[cscheid]

Description

Adds support for Quarto 2's div syntax for definition lists, so authors can move to it while still on Quarto 1, ahead of a migration:

::: {.definition-list}
* `time`
  - A timestamp that identifies when the request was initiated.
  - Type: string
* `id`
  - A unique identifier.
:::

Each top-level bullet is a term, and the bullets nested under it are its definitions. Quarto 2 doesn't support Pandoc's term / : definition syntax, and this is what it uses instead.

This follows the list-table precedent (#13528):

• New module src/resources/filters/modules/definitionlist.lua, registered in _quarto.modules.
• New astpipeline-process-definition-lists pass in normalize/astpipeline.lua, right after the list-table pass, so it runs before crossrefs and user filters.

The structural rules match Quarto 2's (is_valid_definition_list_div / transform_definition_list_div in pampa):

• The div contains exactly one bullet list.
• Each item is a term (Plain or Para) followed by a nested bullet list of definitions.
• A div with any other shape is left as is, without a warning.
• The div's id, classes and attributes are dropped, since DefinitionList has no Attr.

Like listtable.lua, it skips the data-pos wrapper divs added by the sourcepos extension.

No flags gate: compute_flags runs after the AST pipeline, so a flag couldn't skip this pass (the same reason has_list_tables is set but unused).

Open questions

1. Malformed divs. list-table stops the render with an error; Quarto 2 leaves a malformed .definition-list div alone silently. This PR matches Quarto 2. A warning in Quarto 1 could help people learning the syntax, but would flag what Quarto 2 accepts.
2. Dropped attributes. ::: {#foo .definition-list} loses #foo, as in Quarto 2. Keeping it would mean wrapping the list in a Div, which differs from Quarto 2.

Testing

• New smoke test tests/docs/smoke-all/markdown/definition-list/definition-list-div.qmd, with html, latex and typst checks for tight, loose and nested lists and for malformed divs staying divs.
• I checked the module directly with pandoc 3.12.1 (pandoc markdown, and commonmark with +sourcepos), and rendered the test document with a dev quarto pointed at this branch's filters through QUARTO_SHARE_PATH; the html, latex and typst assertions match that output.
• I have not yet run ./run-tests.sh on this test or the wider suite.

Documentation: quarto-dev/quarto-web#2305

Checklist

I have (if applicable):

• referenced the GitHub issue this PR closes
• updated the appropriate changelog in the PR
• ensured the present test suite passes
• added new tests
• created a separate documentation PR in Quarto's website repo and linked it to this PR

AI-assisted PR

• AI tool used: Claude Code
• Codebase grounding: local clones of quarto-cli, quarto-web, and the Quarto 2 source
• Human review: draft opened for review; not yet reviewed and verified by a human.

🤖 Generated with Claude Code

[posit-snyk-bot]

✅ Snyk checks have passed. No issues have been found so far.

Status	Scan Engine	Critical	High	Medium	Low	Total (0)
✅	Open Source Security	0	0	0	0	0 issues
✅	Licenses	0	0	0	0	0 issues

💻 Catch issues earlier using the plugins for VS Code, JetBrains IDEs, Visual Studio, and Eclipse.

[cderv]

Regarding CI, we have linux issues due to Koma script update, I'll fix it.

Regarding this PR:

Thanks for this. I ran the new test locally to see, and I also tried a few cases outside the test with 🤖 : a callout and a figure with a crossref inside a definition, a .definition-list div inside a callout, a div with one valid and one invalid item (stays a div), and from: commonmark_x+sourcepos. All of them render as expected.

One difference with Quarto 2 I think, about nesting. With this PR, a .definition-list div inside a definition becomes a nested DefinitionList too, and the #nested case of the test asserts it. In pampa the rewrite does not recurse into its result, on purpose it seems:

source

/// Items 1 and 2 of the set, for one div: a valid list-table becomes a
/// `Table`, a valid definition-list a `DefinitionList`, an invalid
/// list-table warns Q-2-35 into `collector` and stays a div, and any other
/// div is unchanged. Neither rewrite recurses into its result, matching
/// the reader.
...
            if is_valid_definition_list_div(&div) {
                FilterResult(vec![transform_definition_list_div(div)], false)

I rendered the test document with a q2 nightly (0.36.0-nightly.20261022): #tight, #loose and the two malformed divs give the same output as this PR, but in #nested the inner div stays a <div class="definition-list"> with its bullet list. Same for a .list-table div inside a definition: q2 leaves it as a div, while our earlier list-table pass converts it to a table here. As the docs PR says this syntax lets documents work in both versions, which behavior do we want ? Either Quarto 2 recurses, or we keep this PR to one level and drop the #nested case.

About the two open questions:

1. Staying silent on a malformed div like Quarto 2 looks right to me for now. pampa has a FUTURE: issue linter warning comment at each rejected shape, so if Quarto 2 adds those warnings we could add the same ones here then.
2. Dropping the attributes is fine I think, as Quarto 2 does the same and the docs PR says it.

[cscheid]

As the docs PR says this syntax lets documents work in both versions, which behavior do we want ? Either Quarto 2 recurses, or we keep this PR to one level and drop the #nested case.

Good catch. I think we should match whatever Pandoc's syntax. I'll ask the robot.

EDIT: Pandoc parses nested definition lists easily. You found a Quarto 2 bug! We'll fix there: quarto-dev/q2#895.

[cscheid]

@cderv We have a q2 fix incoming, so let's assume the behavior in this PR is intentional. Feel free to merge it, and there's a corresponding doc PR over at quarto-dev/quarto-web#2305