Leo White

> `.mli` files are subject to pre-processing. That's the reason why we have `cmti` files no ? Sure, and I would expect that a tool to run the examples in...

I should say though, that I think any proposal is going to require the syntax to allow code blocks to be labelled in some way -- so that part seems...

Sure, but how do I map the results of using `hello.ml` back to the `example.mld` file afterwards? We lose the mapping between the comments and the code so you can't...

Perhaps I should clarify what I mean by an "expect test" workflow. I think that one of the most useful aspects of being able to run examples from documentation is...

As examples this is how the [mdx](https://github.com/realworldocaml/mdx) and [toplevel_expect_tests](https://github.com/janestreet/toplevel_expect_test) tools work.

> Well I don't expect `odoc` to make the whole promotion business. Me neither. I don't think `odoc` needs to provide support for any of this. As long as its...

I can see how this proposal would be better than what I'm suggesting for literate programming, is that what you have in mind? In which case, would only having the...

> If I take these [examples](https://erratique.ch/software/uutf/doc/Uutf#examples) I'm not really interested in `topexpecting` them The expect test style approach wouldn't give you anything extra, but it would give you what you...

I just want to discourage people from trying to use it to implement the execution of code in comments in mli files since I don't think it is the best...

Thanks for the design doc @rizo that makes your plan much clearer. I think the aim is a good one but in general I think that the proposed `extract-code` command...