Mathieu Duponchelle

Actually the correct approach for rendering back to commonmark, now that I think of it a bit more, would simply be to dump back the entire, potentially updated list of...

Hm, that's indeed the use case I've got in mind, I guess it shows :) However as I see it, there would be no obstacle to constructing the entire ast...

Right, so that's also programmatic building of the AST, and you wouldn't a priori care about using libcmark to do html -> html passthrough I guess ;)

To be clear, nothing prevents from having a commonmark output of a programatically built AST using the approach I tried to lay out: * Construct your AST * At render...

btw > Then we could peek into the source to determine things like which emphasis character was used, whether a character was escaped, etc. Was that the sort of thing...

Not sure about that, the problem with this approach is portability. The use case for anchors most people are interested in, afaict, isn't table of contents generation, but in-document linking....

> The pandoc method has the drawback that if you re-order sections in your document, and some sections have the same name, links may break. Right, but there is no...

yay MSVC \o/ I'll have a look at the issues tomorrow, hoping it does not just uncover a new layer :)

Thanks for the review, answering the general comment here: > But that's not what your table and strikethrough extensions do. They define their own block types in the cmark core....

I apologize for the noise here (though github should apologize really), I stick to a rebase workflow because I prefer having a clean history to work with, instead of piling...