Design patterns for composability with traits (i.e. typeclasses)?

[shelby3]

@keean

@shelby3

The motivating example is as I had stated upthread, that any time one needs a heterogeneous collection and they want composability to be open in both directions of the Expression Problem, i.e. they can call any function that the union of data types implements on the required trait. Whereas, without my proposal then there can't be dependency injection in terms of which traits a collection's element type (aka upthread ValueType) can implement. It is really elementary and simple. And I am quite perplexed why no one in computer science has entertained such an idea and solution.

This is an abstract motivation, not a concrete motivating example. You need to provide examples of real world programs that people are implementing and that are running into problems with this. For example pick a real problem like implementing a chess player, or a painting program, or something, anything, and start implementing it. When you find you cannot do something, or find that something is really ugly because of this problem then you have a motivating example. Either you will have lots of examples from your own programming history where this has caused you actual problems, or you can search the internet and forums and find other developers developing real programs that have found this to be a problem.

How many do you want because the list of use case examples is unbounded.

For example the web browser's DOM has collections and they are typed according to the known element types, but if we want to make a typed interface with a typeclass language (i.e. not the subtyping+subclassing anti-pattern). So if we want to extend on top of the DOM our own data types, but be interoperable with preexisting code, we could not do it without my proposal. Of course currently the DOM is not modeled with a typeclass language structure, so there are no such preexisting code, but I am speaking about the current way to do it once I'm done replacing the web browser (my current project goal which is why I need a new programming language).

Note again that modeling the DOM with the subtyping+subclassing anti-pattern also does not allow for extension on both directions (dimensions) of the Expression Problem.