I've got a vague idea to use a workflow to for an API. But, I have no idea how to go about it, or if it's even really a good idea.

I'm writing an optimization library, and my goal is to unify the interaction with various strategies (genetic algorithm, simulated annealing, conjugate gradient, etc.). But, the specifics don't matter too much. Essentially, you have these pieces:

1) a function to optimize

2) a strategy to optimize that function

2a) parameters for the strategy

2b) (optional) communication mechanism for main loop (MailboxProcessor)

So, right now:

1
2
3
4
5
6
7
8

let strategy f (alpha, beta) = // ...

let f (v:vector) = // ...

let strategyParams = (1,2)

let answer = strategy f strategyParams

I'd like to do something like this:

1
2
3
4
5
6
7
8
9
10
11
12
13
14

let answer =

optimize{

let! strategy f (alpha, beta) = // ...

let! f (v:vector) = // ...

let! strategyParams = (1,2)

return strategy f strategyParams

}

But, the only examples I find are maybe{} workflows, which don't help too much for more complicated cases:

[link:leibnizdream.wordpress.com]

My question is twofold: First, should I even bother with workflows, or is function composition good enough? Second, how should I go about designing a meaningful workflow?