en el issue contra todo el historial de repo.Commits. Inspecciona el punto de entrada CommitLog.QueryBy y compara cómo se gestionan los commits de merge. Se considera terminado determinar si debería aparecer el commit de merge de Iteration 4 y corregir el comportamiento o documentar el resultado previsto.

I was playing around with using CommitLog.QueryBy() to obtain the history of a specific file, and I am not clear if I'm using the feature correctly, but not getting the output I would expect.

Specifically, in a sample repo with 5 commits where each commit changes the same file, .QueryBy() appears to omit the commit that is merging and changing the file.

Is it possible this is by design, and the merge change in Iteration 4 is being purposefully omitted, or might this be a bug?

Here's the output of my sample program:

History of sample.txt has 4 entries:
 - 742d4c5d1fa8c8b4f3fd6894580cb33e7a3ab4e0: Iteration 5
 - 0cbfdb9a0f98ef3e62ba9e8f2d5dd6ef9bd755f6: Iteration 2
 - 59947c86d20f45b6f19dfd52dcbc4c4b23fc4d16: Iteration 3
 - 020323b5bc36f931442343df6aa0557e8d20581f: Iteration 1
Commit history has 5 entries:
 - 742d4c5d1fa8c8b4f3fd6894580cb33e7a3ab4e0: Iteration 5
 - d096877e0d40c0a002e73c308a50d94d6ffbcb67: Iteration 4
 - 0cbfdb9a0f98ef3e62ba9e8f2d5dd6ef9bd755f6: Iteration 2
 - 59947c86d20f45b6f19dfd52dcbc4c4b23fc4d16: Iteration 3
 - 020323b5bc36f931442343df6aa0557e8d20581f: Iteration 1

Here's the code that i'm running against a sample repo (see below):

using LibGit2Sharp;
using System;
using System.Linq;

namespace ConsoleApplication2
{
    class Program
    {
        // This program is compiled against:
        // - LibGit2Sharp v0.23.0
        // - Native binaries v1.0.129
        static void Main(string[] args)
        {
            var repo = new Repository(Repository.Discover(Environment.CurrentDirectory));

            var filePath = "sample.txt";

            // Look at the history of sample.txt
            var fileHist = repo.Commits.QueryBy(filePath).Select(x => x.Commit).ToList();
            Console.WriteLine("History of {0} has {1} entries:", filePath, fileHist.Count);
            foreach (var commit in fileHist)
                Console.WriteLine(" - {0}: {1}", commit.Sha, commit.MessageShort);

            // Look at the entire history
            var fullHist = repo.Commits.ToList();
            Console.WriteLine("Commit history has {1} entries:", filePath, fullHist.Count);
            foreach (var commit in fullHist)
                Console.WriteLine(" - {0}: {1}", commit.Sha, commit.MessageShort);
        }
    }
}

Here's the script to create a sample git repo that has a diamond pattern of commits around a file:

@REM create a git repo with a single file
git init
echo 1 > sample.txt
git add sample.txt
git commit -m "Iteration 1"

@REM create a branch and make a change
git checkout -b "branch"
echo 2 >> sample.txt
git commit -a -m "Iteration 2"

@REM make a change on the master
git checkout master
echo 3 >> sample.txt
git commit -a -m "Iteration 3"

@REM merge branch into master and overwrite conflict
git merge branch
echo 1 > sample.txt
echo 2 >> sample.txt
echo 3 >> sample.txt
echo 4 >> sample.txt
git commit -a -m "Iteration 4"

@REM one last change for good measure
echo 5 >> sample.txt
git commit -a -m "Iteration 5"