Optimize SimpleInjector processor dispatch

[Copilot]

Eliminates per-dispatch MakeGenericType and dynamic binder work in Solid SimpleInjector processors while preserving handler resolution and decorators. Task 001 is documented with benchmark results and moved to completed.

• Dispatch cache
  • Cache immutable compiled invokers by runtime message type.
  • Resolve handlers from SimpleInjector for every call; scoped/transient lifetimes and decorator chains remain intact.

return await QueryHandlerInvokerCache<TResult>
    .ExecuteAsync(_container, query, ctk)
    .ConfigureAwait(false);

• Coverage and measurement

  • Add decorated handler, cancellation, and exception behavior tests.
  • Add BenchmarkDotNet request/query/command comparisons.
  • Record 77.8–82.8% lower mean dispatch time and reduced allocations.

• Generator decision

  • Document why source generator/interceptor dispatch was not added for existing interface-based processor calls.

[Copilot]

Pull request overview

Optimizes Ark.Tools.Solid.SimpleInjector processor dispatch by caching compiled handler invokers per runtime message type, avoiding per-call MakeGenericType and dynamic binder overhead while preserving per-dispatch handler resolution (scopes/transients/decorators).

Changes:

• Replace dynamic/per-call reflection dispatch with compiled invoker caches for request/query/command processors.
• Add MSTest coverage for decorator execution order plus cancellation/exception propagation.
• Add a BenchmarkDotNet benchmark project and move the performance task doc from pending to completed (with recorded results).

Reviewed changes

Copilot reviewed 15 out of 15 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
tests/Ark.Tools.Solid.SimpleInjector.Tests/SimpleInjectorProcessorTests.cs	New behavioral tests for decorated handlers + cancellation/exception behavior.
tests/Ark.Tools.Solid.SimpleInjector.Tests/Ark.Tools.Solid.SimpleInjector.Tests.csproj	New test project wiring for Solid.SimpleInjector.
tests/Ark.Tools.Solid.SimpleInjector.Tests/packages.lock.json	Lockfile for the new test project.
src/common/Ark.Tools.Solid.SimpleInjector/SimpleInjectorRequestProcessor.cs	Switch request dispatch to cached compiled invoker.
src/common/Ark.Tools.Solid.SimpleInjector/SimpleInjectorQueryProcessor.cs	Switch query dispatch to cached compiled invoker.
src/common/Ark.Tools.Solid.SimpleInjector/SimpleInjectorCommandProcessor.cs	Switch command dispatch to cached compiled invoker.
src/common/Ark.Tools.Solid.SimpleInjector/HandlerInvokerCaches.cs	New compiled-invoker cache implementation.
benchmarks/Ark.Tools.Benchmarks/Ark.Tools.Benchmarks.csproj	New BenchmarkDotNet benchmark project.
benchmarks/Ark.Tools.Benchmarks/Program.cs	BenchmarkDotNet entrypoint.
benchmarks/Ark.Tools.Benchmarks/ProcessorDispatchBenchmarks.cs	Benchmarks comparing reflection/dynamic vs cached invoker dispatch.
benchmarks/Ark.Tools.Benchmarks/packages.lock.json	Lockfile for the new benchmark project.
docs/performance/pending/001-solid-simpleinjector-dispatch.md	Remove pending task doc (moved to completed).
docs/performance/completed/001-solid-simpleinjector-dispatch.md	Completed task doc with benchmark summary + decision record.
Directory.Packages.props	Add BenchmarkDotNet version pin.
Ark.Tools.slnx	Include new benchmark + new test project in the solution.

Suppressed comments (4)

src/common/Ark.Tools.Solid.SimpleInjector/HandlerInvokerCaches.cs:59

• This ExecuteAsync method returns the Task directly. Repo guidance requires async methods to use async/await (even for a single await) for stacktrace clarity in production debugging.

    public static Task<TResponse> ExecuteAsync(Container container, IRequest<TResponse> request, CancellationToken cancellationToken)
    {
        var invoker = _invokers.GetOrAdd(request.GetType(), static requestType => CreateInvoker(requestType));
        return invoker(container, request, cancellationToken);
    }

src/common/Ark.Tools.Solid.SimpleInjector/HandlerInvokerCaches.cs:96

• This ExecuteAsync method returns the Task directly. Repo guidance requires async methods to use async/await (even for a single await) for stacktrace clarity in production debugging.

    public static Task ExecuteAsync(Container container, ICommand command, CancellationToken cancellationToken)
    {
        var invoker = _invokers.GetOrAdd(command.GetType(), static commandType => CreateInvoker(commandType));
        return invoker(container, command, cancellationToken);
    }

benchmarks/Ark.Tools.Benchmarks/ProcessorDispatchBenchmarks.cs:78

• The reflection/dynamic benchmark path awaits without ConfigureAwait(false), while the cached path uses it. Keeping both consistent avoids context-capture noise and matches the repo’s async style.

        var handlerType = typeof(IRequestHandler<,>).MakeGenericType(_request.GetType(), typeof(int));
        dynamic handler = _container.GetInstance(handlerType);
        return await handler.ExecuteAsync((dynamic)_request);
    }

benchmarks/Ark.Tools.Benchmarks/ProcessorDispatchBenchmarks.cs:92

• The reflection/dynamic benchmark path awaits without ConfigureAwait(false), while the cached path uses it. Keeping both consistent avoids context-capture noise and matches the repo’s async style.

        var handlerType = typeof(ICommandHandler<>).MakeGenericType(_command.GetType());
        dynamic handler = _container.GetInstance(handlerType);
        await handler.ExecuteAsync((dynamic)_command);
    }

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.