gaweph / memoizer Goto Github PK

Thanks!

I have this test to illustrate the problem. I assumed calling the same method on a different class would give different results.

While this can be interesting in some scenarios, I consider that caching scope feel more natural in a per-class fashion.

public class MemoizerTests
{
    [Fact]
    public void Cached_method_in_different_instances_should_return_different_results()
    {
        var i1 = new MyClass("hello");
        var i2 = new MyClass("how are you?");

        var one = i1.Get();
        var two = i2.Get();
        Assert.NotEqual(one, two);
    }

    class MyClass
    {
        private string payload;

        public MyClass(string payload)
        {
            this.payload = payload;
        }

        [Cache]
        public string Get() => payload;
    }
}

Does it make sense to create an option in the CacheAttribute to choose the scope level?

• per-type
• per-instance

Python has had @cached decorator for a long time and I just tried it. Sadly, when running unit tests which send different values through cached functions, I learned that unit tests suites typically all share the same process, and in the case of python, the same "lifetime of the application". So basically, @cache makes a a function a singleton/global, which is a problem for unit tests (both parallel and sequential were affected).

Does that all apply to this library as well?