rmc00 / maybe Goto Github PK

Environment:

• Windows 10
• Visual Studio 2017
• C# project with framework 4.6.2
• Maybe.Net v1.0.77

Error Message: (Sorry for Chinese locale, the main meaning is the package not compatible with net462)

正在还原 D:\GitProjects\BigsetExisting\BigsetExisting\BigsetExisting.csproj 的包...
包 Maybe.NET 1.0.77 与 net462 (.NETFramework,Version=v4.6.2) 不兼容。 包 Maybe.NET 1.0.77 支持: release (Release,Version=v0.0)
包 Maybe.NET 1.0.77 与 net462 (.NETFramework,Version=v4.6.2) / win 不兼容。 包 Maybe.NET 1.0.77 支持: release (Release,Version=v0.0)
包 Maybe.NET 1.0.77 与 net462 (.NETFramework,Version=v4.6.2) / win-x64 不兼容。 包 Maybe.NET 1.0.77 支持: release (Release,Version=v0.0)
包 Maybe.NET 1.0.77 与 net462 (.NETFramework,Version=v4.6.2) / win-x86 不兼容。 包 Maybe.NET 1.0.77 支持: release (Release,Version=v0.0)
程序包还原失败。正在回滚“BigsetExisting”的程序包更改。
已用时间: 00:00:00.2589401

Reproduce steps:

1. New Console project with net framework 4.6.2
2. Manage NuGet package
3. Find & Add "Maybe.Net 1.0.77"

I saw, that most methods already have XML Documentation written for them.
Unfortunately, they're not distributed via NuGet. Doing this would make it much easier for developers to use this library:

https://stackoverflow.com/questions/5205738/how-do-you-include-xml-docs-for-a-class-library-in-a-nuget-package

Hi - thanks for creating a library for Bloom filters (with a great API!)

I have run some experiments using your scalable Bloom filters, but they do not seem to perform very well :(

I created a fork containing code for benchmarking the use cases I want to use your library for, as well as experiments with optimizing some of your code.

One performance bottleneck can be found in your choice of method for converting objects to bytes, in order to hash the bytes. I have created a project for benchmarking various methods for the conversion, showing that the BinaryFormatter used in your library performs horribly compared to the alternatives.

The benchmarks in my fork shows the performance improvements achievable by replacing BinaryFormatter with alternatives (I have included the benchmarking results at the bottom of this issue - notice the memory usage column on the far right)

Despite the new optimizations, the memory usage of your Bloom filters (esspecially the scalable version, which I really want to use) is very high compared to an alternative like a HashSet.

Is this just the nature of the implementation, or can it be improved?

(The table below is a part of the output of the benchmarking code. In the tabl, your original implementation of a scalable Bloom filter is called ScalableBloomFilter.)

BenchmarkDotNet=v0.11.5, OS=macOS Mojave 10.14.5 (18F203) [Darwin 18.6.0]
Intel Core i7-8850H CPU 2.60GHz (Coffee Lake), 1 CPU, 12 logical and 6 physical cores
.NET Core SDK=2.2.300
  [Host] : .NET Core 2.2.5 (CoreCLR 4.6.27617.05, CoreFX 4.6.27618.01), 64bit RyuJIT
  Core   : .NET Core 2.2.5 (CoreCLR 4.6.27617.05, CoreFX 4.6.27618.01), 64bit RyuJIT

Job=Core  Runtime=Core  

i fix it this way :

        public IEnumerator<T> GetEnumerator()
        {
            var currentNode = _headNode.Next[0];
            while (currentNode != null && currentNode.HasNextAtLevel(0))
            {
                yield return currentNode.Value;
                currentNode = currentNode.Next[0];
            }

            if (currentNode != null)
            {
                yield return currentNode.Value;
            }

        }