TheOmenDen.Shared 2022.11.13.2149

The Omen Den Shared Library

• This is just a grouping of common classes that are used within every application that The Omen Den aims to provide as a company, and is free to modify, redistribute, and use elsewhere

• Especially since this library is sure to not be unique to us.

• Attributions for credits on each release are defined below

<br /><br />

Goals that we aim for:

1. Relevant extensions across The Omen Den's software applications
2. Exception calling, and custom exceptions
3. Pooling extensions for StringBuilder and Arrays
   • Hopefully working towards far less overhead than constantly newing these types up
   • In the case of arrays, adding "slicing" SubArray functionality
4. Custom Enumerations
   • Providing ways to grade exceptions via gravity
   • Providing ways to use "smarter" enumerations for differing control flow
   • Provides two structs Conditions and Consequences that can be chained together against the EnumerationBase implementations
   • Inspired by:
     1. The Smart Enumerations Library
     2. This article by Jimmy Bogard
5. Provides simple LINQ methods for ReadOnlySpan<T>
6. Base Record types for QueryStrings
7. AsyncLazyInitializer type sourced from: Microsoft Dev Blogs
8. Extensions on the Type Type to allow for easier retrieval of ancestors
9. Provides Progress Bars that can also be threadsafe, as well as a simple progress indicator.
   • Gives the caller the ability to specify a change in the ConsoleColor with any one of the provided colors.
   • Provides a relatively smooth, and rate-adjustable animation on the progress indicator
10. Provides Guard methods that can be used to help control the flow and validtion of models and creation of various components in an application domain.
    • Inspired By:
      1. Guard.Net
      2. This video by Derek Comartin of Code Opinion
      3. This article by Stavros Kasidis

Async Stream Handling Features

1. Provided in the AsynchronousStreamOutcome set of extensions are a few methods for capturing OperationOutcome during the iterations of an IAsyncEnumerable<T>
   • The T in question must implement our IEntity interface.
   • We aim to work solely with asynchronous iterations provided by an await foreach() pattern
   • We also aim to provide a simple, and easy to interpret API for further processing and alignment by avoiding an underlying try...catch within the Async Iterator.
2. Our current implementation allows for the simplicity of just providing a long running delegate of an individual operation. Func<in T, in CancellationToken, out ValueTask<OperationOutcome>>
   • This guarantees that the underlying operation is asynchronous
   • This also ensures that the operation returns an OperationOutcome object
3. With this provided, our intent on design is twofold -
   • Firstly We remove the basis for throwing an exception over the stream.
   • Secondly We allow for individual failures to occur during the stream.
     • With this approach, we aim to allow for multiple failures to occur during the stream, while maintaining a consistent reporting behavior to allow for a more streamlined client experience.
     • We also aim for the Tuple<T,OperationOutcome> coupling to be a launching strategy for further processing.

EnumerationBase

1. These Record Types are implemented as such:
   • EnumerationBase<TKey>
     • Similar to the standard Enumeration Enum type that you will see in C# but provides the ability to work with Condtions (When) and Consequences (Then) allowing for the expansion of logic ontop of Enumerations.
     • We provide a way to customize this further by using EnumrationBase<TKey, TEnumeration> to define a customized "backing field" for your own design.
     • We also provide guards, and relevant exceptions to assist in runtime "Early Swift Failure" practices.
   • EnumerationFlagBase<TKey>
     • This record allows for your enumerations to be extended to have a similar functionality as working with bit flags on your standard C# enumerations.
     • This is further enhanced by the use of our Condtions (When) ... Consequences (Then) pattern - allowing for more robust and well defined logic.
     • Strictly enforced by makign the definition of the enum require an integer backing field using powers of two (2ⁿ)
       1. 2,4,8,16... 2ⁿ

Guards

1. We have created a Guard partial class system
   • Our aim/intention with this is to allow for a "fail early" strategy at your domain's edge.
   • We encourage you NOT to use these classes within your business logic/application domain since that seems to be an anti-pattern.
   • There are also differing typed exceptions that are thrown per condition, as well as provided exception message templates.
   • If you desire to define your own custom Guard - You would want to work with the Guard.FromCondition method. This allows for the passing of a delegate that has a boolean result, and an Exception object as an outcome.
     1. e.g. Guard.FromCondition(Func<Boolean>, Exception exception)

Random Generations

1. The need for generating random values across threads has been shown for us at The Omen Den - especially as we work on Crows Against Humility
   • Our aim/intention was to provide a thread safe implementation of .NET's Random.
   • We of course will credit Alexey from [this stack overflow] post for the basis of our design and documentation - it is important to continue to recognize where our knowledge comes from.
   • ThreadSafeRandom implements both IDisposable and IAsyncDisposable to allow for the destruction/freeing of resources used by ThreadLocal
   • We don't use a locking mechanism here, to avoid the complications and relative expense
   • However, we also recognized that by using a static implementation, there could be complications from "newing" up an instance
2. We have also incorporated this approach in our Linq implementations of random element retrieval.

   • This article by Paul Scharf provided enough basis and inspiration for our approach

   • With the ability to provide some basic optimizations for the following implementations: ICollection<T> T[] and List<T> - we were able to see linear performance, as well as reduced allocations

   • We also have the methods for Arrays (T[]) and Lists List<T> as separate methods with the types appended to their suffix. (e.g. GetRandomElementFromArray)

   • We split up our methods into two main ideas

     1. GetRandomElement(s) - To retrieve the actual TValue from the supplied source collection
     2. GetRandomIndex(Indicies)- To retrieve an Index struct for further usage from the supplied source collection.

   • We do also have benchmarks available, but we also believe that O(n) (mentioned in the above article) - was probably the most optimized we could get. The most optimizations we retrieved were from space saving methods

   • A few Benchmarks, as generated by Benchmark.NET (small array - 100 ints, medium array 100,000 ints, large array 1,000,000 ints)

<br/>

TODOs:

1. Provide better source generation for our generic implementations
2. Provide clearer, and more accessible documentation
3. Add .NET 7 and .NET Standard build targets.
4. General clean-up, and optimizations