Michael Hatherly

Yeah, something to that effect in the readme would be worth adding. In addition probably worth noting other places where interpolation can't work: ``` julia> using CommonMark julia> a =...

Let's adjust CI to run on `1.0 - 1.2` as well here to ensure things are compatible with those versions. I'd like to attempt to retain compat across `1.0 -...

There looks to be some other issues on 1.0-1.2 so I think it's simplest to just restrict compat for `julia` to `1.3` since that's what CI has been testing.

> should not carry state Agreed, stateless if at all possible otherwise we'll just make things too confusing in the long-term. > I do feel that doing this on the...

> I noticed that there is actually some precedent to having the parser functions be stateful closure, with the ji from JuliaInterpolationRule being used in the parser function: Yeah, there's...

That's awesome. I'm quite surprised that commonmark's HTML handling is actually sufficient to parse `htl_output` correctly.

Regarding including it here, I'd like to remove deps where possible since the eventual end goal is a standard lib. Might be a PlutoUI macro, though whether `@md` is actually...

https://github.com/MichaelHatherly/CommonMark.jl/blob/master/src/extensions/interpolation.jl#L121-L152 is the set of extensions that get used in `cm""` which is probably what you'd actually want to use in the "real" `@mdl` macro. It's an internal method so...

Could go with `@mdx` since the syntax supported is _kind of_ like https://mdxjs.com/ though not quite.

If we're inconsistent with https://spec.commonmark.org/dingus/ then we can adjust as needed, but I think we are probably replicating the expected behaviour of the reference implementation here.