Syntax errors are Good, Actually

I consider the following sentence in the CommonMark spec to be a fatal flaw in its design:

Any sequence of characters is a valid CommonMark document.

Syntax errors are good! People make typos; a good language makes it so that common typos are obviously detectable. Imagine ifevery sequence of characters were a valid English word. The reason it isn’t the case, of course, is that languages that are bad like this are selected away.

Why is CommonMark making this obvious mistake? I don’t understand it. A lack of errors isn’t a kindness to novice users. It’s the other way around!

Incidentally, this is another thing that Typst gets right. You get excellent syntax errors and linting when using Typst. That’s how it should be!

And that’s what we’re going to try in Quarto Markdown. A syntax that’s as close to Markdown as we can make it, but for which you get proper syntax errors.

Consider Example 307 from the GFM spec:

Input: `hi`lo`

Output: <p><code>hi</code>lo`</p>

This really should be a syntax error. Mismatched backquotes should be disallowed, and “naked” backquotes should be escaped (likely with a backslash). I’m not sure how many of these classes of examples we will be able to convert, but it should give you an idea