Expanding test

Tests/Serialization/Gvwie/GeneratorPattern.cs

@@ -0,0 +1,16 @@
+using System;
+using System.Collections.Generic;
+using System.Text;
+
+namespace Esiur.Tests.Gvwie
+{
+    public enum GeneratorPattern
+    {
+        Uniform,
+        Positive,
+        Negative,
+        Alternating,
+        Small,
+        Ascending,
+    }
+}

Tests/Serialization/Gvwie/IntArrayGenerator.cs

@@ -8,107 +8,142 @@ namespace Esiur.Tests.Gvwie;
 
 public static class IntArrayGenerator
 {
+
+
+
     private static readonly Random rng = new Random(24241564);
 
-
+    /// <summary>
-    public static long[] GenerateInt32Run(int length)
+    /// Generate an array composed of ascending runs (consecutive integers).
+    /// Example output: [1,2,3,4,5, 5001,5002,5003, 10000001,10000002,...]
+    /// Parameters:
+    ///  - length: total array length
+    ///  - minRunSize / maxRunSize: inclusive bounds for run lengths
+    ///  - minValue / maxValue: allowed value range for run starts
+    ///  - allowNegative: if false, generated values will be non-negative
+    ///  - minGap / maxGap: approximate gap between runs (large gaps produce the jump examples)
+    /// </summary>
+    public static long[] GenerateRuns(int length,
+        int minRunSize = 3,
+        int maxRunSize = 8,
+        long minValue = -10_000_000L,
+        long maxValue = 10_000_000L,
+        bool allowNegative = true,
+        long minGap = 1_000L,
+        long maxGap = 10_000_000L)
     {
+        if (length <= 0)
+            return Array.Empty<long>();
+
+        if (minRunSize < 1) minRunSize = 1;
+        if (maxRunSize < minRunSize) maxRunSize = minRunSize;
+
+        // If negative runs not allowed, clamp minValue to 0
+        if (!allowNegative && minValue < 0) minValue = 0;
+
         var data = new long[length];
+        int idx = 0;
+        long prevEnd = long.MinValue;
 
-        int i = 0;
+        while (idx < length)
-        var inSmallRange = true;
-        var inShortRange = false;
-        var inLargeRange = false;
-        var inLongRange = false;
-
-        long range = 30;
-
-        while (i < length)
         {
-            // stay same range
+            // choose run size
-            if (rng.NextDouble() < 0.9)
+            int runSize = rng.Next(minRunSize, maxRunSize + 1);
-            {
+            if (idx + runSize > length)
-                if (inSmallRange)
+                runSize = length - idx;
-                    data[i++] = rng.Next(-64, 65);
-                else if (inShortRange)
-                    data[i++] = rng.NextInt64(range - 100, range + 100);
-                else if (inLargeRange)
-                    data[i++] = rng.NextInt64(range - 1000, range + 1000);
-                else if (inLongRange)
-                    data[i++] = rng.NextInt64(range - 10000, range + 10000);
-            }
-            else
-            {
-                // switch range
-                var rand = rng.NextDouble();
-                if (rand < 0.25)
-                {
-                    inSmallRange = true;
-                    inShortRange = false;
-                    inLargeRange = false;
-                    inLongRange = false;
-                    data[i++] = rng.Next(-64, 65);
-                }
-                else if (rand < 0.50)
-                {
-                    inSmallRange = false;
-                    inShortRange = true;
-                    inLargeRange = false;
-                    inLongRange = false;
-                    range = rng.NextInt64(1000, short.MaxValue);
-                    data[i++] = rng.NextInt64(range - 100, range + 100);
-                }
-                else if (rand < 0.75)
-                {
-                    inSmallRange = false;
-                    inShortRange = false;
-                    inLargeRange = true;
-                    inLongRange = false;
-                    range = rng.NextInt64(1000, int.MaxValue);
-                    data[i++] = rng.NextInt64(range - 1000, range + 1000);
-                }
-                else
-                {
-                    inSmallRange = false;
-                    inShortRange = false;
-                    inLargeRange = false;
-                    inLongRange = true;
-                    range = rng.NextInt64(10000, long.MaxValue);
-                    data[i++] = rng.NextInt64(range - 10000, range + 10000);
 
+            // pick a start. Aim for gaps between runs by either taking a random value
+            // or basing on previous end + gap. Try a few times to avoid accidental small gaps.
+            long start = 0;
+            long attemptUpper = maxValue - runSize; // inclusive exclusive handled by NextInt64
+            if (attemptUpper < minValue) attemptUpper = minValue;
+
+            bool picked = false;
+            for (int attempt = 0; attempt < 10 && !picked; attempt++)
+            {
+                // decide whether to use a jump based on minGap/maxGap or pick random
+                if (prevEnd != long.MinValue && rng.NextDouble() < 0.7)
+                {
+                    // generate a gap and place start after prevEnd + gap
+                    long gap = rng.NextInt64(minGap, Math.Max(minGap + 1, maxGap));
+                    long candidate = prevEnd + gap;
+                    // if candidate within allowed bounds adjust to fit
+                    if (candidate >= minValue && candidate <= attemptUpper)
+                    {
+                        start = candidate;
+                        picked = true;
+                        break;
+                    }
+                }
+
+                // fallback: pick random start in allowed bounds
+                start = rng.NextInt64(minValue, attemptUpper + 1);
+                // avoid being too close to previous run end if present
+                if (prevEnd == long.MinValue || Math.Abs(start - prevEnd) >= minGap)
+                {
+                    picked = true;
+                    break;
                 }
             }
 
+            if (!picked)
+            {
+                // final fallback: clamp to bounds
+                start = Math.Max(minValue, Math.Min(attemptUpper, prevEnd + minGap));
+            }
+
+            // fill the run with consecutive values, careful with overflow
+            for (int j = 0; j < runSize; j++)
+            {
+                long val;
+                try
+                {
+                    checked
+                    {
+                        val = start + j;
+                    }
+                }
+                catch (OverflowException)
+                {
+                    // clamp if overflow occurs
+                    val = (start >= 0) ? long.MaxValue - (runSize - j - 1) : long.MinValue + (runSize - j - 1);
+                }
+
+                data[idx++] = val;
+            }
+
+            prevEnd = data[idx - 1];
         }
 
         return data;
     }
 
+
     // Generate random int array of given length and distribution
-    public static int[] GenerateInt32(int length, string pattern = "uniform", 
+    public static int[] GenerateInt32(int length, GeneratorPattern pattern = GeneratorPattern.Uniform, 
         int range = int.MaxValue)
     {
         var data = new int[length];
 
-        switch (pattern.ToLower())
+        switch (pattern)
         {
-            case "uniform":
+            case GeneratorPattern.Uniform:
                 // Random values in [-range, range]
                 for (int i = 0; i < length; i++)
                     data[i] = rng.Next(-range, range);
                 break;
 
-            case "positive":
+            case GeneratorPattern.Positive:
                 for (int i = 0; i < length; i++)
                     data[i] = rng.Next(0, range);
                 break;
 
-            case "negative":
+            case GeneratorPattern.Negative:
                 for (int i = 0; i < length; i++)
                     data[i] = -rng.Next(0, range);
                 break;
 
-            case "alternating":
+            case GeneratorPattern.Alternating:
                 for (int i = 0; i < length; i++)
                 {
                     int val = rng.Next(0, range);
@@ -116,14 +151,14 @@ public static class IntArrayGenerator
                 }
                 break;
 
-            case "small":
+            case GeneratorPattern.Small:
                 // Focused on small magnitudes to test ZigZag fast path
                 for (int i = 0; i < length; i++)
                     data[i] = rng.Next(-64, 65);
                 break;
 
 
-            case "ascending":
+            case GeneratorPattern.Ascending:
                 {
                     int start = rng.Next(-range, range);
                     for (int i = 0; i < length; i++)
@@ -247,30 +282,30 @@ public static class IntArrayGenerator
     }
 
     // Generate random int array of given length and distribution
-    public static long[] GenerateInt64(int length, string pattern = "uniform", 
+    public static long[] GenerateInt64(int length, GeneratorPattern pattern = GeneratorPattern.Uniform, 
         long range = long.MaxValue)
     {
         var data = new long[length];
 
-        switch (pattern.ToLower())
+        switch (pattern)
         {
-            case "uniform":
+            case GeneratorPattern.Uniform:
                 // Random values in [-range, range]
                 for (int i = 0; i < length; i++)
                     data[i] = rng.NextInt64(-range, range);
                 break;
 
-            case "positive":
+            case GeneratorPattern.Positive:
                 for (int i = 0; i < length; i++)
                     data[i] = rng.NextInt64(0, range);
                 break;
 
-            case "negative":
+            case GeneratorPattern.Negative:
                 for (int i = 0; i < length; i++)
                     data[i] = -rng.NextInt64(0, range);
                 break;
 
-            case "alternating":
+            case GeneratorPattern.Alternating:
                 for (int i = 0; i < length; i++)
                 {
                     var val = rng.NextInt64(0, range);
@@ -278,14 +313,14 @@ public static class IntArrayGenerator
                 }
                 break;
 
-            case "small":
+            case GeneratorPattern.Small:
                 // Focused on small magnitudes to test ZigZag fast path
                 for (int i = 0; i < length; i++)
                     data[i] = rng.NextInt64(-64, 65);
                 break;
 
 
-            case "ascending":
+            case GeneratorPattern.Ascending:
                 {
                     var start = rng.NextInt64(-range, range);
                     for (int i = 0; i < length; i++)
@@ -300,29 +335,29 @@ public static class IntArrayGenerator
         return data;
     }
 
-    public static short[] GenerateInt16(int length, string pattern = "uniform", 
+    public static short[] GenerateInt16(int length, GeneratorPattern pattern = GeneratorPattern.Uniform, 
         short range = short.MaxValue)
     {
         var data = new short[length];
 
-        switch (pattern.ToLower())
+        switch (pattern)
         {
-            case "uniform":
+            case GeneratorPattern.Uniform:
                 for (int i = 0; i < length; i++)
                     data[i] = (short)rng.Next(-range, range + 1);
                 break;
 
-            case "positive":
+            case GeneratorPattern.Positive:
                 for (int i = 0; i < length; i++)
                     data[i] = (short)rng.Next(0, range + 1);
                 break;
 
-            case "negative":
+            case GeneratorPattern.Negative:
                 for (int i = 0; i < length; i++)
                     data[i] = (short)(-rng.Next(0, range + 1));
                 break;
 
-            case "alternating":
+            case GeneratorPattern.Alternating:
                 for (int i = 0; i < length; i++)
                 {
                     short val = (short)rng.Next(0, range + 1);
@@ -330,13 +365,13 @@ public static class IntArrayGenerator
                 }
                 break;
 
-            case "small":
+            case GeneratorPattern.Small:
                 for (int i = 0; i < length; i++)
                     data[i] = (short)rng.Next(-64, 65);
                 break;
 
 
-            case "ascending":
+            case GeneratorPattern.Ascending:
                 {
                     short start = (short)rng.Next(-range, range);
                     for (int i = 0; i < length; i++)

Tests/Serialization/Gvwie/IntArrayRunner.cs

@@ -21,71 +21,151 @@ namespace Esiur.Tests.Gvwie
         public virtual IList<T>? Values { get; set; }
     }
 
+
     internal class IntArrayRunner
     {
         public void Run()
         {
+
+            const int TEST_ITERATIONS = 100;
+            const int SAMPLE_SIZE = 100;
+
             Console.WriteLine(";Esiur;FlatBuffer;ProtoBuffer;MessagePack;BSON;CBOR;Avro,Optimal");
 
 
 
-            //Console.Write("Cluster (Int32);");
+            Console.Write("Cluster (Int32);");
-            ////CompareInt(int32cluster);
-            //Average(() => CompareInt(IntArrayGenerator.GenerateInt32Run(1000)), 1000);
 
+            PrintAverage(
+                Average(() => CompareInt(IntArrayGenerator.GenerateRuns(SAMPLE_SIZE)), TEST_ITERATIONS)
+            );
 
             Console.Write("Positive (Int32);");
-            Average(() => CompareInt(IntArrayGenerator.GenerateInt32(1000, "positive")), 1000);
+
+            PrintAverage(
+                Average(() => CompareInt(IntArrayGenerator.GenerateInt32(SAMPLE_SIZE, GeneratorPattern.Uniform)), TEST_ITERATIONS)
+            );
 
             Console.Write("Negative (Int32);");
-            Average(() => CompareInt(IntArrayGenerator.GenerateInt32(1000, "negative")), 1000);
+            PrintAverage(
-
+                Average(() => CompareInt(IntArrayGenerator.GenerateInt32(SAMPLE_SIZE, GeneratorPattern.Negative)), TEST_ITERATIONS)
+            );
 
             Console.Write("Small (Int32);");
-            Average(() => CompareInt(IntArrayGenerator.GenerateInt32(1000, "small")), 1000);
+            PrintAverage(
-
+                Average(() => CompareInt(IntArrayGenerator.GenerateInt32(SAMPLE_SIZE, GeneratorPattern.Small)), TEST_ITERATIONS)
+            );
             // CompareInt(int32small);
 
             Console.Write("Alternating (Int32);");
             //CompareInt(int32alter);
 
-            Average(() => CompareInt(IntArrayGenerator.GenerateInt32(1000, "alternating")), 1000);
+            PrintAverage(
-
+                Average(() => CompareInt(IntArrayGenerator.GenerateInt32(SAMPLE_SIZE, GeneratorPattern.Alternating)), TEST_ITERATIONS)
+            );
 
             Console.Write("Ascending (Int32);");
-            //CompareInt(int32asc);
-            Average(() => CompareInt(IntArrayGenerator.GenerateInt32(1000, "ascending")), 1000);
 
+            PrintAverage(
+                Average(() => CompareInt(IntArrayGenerator.GenerateInt32(SAMPLE_SIZE, GeneratorPattern.Ascending)), TEST_ITERATIONS)
+            );
 
             Console.Write("Int64;");
-            Average(() => CompareInt(IntArrayGenerator.GenerateInt64(1000, "uniform")), 1000);
+
-            //CompareInt(int64Uni);
+            PrintAverage( 
+                Average(() => CompareInt(IntArrayGenerator.GenerateInt64(SAMPLE_SIZE)), TEST_ITERATIONS)
+            );
 
             Console.Write("Int32;");
-            //CompareInt(int32Uni);
+
-            Average(() => CompareInt(IntArrayGenerator.GenerateInt32(1000, "uniform")), 1000);
+            PrintAverage(
+                Average(() => CompareInt(IntArrayGenerator.GenerateInt32(SAMPLE_SIZE)), TEST_ITERATIONS)
+            );
 
             Console.Write("Int16;");
-            //CompareInt(int16Uni);
-
-            Average(() => CompareInt(IntArrayGenerator.GenerateInt16(1000, "uniform")), 1000);
 
+            PrintAverage(
+                Average(() => CompareInt(IntArrayGenerator.GenerateInt16(SAMPLE_SIZE)), TEST_ITERATIONS)
+            );
 
             Console.Write("UInt64;");
-            //CompareInt(uint64Uni);
-            Average(() => CompareInt(IntArrayGenerator.GenerateUInt64(1000, "uniform")), 1000);
 
+            PrintAverage(
+                Average(() => CompareInt(IntArrayGenerator.GenerateUInt64(SAMPLE_SIZE)), TEST_ITERATIONS)
+            );
 
             Console.Write("UInt32;");
-            //CompareInt(uint32Uni);
+
-            Average(() => CompareInt(IntArrayGenerator.GenerateUInt32(1000, "uniform")), 1000);
+            PrintAverage(
+                Average(() => CompareInt(IntArrayGenerator.GenerateUInt32(SAMPLE_SIZE)), TEST_ITERATIONS)
+            );
 
             Console.Write("UInt16;");
-            //CompareInt(uint16Uni);
-            Average(() => CompareInt(IntArrayGenerator.GenerateUInt16(1000, "uniform")), 1000);
 
+            PrintAverage(
+                Average(() => CompareInt(IntArrayGenerator.GenerateUInt16(SAMPLE_SIZE)), TEST_ITERATIONS)
+            );
         }
 
+        // Generate CSV suitable for Office Word chart where the sample size varies.
+        // Produces a CSV with header: SampleSize;Esiur;FlatBuffer;ProtoBuffer;MessagePack;BSON;CBOR;Avro;Optimal
+        public void RunChart()
+        {
+            var sizes = new int[] { 10, 100, 1000, 10000, 100000 };
+
+            // Define generators to evaluate. Each entry maps a name to a function that
+            // given a sample size returns the averages (double[]) by calling Average(...).
+            var generators = new List<(string name, Func<int, int, double[]> fn)>()
+            {
+                ("GenerateRuns", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateRuns(size)), iterations)),
+                ("Int32_Uniform", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateInt32(size, GeneratorPattern.Uniform)), iterations)),
+                ("Int32_Negative", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateInt32(size, GeneratorPattern.Negative)), iterations)),
+                ("Int32_Small", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateInt32(size, GeneratorPattern.Small)), iterations)),
+                ("Int32_Alternating", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateInt32(size, GeneratorPattern.Alternating)), iterations)),
+                ("Int32_Ascending", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateInt32(size, GeneratorPattern.Ascending)), iterations)),
+                ("Int64", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateInt64(size)), iterations)),
+                ("Int32_Default", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateInt32(size)), iterations)),
+                ("Int16", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateInt16(size)), iterations)),
+                ("UInt64", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateUInt64(size)), iterations)),
+                ("UInt32", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateUInt32(size)), iterations)),
+                ("UInt16", (size, iterations) => Average(() => CompareInt(IntArrayGenerator.GenerateUInt16(size)), iterations)),
+            };
+
+            foreach (var gen in generators)
+            {
+                var sb = new System.Text.StringBuilder();
+                sb.AppendLine("SampleSize;Esiur;FlatBuffer;ProtoBuffer;MessagePack;BSON;CBOR;Avro;Optimal");
+
+                foreach (var size in sizes)
+                {
+                    // Choose iterations depending on size to keep total runtime reasonable
+                    int iterations;
+                    if (size <= 100) iterations = 1000;
+                    else if (size <= 1000) iterations = 200;
+                    else if (size <= 10000) iterations = 50;
+                    else iterations = 10;
+
+                    Console.WriteLine($"Running {gen.name} sample size={size}, iterations={iterations}...");
+
+                    var averages = gen.fn(size, iterations);
+
+                    sb.Append(size);
+                    for (int i = 0; i < averages.Length; i++)
+                    {
+                        sb.Append(';');
+                        sb.Append(Math.Round(averages[i]));
+                    }
+
+                    sb.AppendLine();
+                }
+
+                var file = $"run_chart_{gen.name}.csv";
+                System.IO.File.WriteAllText(file, sb.ToString());
+                Console.WriteLine($"Chart CSV written to: {file}");
+            }
+        }
+
+
         public static (int, int, int, int, int, int, int, int) CompareInt(long[] sample)
         {
             var intRoot = new ArrayRoot<long>() { Values = sample };
@@ -300,7 +380,7 @@ namespace Esiur.Tests.Gvwie
         }
 
 
-        static (double, double, double, double, double, double, double, double) Average(Func<(int, int, int, int, int, int, int, int)> call, int count)
+        static double[] Average(Func<(int, int, int, int, int, int, int, int)> call, int count)
         {
             var sum = new List<(int, int, int, int, int, int, int, int)>();
 
@@ -308,7 +388,7 @@ namespace Esiur.Tests.Gvwie
                 sum.Add(call());
 
 
-            var rt = (sum.Average(x => x.Item1),
+            var rt = new double[]{ sum.Average(x => x.Item1),
                     sum.Average(x => x.Item2),
                     sum.Average(x => x.Item3),
                     sum.Average(x => x.Item4),
@@ -316,13 +396,36 @@ namespace Esiur.Tests.Gvwie
                     sum.Average(x => x.Item6),
                     sum.Average(x => x.Item7),
                     sum.Average(x => x.Item8)
-                   );
+            };
+
+            Console.WriteLine($"{rt[0]};{rt[1]};{rt[2]};{rt[3]};{rt[4]};{rt[5]};{rt[6]};{rt[7]}");
 
-            Console.WriteLine($"{rt.Item1};{rt.Item2};{rt.Item3};{rt.Item4};{rt.Item5};{rt.Item6};{rt.Item7};{rt.Item8}");
 
             return rt;
         }
 
+        static string PrintAverage(double[] values)
+        {
+            // Determine winner (lowest average size)
+            var names = new string[] { "Esiur", "FlatBuffer", "ProtoBuffer", "MessagePack", "BSON", "CBOR", "Avro", "Optimal" };
+            var min = values.SkipLast(1).Min();
+            var idx = Array.IndexOf(values, min);
+            if (idx >= 0 && idx < names.Length)
+            {
+                if (idx == 0)
+
+                    Console.ForegroundColor = ConsoleColor.Green;
+                else
+                    Console.ForegroundColor = ConsoleColor.Red;
+
+                Console.WriteLine($"Winner: {names[idx]} ({min:F0})");
+                Console.ForegroundColor = ConsoleColor.White;
+
+                return names[idx];
+            }
+
+            return "Unknown";
+        }
 
         public static byte[] SerializeFlatBuffers<T>(ArrayRoot<T> array)
         {

Tests/Serialization/Gvwie/Program.cs

@@ -11,4 +11,5 @@ MessagePack.MessagePackSerializer.DefaultOptions = MessagePackSerializerOptions.
 
 var ints = new IntArrayRunner();
 ints.Run();
+ints.RunChart();