More duck typing and ambiguity

Ideas
1

So I believe in WREC. We Wrote Enough Code.

I propose to introduce a “mode” to the language that allows gradually more ambiguity as more code is written.

Here is what I mean:

# entire code
obj_a.my_collection.add(1)

It is easy to guess that the programmer meant there is an object obj_a and that it likely has set as it is the most commonly used type for the attribute my_collection, which existence is obvious.

The thesis is that this allows to write much less code to get same result, and with much more fun. Resolving the ambiguities is done by a combination of machine learning and automatic proof.

The idea of WREC is to go “all the way duck” to allow constructs (in the future) like

# entire code
a = create_canvas_in_a_browser()
a.draw_a_straight_line()

I’ve been experimenting with this for a while and can say it can be done.

What do you think? Would you use this? Would you join the WREC project if one existed?

2

You mean like GitHub Copilot?

3

Not exactly. I mean yes, analyze code of other people,

but make a gradual change to the language instead of the generate-everything-at-one-shot

So the plan would be like:

  1. allow slightly more ambiguity, like inferring there is a “set” attribute probably initialized as empty from just the use pattern
  2. do some more ambiguity like auto-inferring metaclasses for the expected behavior that we derive from the code
  3. gradually get to the point that we can “infer” what a function means from its name and surrounding code

not the code-generation of copilot. It’s fun to write small code and not fun to debug large code, especially if it was generated

I have a pretty non-zero experience in proof, planning and meta to say it’s absolutely doable.

4

Okay, then I think you should probably write a small prototype implementation and publish a paper about it, right? Because with my limited understanding I have no idea how to do that, and I doubt you’ll get much help from the Python community (or its typing subcommunity) other than skepticism.

4 Likes
5

Sure. That’s exactly what I was thinking (and afraid) about.