Closed nrnrnr closed 8 months ago
I think "like pattern matching" means you can compile it to a decision tree.
Writing this here so I don't forget, and so you see it.
I suspect this property will be difficult to formalize in a useful way. But I will look forward to seeing what you have by Thursday.
FYI, github is rendering the $\;$ as a semicolon: https://github.com/rogerburtonpatel/vml/blob/main/like-pattern-matching.md
Other comments:
I'm not seeing the cycle among $x, y, z$. I'm seeing "x depends on y; x depends on z". But those equations aren't let
bindings. What are you seeing for dependencies?
According to Figure 1 in the Verse paper, a logical variable can be a value. So your value_bound_var
doesn't create the restriction you are hoping for.
The draft is still from formal. And I don't yet grasp the idea well enough to be confident that it is ready to formalize. In particular, I worry that "dependency graph" is a complex concept. Perhaps there is a simpler piece of spaghetti to pull on?
FYI, github is rendering the ; as a semicolon: https://github.com/rogerburtonpatel/vml/blob/main/like-pattern-matching.md
Gosh darn it. Missed this. Just want my latex \; spaces. Thanks for the heads-up.
Other comments:
* I'm not seeing the cycle among x,y,z. I'm seeing "x depends on y; x depends on z". But those equations aren't `let` bindings. What are you seeing for dependencies?
I see a logical variable $x$ that depend on tuples containing logical variables $y, z$. Those variables, in turn, don't depend on external values. They're only related to $x$. I agree that I need to formalize this idea. Does what I'm getting at- that logical variables that depend only on logical variables with no overall scrutinee is what creates logical-variables-at-runtime scenarios?
* According to Figure 1 in the Verse paper, a logical variable can be a value. So your `value_bound_var` doesn't create the restriction you are hoping for.
Yep. I'll fix this.
* The draft is still from formal. And I don't yet grasp the idea well enough to be confident that it is ready to formalize. In particular, I worry that "dependency graph" is a complex concept. Perhaps there is a simpler piece of spaghetti to pull on?
I agree. I'll think on this more. Thank you for your sincerity.
Revising tomorrow with translations.
Pick something that might mean "like pattern matching" and try to formalize it. Issue #4 is an option but not the only option.