Clarify whether loose lists are transitive

[davidje13]

In the current spec:

A list is loose if any of its constituent list items are separated by blank lines, or if any of its constituent list items directly contain two block-level elements with a blank line between them. Otherwise a list is tight. (The difference in HTML output is that paragraphs in a loose list are wrapped in <p> tags, while paragraphs in a tight list are not.)

This is ambiguous in the case of nested lists (specifically whether "directly contain two block-level elements with a blank line between them" should apply to the sub-list items, or only to the sub-list as a whole), and there is no test case for it.

An example input:

- foo
- bar
  - zig

  - zag

generated output from "dingus":

<ul>
  <li>foo</li>
  <li>
    bar
    <ul>
      <li>
        <p>zig</p>
      </li>
      <li>
        <p>zag</p>
      </li>
    </ul>
  </li>
</ul>

alternative interpretation:

<ul>
  <li>
    <p>foo</p>
  </li>
  <li>
    <p>bar</p>
    <ul>
      <li>
        <p>zig</p>
      </li>
      <li>
        <p>zag</p>
      </li>
    </ul>
  </li>
</ul>

It is not entirely clear if this is intended or not, and some implementations which try to conform to commonmark take the approach of making both lists loose (among other partial outputs which are unlikely to be intentional)

What is the intended behaviour here? Can it be clarified / added as an example?

[taufik-nurrohman]

[...] contain two block-level elements with a blank line between them [...]

It does not contain a blank line between the text (to be a paragraph) and the list block. Try:

- foo
- bar

  - zig

  - zag

[vassudanagunta]

No ambiguity because "directly" means immediate children, not children of children, and therefore not blank lines between sub list items. As confirmed by the dingus output.

[davidje13]

No ambiguity because "directly" means immediate children, not children of children, and therefore not blank lines between sub list items. As confirmed by the dingus output.

I guess that depends on whether you're looking at it from the perspective of the input markdown or the output HTML;

In the markdown, the direct children are the list items themselves (because markdown doesn't have a syntax for the containing list; a list is just defined as a sequence of list items)

A list is a sequence of one or more list items of the same type

In the HTML, there is of course a <ul> or <ol> wrapper, meaning the list items are no-longer direct children. Though not all markdown implementations generate HTML.

But it sounds as though the dingus behaviour is the intended one. I think an explicit example / test case here would be beneficial, since it has apparently tripped up a few implementations already.

[vassudanagunta]

I guess that depends on whether you're looking at it from the perspective of the input markdown or the output HTML;

Since this is a spec for Markdown syntax, its perspective is the input Markdown.

But I can see that this subtlety can be easily missed, and it does make sense that a test case exists to enforce it. I'm sure a PR would be most welcome by @jgm. It would take fewer lines than this discussion ;)