File the audits' pre-existing findings, and mark item 43 cited by decisions

AGENTS.md

@@ -10,8 +10,14 @@ In `docs/decisions.md`:
 - Plain markdown is a flavour of the grammar
 - The round-trip is the product
 - Markdown in is a canonical fixpoint
-- Equality is editor-normal
+- Equality is deep
+- `!adf:textBreak{}` parts text CommonMark would join
+- An empty key spells `empty`
+- `-0` is spelled `-0`
+- Empty markdown is a document of no blocks
 - Unknown nodes ride the carry
+- The carry fence names the node type
+- A code block is a fence per text node
 - Foreign HTML sorts three ways
 - Names stay text
 - Directives under `!adf:`
@@ -162,12 +168,16 @@ nearest text, a candidate entry in that file's voice, and the instance it yields
 keeps collecting instances is wrong: rewrite it.
 
 Which output the audience expects — README goal 5 — is settled by a reader panel rather than
-asked: three fresh-context readers, one per README persona the conversion serves, each given only
+asked: three fresh-context readers, one per README persona the question serves, each given only
 `## Audience` and the input, writing what they expect before picking among outputs the goals
 allow, rendered, shuffled, with no rationale and nothing saying what is implemented. Three agreeing
 settle it; otherwise four more read, five of seven settle it, and less is a missing goal, asked.
 The verdict lands in `docs/decisions.md`.
 
+Every new or changed markdown or HTML spelling goes to such a panel, seated by the people who read
+and write that format (README `## Audience`); a question about what an app relies on seats the
+developer personas.
+
 ### Stated numbers
 
 A stated number — 500 levels, the branch floor — is kept; a chunk that cannot keep it asks, naming
@@ -180,3 +190,14 @@ and `todo.md` and trusting them over anything remembered from earlier iterations
 is a thin driver: each chunk's work runs in a fresh-context subagent holding this file as its
 charter, and the driver only relays maintainer questions, runs the review flow, merges, and cleans
 up. The loop stops when only maintainer-reserved acts remain.
+
+## 8. Scoring run
+
+The comprehension panel's fill-ins:
+
+- Language: TypeScript.
+- Kind: a pure-function document converter with hand-written parsers and emitters.
+- Domain: Atlassian Document Format and CommonMark parsing.
+- Domain docs: the CommonMark spec, ADF's JSON schema and `spec/flavour.md`.
+- 3am question: a viewer/editor app reports that a document it saved comes back with two text
+  nodes merged and a mark gone after `markdownToAdf(adfToMarkdown(doc))`.

CHANGELOG.md

@@ -2,10 +2,20 @@
 
 ## Unreleased
 
-- **Breaking:** directives, the opaque carry among them (now `carry`), are spelled under an `!adf:`
+- **Breaking:** directives, the inline opaque carry among them (now `!adf:carry{json="…"}`), are
-  prefix (`!adf:name … !adf:/name`, `!adf:name[content]{attrs}`, `!adf:name arg {attrs}`) in place
+  spelled under an `!adf:` prefix (`!adf:name … !adf:/name`, `!adf:name[content]{attrs}`,
-  of the `:::`/`::`/`:name` forms: text holding an unescaped `!adf:` is claimed, and `adf` is an
+  `!adf:name arg {attrs}`) in place of the `:::`/`::`/`:name` forms, and the block carry is a code
-  ordinary code block language. Convert stored markdown per `MIGRATION.md`.
+  fence whose info string `adf:<type>` names the node's type, its body the node's JSON without
+  `type`: text holding an unescaped `!adf:` and a code fence whose info string opens `adf:` are
+  claimed, and `adf` is an ordinary code block language. Convert stored markdown per `MIGRATION.md`.
+- **Breaking:** `markdownToAdf` and `plainMarkdownToAdf` read markdown holding no block as a
+  document whose `content` is empty, as Atlassian's schema requires; `!adf:doc {content=none}` spells a document holding no
+  `content` key.
+- `markdownToAdf(adfToMarkdown(doc))` deep-equals `doc`: two adjacent text nodes a reader would
+  join are parted by `!adf:textBreak{}`, an empty `attrs`, `content` or `marks` is spelled
+  `{attrs=empty}`, `{content=empty}` or `{marks=empty}`, `-0` is spelled `-0`, and a `codeBlock`
+  of several text nodes is a fence per node. A `codeBlock` holding other than plain text nodes
+  rides the block carry, where it was refused.
 - **Breaking:** `unspellable-link` leaves `ConvertErrorCode`; a link whose `href` or `title` no
   CommonMark escape spells is written as `!adf:link[text]{attrs}`.
 - **Breaking:** some directive refusals carry `malformed-directive` where they carried

MIGRATION.md

@@ -22,7 +22,8 @@ import { markdownToAdf as markdownToAdf010 } from 'adf-codec-0.1'
 
 function migrateMarkdown(stored: string) {
   const parsed = markdownToAdf010(stored)
-  return parsed.ok ? adfToMarkdown(parsed.value) : parsed
+  // 0.1.0's reader was editor-normal, so a document with no content key always meant an empty one.
+  return parsed.ok ? adfToMarkdown({ ...parsed.value, content: parsed.value.content ?? [] }) : parsed
 }
 ```
 
@@ -40,12 +41,12 @@ function migrateMarkdown(stored: string) {
 | `::media {id=a type=file}` | `!adf:media {id=a type=file}` |
 | `::taskItem TODO {localId=i}`: an empty `caption`, `decisionItem`, `paragraph` or `taskItem`, or an empty `heading` carrying `localId` | `!adf:taskItem TODO {localId=i}` then `!adf:/taskItem` |
 | `:mention[@Mikael]{id=5b10a2}` | `!adf:mention[@Mikael]{id=5b10a2}` |
-| the `adf` code fence and `:adf{json="…"}` | the `carry` code fence and `!adf:carry{json="…"}` |
+| the `adf` code fence and `:adf{json="…"}` | the `adf:<type>` code fence, its JSON without `type`, and `!adf:carry{json="…"}` |
 | `\:` keeps a directive literal | `\!adf:` keeps a directive literal |
 | `:adf{json="…"}` carrying a link for its `collection`, `id` or `occurrenceKey` | `!adf:link[text]{attrs}` |
 
 A colon run and `:name[` are plain text now, and `adf` an ordinary code block language; text
-holding an unescaped `!adf:` and a `carry` fence are claimed instead.
+holding an unescaped `!adf:` and a code fence whose info string opens `adf:` are claimed instead.
 
 ### Readings
 
@@ -54,6 +55,8 @@ Markdown the spelling table leaves alone, which `0.2.0` reads as a different doc
 | Input | `0.1.0` | `0.2.0` |
 | --- | --- | --- |
 | a link whose text already holds one (`[a<https://example.com/>b](/v)`) | marks every node the inner link does not, splitting the outer link around it | leaves the outer brackets literal text; write the pieces as separate links to keep them |
+| markdown holding no block (`markdownToAdf("")`) | `{ type: 'doc', version: 1 }` | `{ content: [], type: 'doc', version: 1 }`; `!adf:doc {content=none}` reads as the former |
+| a code fence whose info string opens `adf:` (```` ```adf:x ````) | a `codeBlock` with that language | the block carry; write `!adf:codeBlock {language="adf:x"}` around a bare fence to keep the code block |
 
 ### Error codes
 
@@ -63,6 +66,7 @@ it named converts.
 | Input | `0.1.0` | `0.2.0` |
 | --- | --- | --- |
 | a link whose `href` or `title` no CommonMark escape spells, on emit | `unspellable-link` | spells `!adf:link[text]{attrs}` |
+| a `codeBlock` holding other than plain text nodes, on emit | `unsupported-node-shape` | rides the block carry |
 | a leaf node given a body (`media`, `listBreak`) | `unsupported-node-shape` | `malformed-directive` |
 | a node with a block body written as a leaf (`panel`) | `unsupported-node-shape` | `malformed-directive` |
 | an empty node the `::taskItem` spelling row names, written as a leaf | parses | `malformed-directive` |

README.md

@@ -52,6 +52,10 @@ which is free text.
 - **LLM/agent pipeline** — hands documents to a model as markdown and writes the edits back.
   Relies on the round-trip and on markdown a reader half-knowing the lossless flavour can still edit.
 
+Behind those apps, the people who read and write the markdown, and later the HTML: product
+managers, engineers and support agents working in Atlassian products through a plugin or another
+UI. They know markdown and not ADF, and rely on every spelling saying what it means to them.
+
 ## The shape
 
 ```sh
@@ -171,7 +175,7 @@ Parsing — `markdownToAdf` and `plainMarkdownToAdf`, and `htmlToAdf` at `0.2.0`
 
 | Code | Fires when | What you can do |
 | --- | --- | --- |
-| `malformed-directive` | an `!adf:` the grammar cannot read — a prefix completing no directive, an unclosed container, `[content]` or `{attrs}`, a closer with no container of its name open, a leaf given a body, `{attrs}` out of order or duplicated, invalid JSON in a `carry` | write the spelling the message names, or escape the prefix — `\!adf:`, block and inline alike — to keep it literal text |
+| `malformed-directive` | an `!adf:` the grammar cannot read — a prefix completing no directive, an unclosed container, `[content]` or `{attrs}`, a closer with no container of its name open, a leaf given a body, `{attrs}` out of order or duplicated, invalid JSON in an opaque carry | write the spelling the message names, or keep it literal: escape the prefix — `\!adf:`, block and inline alike — or wrap a code fence whose info string opens `adf:` in `!adf:codeBlock {language="adf:…"}` with a bare fence |
 | `malformed-pipe-table` | a pipe row that is no pipe table — a missing or ragged `---` delimiter row, an alignment colon in it, or a row not opening with a pipe | open every row with a pipe and give the delimiter row the header's cell count; to keep the lines literal text instead, escape the leading pipe of every one — escaping a single row leaves the next to open a fresh table and fail the same way |
 | `unknown-directive-name` | a directive whose name is no node or mark this version spells | check the name in `spec/flavour.md`, or escape the prefix as `\!adf:`; the spelling itself is well formed, so a later minor may give the name meaning |
 | `unmappable-html` | the input holds an HTML construct the documented element set does not map, a comment and a processing instruction among them — at this version that is every raw HTML construct in markdown, the element set landing at `0.2.0` | remove the construct, or write what it holds in the lossless flavour |
@@ -200,14 +204,17 @@ emit refuses:
 
 Serves Goals 1, 3 and 4.
 
-- `markdownToAdf(adfToMarkdown(doc))` equals `doc` — unknown node types included, carried opaquely
+- `markdownToAdf(adfToMarkdown(doc))` deep-equals `doc` — every key and value as `doc` holds it,
+  adjacent text nodes, an empty `attrs`, `content` or `marks` and `-0` included, and unknown node
+  types carried opaquely
   ([`docs/decisions.md`](https://gitea.larvit.se/larvit/adf-codec/src/branch/main/docs/decisions.md#unknown-nodes-ride-the-carry)).
 - Markdown this library reads, and markdown it writes, means what the CommonMark spec says; from
   `0.2.0`, well-formed HTML means what the HTML standard says, read or written. The bullets below
   name every exception.
 - Plain CommonMark is valid input to `markdownToAdf` apart from the raw HTML `unmappable-html`
-  names, with three carve-outs — literal text matching directive, pipe-table or strikethrough
+  names, with four carve-outs — literal text matching directive, pipe-table or strikethrough
-  syntax is claimed (escapable — `spec/flavour.md`) — and one gap: a CommonMark image fits only as
+  syntax, and a code fence whose info string opens `adf:`, is claimed (escapable — `spec/flavour.md`)
+  — and one gap: a CommonMark image fits only as
   its own title-less paragraph; mid-text and titled images are error results, save an image inside
   another's description, which flattens into the alt text. Converting back yields the library's
   canonical spelling, which round-trips byte-identically — where it converts back at all: a parse
@@ -235,7 +242,7 @@ Serves Goals 1, 3 and 4.
   makes a call loop forever.
 - The emitted formats are semver surface
   ([`docs/decisions.md`](https://gitea.larvit.se/larvit/adf-codec/src/branch/main/docs/decisions.md#the-formats-are-api)).
-- **`0.2.0`** — `htmlToAdf(adfToHtml(doc))` equals `doc`; fidelity HTML cannot express rides
+- **`0.2.0`** — `htmlToAdf(adfToHtml(doc))` deep-equals `doc`; fidelity HTML cannot express rides
   `data-*` attributes. Foreign HTML maps a documented element set, which markdown's raw HTML reads
   through as well, and a construct outside it is an error; well-formed HTML only — no tag-soup
   recovery.

browser-tests/convert-corpus.js

@@ -1,16 +1,18 @@
 try {
   const { adfToMarkdown, isAdfDocument, markdownToAdf } = await import('/dist/index.js')
+  // WebDriver's JSON reads -0 back as 0, so a document crosses as JSON text with -0 tagged; run.js revives it.
+  const spelled = (result) => (result.ok ? { ok: true, value: JSON.stringify(result.value, (_, value) => (Object.is(value, -0) ? '\u0000-0' : value)) } : result)
   window.convertCorpus = (corpus) => ({
     errors: corpus.errors.map(({ markdown, name }) => ({ name, parsed: markdownToAdf(markdown) })),
-    normalization: corpus.normalization.map(({ markdown, name }) => ({ name, parsed: markdownToAdf(markdown) })),
+    normalization: corpus.normalization.map(({ markdown, name }) => ({ name, parsed: spelled(markdownToAdf(markdown)) })),
     realPayloads: corpus.realPayloads.map(({ json, name }) => {
       const adf = JSON.parse(json)
       const emitted = adfToMarkdown(adf)
-      return { emitted, isDocument: isAdfDocument(adf), name, parsed: emitted.ok ? markdownToAdf(emitted.value) : undefined }
+      return { emitted, isDocument: isAdfDocument(adf), name, parsed: emitted.ok ? spelled(markdownToAdf(emitted.value)) : undefined }
     }),
     roundTrip: corpus.roundTrip.map(({ json, markdown, name }) => {
       const adf = JSON.parse(json)
-      return { emitted: adfToMarkdown(adf), isDocument: isAdfDocument(adf), name, parsed: markdownToAdf(markdown) }
+      return { emitted: adfToMarkdown(adf), isDocument: isAdfDocument(adf), name, parsed: spelled(markdownToAdf(markdown)) }
     }),
   })
 } catch (cause) {

browser-tests/run.js

@@ -2,7 +2,6 @@ import assert from 'node:assert/strict'
 import { createServer } from 'node:http'
 import { extname, join } from 'node:path'
 import { readFileSync, readdirSync } from 'node:fs'
-import { toEditorNormal } from '../dist/adf/editor-normal.js'
 
 const contentTypes = { '.html': 'text/html; charset=utf-8', '.js': 'text/javascript' }
 const driver = 'http://127.0.0.1:4444'
@@ -50,6 +49,11 @@ function fixtureNames(kind, extension) {
     .sort()
 }
 
+// convert-corpus.js tags -0 so it survives WebDriver's JSON.
+function revived(text) {
+  return JSON.parse(text, (_, value) => (value === '\u0000-0' ? -0 : value))
+}
+
 function refusal(result) {
   return result.ok ? '' : `${result.error.code}: ${result.error.message}`
 }
@@ -107,7 +111,7 @@ for (const [index, { json, markdown, name }] of corpus.roundTrip.entries()) {
     assert.ok(result.emitted.ok, `it did not emit — ${refusal(result.emitted)}`)
     assert.equal(result.emitted.value, markdown)
     assert.ok(result.parsed.ok, `it did not parse — ${refusal(result.parsed)}`)
-    assert.deepEqual(toEditorNormal(result.parsed.value), JSON.parse(json))
+    assert.deepEqual(revived(result.parsed.value), JSON.parse(json))
   })
 }
 
@@ -115,7 +119,7 @@ for (const [index, { name }] of corpus.normalization.entries()) {
   const result = results.normalization[index]
   checking(name, result, () => {
     assert.ok(result.parsed.ok, `it did not parse — ${refusal(result.parsed)}`)
-    assert.deepEqual(toEditorNormal(result.parsed.value), JSON.parse(fixture(name, '.json')))
+    assert.deepEqual(revived(result.parsed.value), JSON.parse(fixture(name, '.json')))
   })
 }
 
@@ -125,7 +129,7 @@ for (const [index, { json, name }] of corpus.realPayloads.entries()) {
     assert.ok(result.isDocument, `${name}.json is no ADF document`)
     assert.ok(result.emitted.ok, `it did not emit — ${refusal(result.emitted)}`)
     assert.ok(result.parsed.ok, `it did not parse back — ${refusal(result.parsed)}`)
-    assert.deepEqual(toEditorNormal(result.parsed.value), JSON.parse(json))
+    assert.deepEqual(revived(result.parsed.value), JSON.parse(json))
   })
 }
 

corpus/README.md

@@ -18,8 +18,8 @@ One directory per contract kind:
   `reason`. `kind` is `mark-model` (the permanent count divergence from ADF's mark-per-text-node
   model), `unspellable` (parses but the flavour has no spelling) or `pending` (a parser gap).
 
-JSON is editor-normal (`docs/decisions.md` §Equality is editor-normal), two-space indent, keys
+JSON is two-space indent, keys sorted, and a document read back must deep-equal the fixture's
-sorted. `spec.json` is the vendored, upstream machine-readable suite, byte-exact from
+(`docs/decisions.md` §Equality is deep). `spec.json` is the vendored, upstream machine-readable suite, byte-exact from
 [spec.commonmark.org](https://spec.commonmark.org/0.31.2/spec.json) (CommonMark 0.31.2, © John
 MacFarlane, [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/)), and is not
 re-serialized by the corpus gate.

corpus/errors/carry-fence-type-spellable.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/carry-fence-type-spellable.md

@@ -0,0 +1,5 @@
+```adf:
+{
+  "type": "blockCard"
+}
+```

corpus/errors/carry-fence-type-unspellable.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/carry-fence-type-unspellable.md

@@ -0,0 +1,3 @@
+```adf:\\
+{}
+```

corpus/errors/carry-fence-type.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/carry-fence-type.md

@@ -0,0 +1,5 @@
+```adf:blockCard
+{
+  "type": "blockCard"
+}
+```

corpus/errors/carry-invalid-json.md

@@ -1,3 +1,3 @@
-```carry
+```adf:blockCard
-{"type":
+{"attrs":
 ```

corpus/errors/code-block-fence-languages.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/code-block-fence-languages.md

@@ -0,0 +1,8 @@
+!adf:codeBlock
+```js
+a
+```
+```ts
+b
+```
+!adf:/codeBlock

corpus/errors/directive-slot-carried-marks.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/directive-slot-carried-marks.md

@@ -0,0 +1 @@
+!adf:status[!adf:carry{json="{\"marks\":[],\"text\":\"x\",\"type\":\"text\"}"}]

corpus/errors/document-directive-bare.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/document-directive-bare.md

@@ -0,0 +1 @@
+!adf:doc

corpus/errors/document-directive-misplaced.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/document-directive-misplaced.md

@@ -0,0 +1,3 @@
+!adf:doc {content=none}
+
+Text.

corpus/errors/empty-attrs-with-attributes.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/empty-attrs-with-attributes.md

@@ -0,0 +1,3 @@
+!adf:panel info {attrs=empty}
+Text.
+!adf:/panel

corpus/errors/empty-content-with-body.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/empty-content-with-body.md

@@ -0,0 +1,3 @@
+!adf:paragraph {content=empty}
+Text.
+!adf:/paragraph

corpus/errors/empty-key-value.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/empty-key-value.md

@@ -0,0 +1,3 @@
+!adf:paragraph {attrs=none}
+Text.
+!adf:/paragraph

corpus/errors/empty-marks-wrapped.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/empty-marks-wrapped.md

@@ -0,0 +1 @@
+**!adf:date{marks=empty}**

corpus/errors/text-break-attributes.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/text-break-attributes.md

@@ -0,0 +1 @@
+a!adf:textBreak{x=y}b

corpus/errors/text-break-content.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/text-break-content.md

@@ -0,0 +1 @@
+a!adf:textBreak[x]b

corpus/errors/text-break-marks-differ.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/text-break-marks-differ.md

@@ -0,0 +1 @@
+**a**!adf:textBreak{}b

corpus/errors/text-break-misplaced.error

@@ -0,0 +1 @@
+unsupported-node-shape

corpus/errors/text-break-misplaced.md

@@ -0,0 +1 @@
+!adf:textBreak{}Hello

corpus/round-trip/block-nodes/code-block-carry.json

@@ -14,7 +14,19 @@
     },
     {
       "attrs": {
-        "language": "carry"
+        "language": "adf:blockCard"
+      },
+      "content": [
+        {
+          "text": "{\n  \"attrs\": {}\n}",
+          "type": "text"
+        }
+      ],
+      "type": "codeBlock"
+    },
+    {
+      "attrs": {
+        "language": "adf:"
       },
       "content": [
         {

corpus/round-trip/block-nodes/code-block-carry.md

@@ -1,12 +1,18 @@
-!adf:codeBlock {language=carry}
+```carry
-```
 {
   "type": "blockCard"
 }
 ```
+
+!adf:codeBlock {language="adf:blockCard"}
+```
+{
+  "attrs": {}
+}
+```
 !adf:/codeBlock
 
-!adf:codeBlock {language=carry}
+!adf:codeBlock {language="adf:"}
 ````
 ```
 ````

corpus/round-trip/block-nodes/code-block-text-nodes.json

@@ -0,0 +1,56 @@
+{
+  "content": [
+    {
+      "attrs": {
+        "language": "js"
+      },
+      "content": [
+        {
+          "text": "const a = 1\n",
+          "type": "text"
+        },
+        {
+          "text": "const b = 2",
+          "type": "text"
+        }
+      ],
+      "type": "codeBlock"
+    },
+    {
+      "content": [
+        {
+          "text": "a",
+          "type": "text"
+        },
+        {
+          "text": "```",
+          "type": "text"
+        },
+        {
+          "text": "\n",
+          "type": "text"
+        }
+      ],
+      "type": "codeBlock"
+    },
+    {
+      "attrs": {
+        "language": "has`tick",
+        "wrap": true
+      },
+      "content": [
+        {
+          "text": "x",
+          "type": "text"
+        },
+        {
+          "text": " y ",
+          "type": "text"
+        }
+      ],
+      "type": "codeBlock"
+    }
+  ],
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/block-nodes/code-block-text-nodes.md

@@ -0,0 +1,31 @@
+!adf:codeBlock
+```js
+const a = 1
+
+```
+```js
+const b = 2
+```
+!adf:/codeBlock
+
+!adf:codeBlock
+```
+a
+```
+````
+```
+````
+```
+
+
+```
+!adf:/codeBlock
+
+!adf:codeBlock {language="has\u0060tick" wrap=true}
+```
+x
+```
+```
+ y 
+```
+!adf:/codeBlock

corpus/round-trip/block-nodes/document-without-content.json

@@ -0,0 +1,4 @@
+{
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/block-nodes/document-without-content.md

@@ -0,0 +1 @@
+!adf:doc {content=none}

corpus/round-trip/block-nodes/empty-keys.json

@@ -0,0 +1,89 @@
+{
+  "content": [
+    {
+      "content": [],
+      "type": "paragraph"
+    },
+    {
+      "attrs": {},
+      "content": [
+        {
+          "text": "Plain words.",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [],
+      "type": "rule"
+    },
+    {
+      "attrs": {
+        "level": 2
+      },
+      "content": [
+        {
+          "text": "Title",
+          "type": "text"
+        }
+      ],
+      "marks": [],
+      "type": "heading"
+    },
+    {
+      "attrs": {
+        "panelType": "info"
+      },
+      "content": [],
+      "type": "panel"
+    },
+    {
+      "content": [
+        {
+          "attrs": {},
+          "content": [
+            {
+              "content": [
+                {
+                  "text": "Item",
+                  "type": "text"
+                }
+              ],
+              "type": "paragraph"
+            }
+          ],
+          "type": "listItem"
+        },
+        {
+          "content": [
+            {
+              "content": [
+                {
+                  "text": "Next",
+                  "type": "text"
+                }
+              ],
+              "type": "paragraph"
+            }
+          ],
+          "type": "listItem"
+        }
+      ],
+      "type": "bulletList"
+    },
+    {
+      "content": [],
+      "type": "codeBlock"
+    },
+    {
+      "attrs": {
+        "language": "js"
+      },
+      "marks": [],
+      "type": "codeBlock"
+    }
+  ],
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/block-nodes/empty-keys.md

@@ -0,0 +1,32 @@
+!adf:paragraph {content=empty}
+!adf:/paragraph
+
+!adf:paragraph {attrs=empty}
+Plain words.
+!adf:/paragraph
+
+!adf:rule {content=empty}
+
+!adf:heading {level=2 marks=empty}
+Title
+!adf:/heading
+
+!adf:panel info {content=empty}
+!adf:/panel
+
+!adf:bulletList
+!adf:listItem {attrs=empty}
+Item
+!adf:/listItem
+!adf:listItem
+Next
+!adf:/listItem
+!adf:/bulletList
+
+!adf:codeBlock {content=empty}
+!adf:/codeBlock
+
+!adf:codeBlock {marks=empty}
+```js
+```
+!adf:/codeBlock

corpus/round-trip/combinations/carry-attributes.md

@@ -1,4 +1,4 @@
-```carry
+```adf:panel
 {
   "attrs": {
     "rounded": true
@@ -13,12 +13,11 @@
       ],
       "type": "paragraph"
     }
-  ],
+  ]
-  "type": "panel"
 }
 ```
 
-```carry
+```adf:panel
 {
   "attrs": {
     "panelType": "extra info"
@@ -33,8 +32,7 @@
       ],
       "type": "paragraph"
     }
-  ],
+  ]
-  "type": "panel"
 }
 ```
 

corpus/round-trip/combinations/negative-zero.json

@@ -0,0 +1,83 @@
+{
+  "content": [
+    {
+      "attrs": {
+        "order": -0
+      },
+      "content": [
+        {
+          "content": [
+            {
+              "content": [
+                {
+                  "text": "First",
+                  "type": "text"
+                }
+              ],
+              "type": "paragraph"
+            }
+          ],
+          "type": "listItem"
+        }
+      ],
+      "type": "orderedList"
+    },
+    {
+      "attrs": {
+        "level": -0
+      },
+      "content": [
+        {
+          "text": "Title",
+          "type": "text"
+        }
+      ],
+      "type": "heading"
+    },
+    {
+      "attrs": {
+        "extensionKey": "toc",
+        "extensionType": "com.atlassian.confluence.macro.core",
+        "parameters": {
+          "maxLevel": -0
+        }
+      },
+      "type": "extension"
+    },
+    {
+      "content": [
+        {
+          "attrs": {
+            "data": [
+              -0
+            ],
+            "url": "https://example.com/"
+          },
+          "type": "inlineCard"
+        },
+        {
+          "marks": [
+            {
+              "attrs": {
+                "color": "#000000",
+                "size": -0
+              },
+              "type": "border"
+            }
+          ],
+          "text": "x",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "attrs": {
+        "x": -0
+      },
+      "type": "blockCard"
+    }
+  ],
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/combinations/negative-zero.md

@@ -0,0 +1,21 @@
+!adf:orderedList {order=-0}
+!adf:listItem
+First
+!adf:/listItem
+!adf:/orderedList
+
+!adf:heading {level=-0}
+Title
+!adf:/heading
+
+!adf:extension {extensionKey=toc extensionType="com.atlassian.confluence.macro.core" parameters="{\"maxLevel\":-0}"}
+
+!adf:inlineCard{data="[-0]" url="https://example.com/"}!adf:border[x]{color="#000000" size=-0}
+
+```adf:blockCard
+{
+  "attrs": {
+    "x": -0
+  }
+}
+```

corpus/round-trip/commonmark-subset/document-empty.json

@@ -1,4 +1,5 @@
 {
+  "content": [],
   "type": "doc",
   "version": 1
 }

corpus/round-trip/inline-nodes/empty-keys.json

@@ -0,0 +1,98 @@
+{
+  "content": [
+    {
+      "content": [
+        {
+          "attrs": {},
+          "type": "date"
+        },
+        {
+          "text": " ",
+          "type": "text"
+        },
+        {
+          "content": [],
+          "type": "date"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "text": "a",
+          "type": "text"
+        },
+        {
+          "marks": [],
+          "type": "hardBreak"
+        },
+        {
+          "text": "b",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "attrs": {
+            "id": "x",
+            "text": "@M"
+          },
+          "content": [],
+          "type": "mention"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "text": "bare",
+          "type": "text"
+        },
+        {
+          "marks": [],
+          "text": "marked",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "marks": [
+            {
+              "attrs": {},
+              "type": "em"
+            }
+          ],
+          "text": "x",
+          "type": "text"
+        },
+        {
+          "text": " and ",
+          "type": "text"
+        },
+        {
+          "attrs": {
+            "shortName": ":a:"
+          },
+          "marks": [
+            {
+              "attrs": {},
+              "type": "strong"
+            }
+          ],
+          "type": "emoji"
+        }
+      ],
+      "type": "paragraph"
+    }
+  ],
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/inline-nodes/empty-keys.md

@@ -0,0 +1,9 @@
+!adf:date{attrs=empty} !adf:date{content=empty}
+
+a!adf:hardBreak{marks=empty}b
+
+!adf:mention[@M]{content=empty id=x}
+
+bare!adf:carry{json="{\"marks\":[],\"text\":\"marked\",\"type\":\"text\"}"}
+
+!adf:carry{json="{\"marks\":[{\"attrs\":{},\"type\":\"em\"}],\"text\":\"x\",\"type\":\"text\"}"} and !adf:carry{json="{\"attrs\":{\"shortName\":\":a:\"},\"marks\":[{\"attrs\":{},\"type\":\"strong\"}],\"type\":\"emoji\"}"}

corpus/round-trip/inline-nodes/text-break.json

@@ -0,0 +1,154 @@
+{
+  "content": [
+    {
+      "content": [
+        {
+          "text": "Hello, ",
+          "type": "text"
+        },
+        {
+          "text": "world",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "marks": [
+            {
+              "type": "strong"
+            }
+          ],
+          "text": "Hello, ",
+          "type": "text"
+        },
+        {
+          "marks": [
+            {
+              "type": "strong"
+            }
+          ],
+          "text": "world",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "marks": [
+            {
+              "type": "code"
+            }
+          ],
+          "text": "a",
+          "type": "text"
+        },
+        {
+          "marks": [
+            {
+              "type": "code"
+            }
+          ],
+          "text": "b",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "marks": [
+            {
+              "attrs": {
+                "href": "https://example.com/"
+              },
+              "type": "link"
+            }
+          ],
+          "text": "a",
+          "type": "text"
+        },
+        {
+          "marks": [
+            {
+              "attrs": {
+                "href": "https://example.com/"
+              },
+              "type": "link"
+            }
+          ],
+          "text": "b",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "text": "Line\n",
+          "type": "text"
+        },
+        {
+          "text": "next",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "marks": [
+            {
+              "type": "underline"
+            }
+          ],
+          "text": "a",
+          "type": "text"
+        },
+        {
+          "marks": [
+            {
+              "type": "underline"
+            }
+          ],
+          "text": "b",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    },
+    {
+      "content": [
+        {
+          "marks": [
+            {
+              "attrs": {},
+              "type": "underline"
+            }
+          ],
+          "text": "a",
+          "type": "text"
+        },
+        {
+          "marks": [
+            {
+              "type": "underline"
+            }
+          ],
+          "text": "b",
+          "type": "text"
+        }
+      ],
+      "type": "paragraph"
+    }
+  ],
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/inline-nodes/text-break.md

@@ -0,0 +1,13 @@
+Hello, !adf:textBreak{}world
+
+**Hello, !adf:textBreak{}world**
+
+`a`!adf:textBreak{}`b`
+
+[a!adf:textBreak{}b](https://example.com/)
+
+Line!adf:text{text="\n"}!adf:textBreak{}next
+
+!adf:underline[a!adf:textBreak{}b]
+
+!adf:underline[a]{attrs=empty}!adf:underline[b]

corpus/round-trip/opaque-carry/carry-in-wrappers.md

@@ -1,17 +1,15 @@
-> ```carry
+> ```adf:blockCard
 > {
 >   "attrs": {
 >     "url": "https://example.com/quoted"
->   },
+>   }
->   "type": "blockCard"
 > }
 > ```
 
-- ```carry
+- ```adf:blockCard
   {
     "attrs": {
       "url": "https://example.com/listed"
-    },
+    }
-    "type": "blockCard"
   }
   ```

corpus/round-trip/opaque-carry/code-block-children.json

@@ -0,0 +1,35 @@
+{
+  "content": [
+    {
+      "attrs": {
+        "language": "js"
+      },
+      "content": [
+        {
+          "marks": [
+            {
+              "type": "strong"
+            }
+          ],
+          "text": "const a = 1",
+          "type": "text"
+        }
+      ],
+      "type": "codeBlock"
+    },
+    {
+      "content": [
+        {
+          "text": "a",
+          "type": "text"
+        },
+        {
+          "type": "hardBreak"
+        }
+      ],
+      "type": "codeBlock"
+    }
+  ],
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/opaque-carry/code-block-children.md

@@ -0,0 +1,32 @@
+```adf:codeBlock
+{
+  "attrs": {
+    "language": "js"
+  },
+  "content": [
+    {
+      "marks": [
+        {
+          "type": "strong"
+        }
+      ],
+      "text": "const a = 1",
+      "type": "text"
+    }
+  ]
+}
+```
+
+```adf:codeBlock
+{
+  "content": [
+    {
+      "text": "a",
+      "type": "text"
+    },
+    {
+      "type": "hardBreak"
+    }
+  ]
+}
+```

corpus/round-trip/opaque-carry/unknown-block.md

@@ -1,23 +1,21 @@
-```carry
+```adf:blockCard
 {
   "attrs": {
     "url": "https://example.com/roadmap"
-  },
+  }
-  "type": "blockCard"
 }
 ```
 
 !adf:panel info
 The card below has no spelling yet.
 
-```carry
+```adf:embedCard
 {
   "attrs": {
     "layout": "wide",
     "url": "https://example.com/board",
     "width": 100
-  },
+  }
-  "type": "embedCard"
 }
 ```
 !adf:/panel

corpus/round-trip/opaque-carry/unknown-type-spelling.json

@@ -0,0 +1,21 @@
+{
+  "content": [
+    {
+      "attrs": {
+        "url": "https://example.com/"
+      },
+      "type": "has`tick"
+    },
+    {
+      "type": ""
+    },
+    {
+      "type": " padded"
+    },
+    {
+      "type": "two words"
+    }
+  ],
+  "type": "doc",
+  "version": 1
+}

corpus/round-trip/opaque-carry/unknown-type-spelling.md

@@ -0,0 +1,24 @@
+```adf:
+{
+  "attrs": {
+    "url": "https://example.com/"
+  },
+  "type": "has`tick"
+}
+```
+
+```adf:
+{
+  "type": ""
+}
+```
+
+```adf:
+{
+  "type": " padded"
+}
+```
+
+```adf:two words
+{}
+```

docs/decisions.md

@@ -14,7 +14,7 @@ a backslash reach them intact; what the flavour cannot spell reduces ADF→ADF a
 2026-08-23, real payloads 2026-09-15, the maintainer. Goal 1. Valid while a consumer saves back
 through the lossless pair.
 
-`markdownToAdf(adfToMarkdown(doc))` and `htmlToAdf(adfToHtml(doc))` must equal `doc` — anything
+`markdownToAdf(adfToMarkdown(doc))` and `htmlToAdf(adfToHtml(doc))` must deep-equal `doc` — anything
 less silently destroys content an editor could not represent, in a document it did not author.
 When losslessness and readability conflict, losslessness wins. Round-trip equality is a property
 tested over a checked-in corpus (`corpus/README.md`), not a claim made in prose. Its real payloads
@@ -31,27 +31,103 @@ where there is a way back. CommonMark spells some things the flavour has no esca
 paragraph opening with a code span whose backticks read back as a fence — so a parse succeeding
 does not imply a spellable document; `corpus/commonmark-spec/exceptions.json` names those.
 
-## Equality is editor-normal
+## Equality is deep
 
-2026-08-24, the maintainer. Goal 1. Valid while markdown cannot tell apart the ADF shapes this
+2026-08-24, deep 2026-10-03, the maintainer. Goal 1. Valid while a pipeline or a bot can build a
-merges.
+shape the editor would not.
 
-"Equals" is structural equality over editor-normal ADF — adjacent text nodes with identical marks
+"Equals" is `assert.deepStrictEqual`: every key and value `doc` holds, two adjacent text nodes, an
-and no attributes merged, JSON number semantics, an empty attrs object, marks array or content
+empty `attrs`, `content` or `marks` and `-0` included, with no normalization on either side.
-array the absent key — the only domain markdown can restore.
+CommonMark's spelling stays wherever a document holds none of those shapes. The plain reader
-`todo.md` item 40 replaces this with deep equality (2026-09-28, the maintainer).
+builds what is written, as `markdownToAdf` does. Only the plain writer is lossy: its reduction
+reads and writes editor-normal ADF — adjacent text nodes of identical marks and no attributes
+merged, `-0` as `0`, and an empty `attrs`, `content` or `marks` the absent key — so two documents
+the editor holds equal write the same plain markdown.
+
+## `!adf:textBreak{}` parts text CommonMark would join
+
+2026-10-03, the maintainer. Goal 1. Valid while CommonMark reads adjacent text as one run.
+
+Two adjacent text nodes a reader would build back as one — neither carried, neither holding
+`attrs` or an empty key, their marks identical, attributes included — are parted by the reserved
+inline leaf `!adf:textBreak{}`, mirroring `!adf:listBreak`: it builds no node, and anywhere but
+between two such nodes, or given `[content]` or `{attrs}`, it is `unsupported-node-shape`. It sits
+inside every mark spelling the pair shares, `**Hello, !adf:textBreak{}world**`; a code span holds
+no directive, so it closes and reopens, `` `a`!adf:textBreak{}`b` ``. A carried node never joins
+its neighbour, so a carry needs no break. A mark run breaks on any difference in the mark, `attrs:
+{}` against no `attrs` included.
+
+## An empty key spells `empty`
+
+2026-10-03, a writer panel and the maintainer. Goals 1 and 5. Valid while no attribute value
+spells an empty object or array.
+
+An `attrs`, `content` or `marks` key holding an empty object or array is the reserved key with the
+bare value `empty` on every directive — block, inline node and mark: `{attrs=empty}`,
+`{content=empty}`, `{marks=empty}`. A writer panel chose the spelling, 5 of 7. A container
+spelling `content=empty` closes with no body, so an empty pair stays the node holding no content
+key; a leaf spells it too, `!adf:rule {content=empty}`. A node CommonMark spells takes its
+directive form to hold an empty key, and a text node holding one rides the inline carry, as does a
+node under an `em`, `strong`, `strike`, `code` or `link` mark whose `attrs` is empty, since none of
+those spellings holds attributes. Any other value of a reserved key is `unsupported-node-shape`,
+except on a block's `marks`, which reads it as the marks array in JSON.
+
+## `-0` is spelled `-0`
+
+2026-10-03, the maintainer. Goal 1. Valid while JSON's own serialization writes `-0` as `0`.
+
+Wherever the flavour writes a number or a JSON value — an attribute, a `json` value, the carry —
+`-0` is `-0`, which JSON's grammar reads back as `-0`. An `orderedList` whose `order` is `-0` takes
+the directive form, since no list marker spells the sign.
+
+## Empty markdown is a document of no blocks
+
+2026-10-03, a panel and the maintainer. Goals 1 and 5. Valid while ADF's schema requires
+`content` on `doc`.
+
+Markdown holding no block reads as `{ content: [], type: 'doc', version: 1 }`, the document
+`spec/adf-schema/full.json` requires. A document holding no `content` key is
+`!adf:doc {content=none}` standing alone as its only block, and a named error anywhere else. The
+writer panel split 4 for `none` and 3 for `absent`, and the maintainer chose `none`; all seven rejected a
+bare `!adf:doc`.
 
 ## Unknown nodes ride the carry
 
-2026-08-23, extended to misplaced known nodes 2026-08-26, the maintainer. Goal 1. Valid while ADF
+2026-08-23, extended to misplaced known nodes 2026-08-26 and to code block children 2026-10-03, the
-holds nodes, or node positions, this library does not spell.
+maintainer. Goal 1. Valid while ADF holds nodes, or node positions, this library does not spell.
 
 An unknown ADF node is carried opaquely — raw JSON rides a dedicated syntax in both formats and
 restores to a deep-equal node. The round-trip holds for documents newer than the library. So does
 a known node no section spells where it stands: a markdown serializer spells a node by type without
-checking its position, and refusing loses a document ADF itself keeps in an `unsupportedBlock`.
+checking its position, and refusing loses a document ADF itself keeps in an `unsupportedBlock`. A
-Where a container's own spelling cannot hold the child it has — a `bulletList` holding other than
+`codeBlock` holding a child no fence holds — anything but a text node carrying no marks, `attrs` or
-`listItem`, a `codeBlock` other than text — the error result names that instead.
+`content` — rides the carry whole.
+
+## The carry fence names the node type
+
+2026-10-03, the maintainer. Goals 1 and 5. Valid while a code fence's info string reads back
+verbatim.
+
+The block carry is a code fence whose info string `adf:<type>` names the node's type, its body the
+node's JSON without `type`: ```` ```adf:blockCard ````. A type no info string carries back — by the
+rule a code language follows — leaves the info string `adf:` and keeps `type` in the body. Every
+info string opening `adf:` is reserved, so a `codeBlock` whose language opens so takes the
+`language` attribute, and `carry` is an ordinary language. A body holding `type` under a named
+type, or an `adf:` fence whose type an info string carries, is `unsupported-node-shape`. The
+reservation claims a fence CommonMark reads as code until `todo.md` item 43 gives CommonMark its own
+reader.
+
+## A code block is a fence per text node
+
+2026-10-03, the maintainer. Goal 1. Valid while ADF holds a code block's text in more than one
+node.
+
+A `codeBlock` holding several text nodes is the `!adf:codeBlock` container holding one fence per
+node, each fence's info string the language; its other attributes sit on the opener, and a
+language no info string carries stays the opener's `language` with bare fences. A `codeBlock`
+spelling `content=empty` has no fence to carry the language, so the opener does. Fences with
+differing info strings are `unsupported-node-shape` — ADF holds one language — and so is an empty
+fence beside another, since a text node holds text.
 
 ## Foreign HTML sorts three ways
 
@@ -97,7 +173,9 @@ opener nests by itself and leaf versus container falls out of the node's content
 2026-08-23, the maintainer. Goal 4. Valid while prose rarely writes the shapes the carve-outs claim.
 
 Plain CommonMark is a subset, with carve-outs (`spec/flavour.md`): literal text shaped like a
-directive, a pipe table or a `~~` pair is claimed — plus one image gap.
+directive, a pipe table or a `~~` pair is claimed, and so is a code fence whose info string opens
+`adf:` (§The carry fence names the node type) — plus one image gap. `todo.md` item 43 ends the
+claims, giving CommonMark its own reader.
 
 ## Tables
 
@@ -295,10 +373,11 @@ handles one cause alike whichever node, attribute or direction raised it.
 - A directive whose name reads back to no node is `unknown-directive-name` rather than a claim
   code — the spelling is well formed, and telling that apart from a typo is what a consumer
   switches on when a later MINOR gives the name meaning. A reserved name is a known name, so never
-  that code, and the two the flavour reserves part on form: a form the grammar does not have is a
+  that code, and the names the flavour reserves part on form: a form the grammar does not have is
-  claim code — `!adf:carry`, whose carry is the fence — and a well-formed form in the wrong place
+  a claim code — `!adf:carry`, whose carry is the fence — and a well-formed form in the wrong place
   is `unsupported-node-shape`, `!adf:listBreak` parting anything but two adjacent lists of one
-  type (2026-09-01).
+  type, `!adf:textBreak{}` anything but two text nodes a reader joins, `!adf:doc` standing beside
+  another block (2026-09-01, the text break and `doc` 2026-10-03).
 - A well-formed directive the node tables refuse — an attribute a node does not hold or spells
   elsewhere, a value outside its kind or its canonical spelling, an argument, or a body of a shape
   its content model does not take — is `unsupported-node-shape`, the emitter's code for the same

spec/flavour.md

@@ -1,9 +1,10 @@
 # The markdown flavour
 
 The grammar of the extended markdown `adfToMarkdown` emits and `markdownToAdf` parses. Plain
-CommonMark is a subset apart from raw HTML (below), with three carve-outs: literal text that matches
+CommonMark is a subset apart from raw HTML (below), with four carve-outs: literal text that matches
-directive syntax below or reads as a pipe table is claimed by the flavour, and a matched `~~` pair
+directive syntax below or reads as a pipe table is claimed by the flavour, a matched `~~` pair
-spells `strike` (escape the `!adf:`, `|` or `~` to keep it literal) — and one gap: a CommonMark
+spells `strike` (escape the `!adf:`, `|` or `~` to keep it literal), and a code fence whose info
+string opens `adf:` is the opaque carry (The opaque carry says how to keep it code) — and one gap: a CommonMark
 image fits only as its own title-less paragraph — mid-text and titled images are named errors. The
 emitted form is contract (`docs/decisions.md` §The formats are API). Per-node syntaxes build on this
 grammar in the sections below.
@@ -44,7 +45,8 @@ normalizes to it through the round-trip.
   CommonMark admits no spelling — the end of a block, inside an ATX heading — or where the node
   carries an attribute, it is the inline directive.
 - An empty paragraph — real payloads carry them — is an `!adf:paragraph` … `!adf:/paragraph` pair
-  holding nothing.
+  holding nothing, and one whose `content` is an empty array the pair
+  `!adf:paragraph {content=empty}` … `!adf:/paragraph` (Attributes).
 - Links `[text](url)`; `<…>` around a destination containing spaces, `<>` an empty one beside a
   title; title in double quotes. A backslash escapes a parenthesis the destination leaves
   unbalanced, and a quote inside the title; a balanced pair stays bare. `<url>` autolink form only
@@ -67,7 +69,10 @@ normalizes to it through the round-trip.
   matching below, which is what lets the emitter decide its own pairings.
 - Blocks separated by one blank line at document level, inside a blockquote and between CommonMark
   blocks; inside a directive container a pair holding a directive block takes none. No trailing
-  whitespace outside a code block's content, single trailing newline; a document with no blocks is the empty string.
+  whitespace outside a code block's content, single trailing newline. A document whose `content` is
+  an empty array is the empty string, and markdown holding no block reads back to it; a document
+  holding no `content` key is the leaf `!adf:doc {content=none}` as its only block, which is a
+  named error anywhere else or spelled any other way.
 
 ## Directives
 
@@ -79,9 +84,11 @@ result naming it at the opener, whatever follows it — so output an old emitter
 escaped, and erroring input gaining meaning later is MINOR, never a reparse (`docs/decisions.md`
 §The formats are API). Each name belongs to one position, and a name the other one spells — a mark
 or an inline node written as a block directive, a block node written inline — is a different error,
-naming the spelling it takes. Two reserved names read back to no node: `carry` for the opaque carry,
+naming the spelling it takes. Four reserved names read back to no node: `carry` for the inline
-as both directive name and fence info string, and `listBreak` for the leaf that parts two adjacent
+opaque carry, `listBreak` for the leaf that parts two adjacent lists and `doc` for a document
-lists (Canonical form).
+holding no `content` key (Canonical form), and `textBreak` for the leaf that parts two text nodes
+(Inline nodes). Every fence info string opening `adf:` is reserved for the block carry (The opaque
+carry).
 
 **Claiming**: an unescaped `!adf:` claims wherever it stands. What follows picks the form: `/name`
 closes a container, and a name picks by what follows it in turn — a space or the line's end a block
@@ -142,6 +149,13 @@ ends the name (`!adf:hardBreak{}`). Input reads that spelling alone: keys out of
 quoted where bare carries it, an escape longer than it need be, an empty `{attrs}` on a block line
 or after a `[content]`, and a number or `json` value outside its canonical JSON spelling are each a
 named error naming the spelling to write instead.
+`attrs`, `content` and `marks` are reserved keys on every directive — block, inline node and mark —
+whose bare value `empty` spells the node's or mark's key holding an empty object or array:
+`!adf:underline[a]{attrs=empty}`, `!adf:date{content=empty}`, `!adf:hardBreak{marks=empty}`. A
+container spelling `content=empty` closes with no body; `attrs=empty` stands beside no other
+attribute, argument or content slot; and an inline node spelling `marks=empty` stands inside no
+mark spelling. Any other value of a reserved key is a named error, except on a block's `marks`
+(Block nodes).
 
 **Escaping**: the emitter backslash-escapes whatever literal text would otherwise parse as
 directive syntax — every literal `!adf:`, `]` inside content, a bracket a link's destination and
@@ -163,15 +177,18 @@ and restores to a deep-equal node. A carry may hold a node the emitter spells na
 unreinterpreted, and the next emit spells it canonically (`docs/decisions.md` §The round-trip is the
 product). Block and inline positions canonicalize differently, each fitting where it sits:
 
-- **Block position**: a fenced code block with info string `carry`, body = the node's JSON —
+- **Block position**: a fenced code block with info string `adf:` and the node's type, body = the
-  two-space indent, object keys sorted.
+  node's JSON without its `type` — two-space indent, object keys sorted: ```` ```adf:blockCard ````.
+  A type no info string carries back, by the rule a `codeBlock`'s language follows, leaves the info
+  string `adf:` and keeps `type` in the body. A body holding `type` under a named type, or an
+  `adf:` fence whose type an info string carries, is a named error.
 - **Inline position**: `!adf:carry{json="…"}` — compact serialization (keys sorted, no whitespace),
   JSON-string-escaped into the attribute.
 
-The info string `carry` is reserved: a genuine `codeBlock` whose `language` is exactly `carry` takes
+Every info string opening `adf:` is reserved: a genuine `codeBlock` whose `language` opens so takes
 the attribute the section below keeps for a language no info string holds, so the reservation
-stays absolute.
+stays absolute. In block-directive position `!adf:carry` is a named error — the carry's block form
-In block-directive position `!adf:carry` is a named error — the carry's block form is the fence.
+is the fence.
 
 ## Raw HTML in input
 
@@ -193,13 +210,14 @@ Each section lists attributes as `name (type)`. A parenthesized value set docume
 payloads hold; the type stays string and any value round-trips verbatim. Values map to attrs by
 type: strings verbatim, numbers and booleans in canonical JSON spelling — quoted where not bare
 (`width="33.33"`) — and `json` values as the inline carry's serialization (compact, keys sorted),
-quoted. `markdownToAdf` emits `attrs`, `content` and `marks` keys only when non-empty; editor-normal
+quoted, `-0` spelled `-0`. `markdownToAdf` builds an `attrs`, `content` or `marks` key only where the
-ADF reads an empty attrs object, marks array or content array as the absent key (`docs/decisions.md`
+markdown spells one, an empty one through its reserved key (Attributes), so a document reads back
-§Equality is editor-normal) — the grammar's empty-`{attrs}` omission already collapses the two
+deep-equal (`docs/decisions.md` §Equality is deep). A node CommonMark spells takes the directive
-spellings.
+form to hold an empty key.
 
 Marks on a block node ride the reserved attribute key `marks` — the node's marks array as a
-`json` value: `!adf:layoutSection {marks="[{\"attrs\":{\"mode\":\"wide\"},\"type\":\"breakout\"}]"}`.
+`json` value, `marks=empty` where it is empty:
+`!adf:layoutSection {marks="[{\"attrs\":{\"mode\":\"wide\"},\"type\":\"breakout\"}]"}`.
 A section saying its body is inline takes at most one paragraph, whose inline content becomes
 the node's `content`; any other body is a named error, and an empty pair is a node holding none.
 
@@ -215,14 +233,17 @@ cannot — `localId` (string) on any of them, marks, and the values below — ta
 form.
 
 - `blockquote`, `bulletList`, `listItem` — containers, block body. Attributes: `localId` (string).
-- `codeBlock` — container, body one fenced code block whose info string is the language and whose
+- `codeBlock` — container, body one fenced code block per text node, each fence's info string the
-  content is the node's. Attributes: `hideLineNumbers` (boolean), `language` (string), `localId`
+  language and its content the node's text; a node holding no `content` key is one empty fence.
-  (string), `uniqueId` (string), `wrap` (boolean). A language no info string carries back — empty,
+  Attributes: `hideLineNumbers` (boolean), `language` (string), `localId` (string), `uniqueId`
-  the reserved `carry`, or holding a backtick, a backslash, a control character, edge whitespace or
+  (string), `wrap` (boolean). A language no info string carries back — empty, opening the reserved
-  an entity reference — rides the `language` attribute instead and the fence carries no info
+  `adf:`, or holding a backtick, a backslash, a control character, edge whitespace or an entity
-  string; writing it in the slot that rule leaves empty, or in both, is a named error. The body is
+  reference — rides the `language` attribute instead and the fences carry no info string, and so
-  one ordinary code block, and a fence's info string decodes escapes and entity references as any
+  does a language beside `content=empty`, which has no fence; writing it in the slot that rule
-  other does.
+  leaves empty, or in both, is a named error, and so are fences whose info strings differ and an
+  empty fence beside another. Each fence is an ordinary code block, and its info string decodes
+  escapes and entity references as any other does. A node holding a child no fence holds — any but
+  a text node carrying no marks, `attrs` or `content` — rides the block carry.
 - `heading` — container, inline body. Attributes: `level` (number), `localId` (string). `level` is
   the `#` count, so a heading carrying none, or one that is no whole number from 1 to 6, has no
   CommonMark spelling.
@@ -424,9 +445,8 @@ Right.
 Attributes and the carry fallback read as in the block sections, the carry in its inline form. Of
 the nodes below, `emoji`, `mention` and `status` spell their `text` attribute in the content slot as
 plain text: `[]` is the empty string, absent content is the absent attribute, non-empty content
-parsing to anything but one text node carrying neither marks, attributes nor content — adjacent text
+parsing to anything but one text node carrying neither marks, attributes nor content is a named
-nodes with identical marks and no attributes merged first — is a named error, and so is a `text` key
+error, and so is a `text` key in `{attrs}`. An enclosing mark spelling does not reach into the slot. The rest take no content,
-in `{attrs}`. An enclosing mark spelling does not reach into the slot. The rest take no content,
 `!adf:text` included; content on a node that takes none is a named error.
 
 - `date` — Attributes: `localId` (string), `timestamp` (string, epoch milliseconds).
@@ -453,16 +473,27 @@ Shipped !adf:emoji[🎉]{shortName=":tada:"} on !adf:date{timestamp=175608000000
 CommonMark strips or refuses one — a block's inline content edges, either side of a line break, an
 em, strong or strike spelling's inner edges, a pipe cell's edges — is spelled `!adf:text{text="…"}`,
 the reserved key carrying the node's text, escaped by the attribute grammar and never literal: pipe
-cells trim and pad. The emitter wraps the whitespace run alone and leaves the rest plain text;
+cells trim and pad. The emitter wraps the whitespace run alone and leaves the rest plain text, which
-`markdownToAdf` merges adjacent text nodes carrying identical marks and no attributes
+the spelled run joins on reading. Input reads that spelling alone: the value is one run of spaces
-(`docs/decisions.md` §Equality is editor-normal). Input reads that spelling alone: the value is one
+and tabs, or one run of newlines, and anything else — a mixed run, or text CommonMark carries
-run of spaces and tabs, or one run of newlines, and anything else — a mixed run, or text CommonMark
+plainly — is a named error.
-carries plainly — is a named error.
 
 ```
 !adf:text{text="  "}Two leading spaces held, and one text node split!adf:text{text="\n"}over two lines.
 ```
 
+**Adjacent text nodes.** CommonMark reads two adjacent text nodes back as one where neither is
+carried, neither holds `attrs` or an empty key, and their marks are identical, attributes included.
+The reserved leaf `!adf:textBreak{}` parts such a pair, inside every mark spelling the two share; a
+code span holds no directive, so it closes and reopens. It builds no node and reads only between
+two such nodes: elsewhere, or with `[content]` or `{attrs}`, it is a named error (`docs/decisions.md`
+§`!adf:textBreak{}` parts text CommonMark would join). A text node holding `attrs` or an empty key
+rides the inline carry.
+
+```
+Hello, !adf:textBreak{}world — **Hello, !adf:textBreak{}world** — `a`!adf:textBreak{}`b`
+```
+
 ## Marks
 
 An inline node's marks ride the spelling wrapped around them, never the block sections' reserved
@@ -490,14 +521,14 @@ the directive form, open to no literal reading, is a named error.
 A spelling adds its mark to every inline node it wraps, and nesting is the marks array in order,
 outermost first: `_!adf:underline[x]_` gives marks `[em, underline]`, `!adf:underline[_x_]` the
 reverse. `adfToMarkdown` nests in the order the array holds rather than sorting it —
-`docs/decisions.md` §Equality is editor-normal restores the array, not a set — and opens each
+`docs/decisions.md` §Equality is deep restores the array, not a set — and opens each spelling once
-spelling once over the longest run of adjacent inline nodes carrying an identical mark, attributes
+over the longest run of adjacent inline nodes carrying an identical mark, attributes included, at
-included, at that depth. A run breaks at every node the emitter carries, so no emitted carry sits
+that depth: `attrs: {}` differs from no `attrs`, and a directive spells it `{attrs=empty}`. A run
-inside a mark spelling.
+breaks at every node the emitter carries, so no emitted carry sits inside a mark spelling.
 
 An inline node whose marks no nesting spells — a mark type not listed here, an attrs key its
 spelling does not list, a value that is not the spelling's type, an attribute the spelling needs
-and the mark lacks, an order putting a code span outside another mark, `code` over anything but a
+and the mark lacks, an empty `attrs` on a mark CommonMark spells, an order putting a code span outside another mark, `code` over anything but a
 text node or over text holding a newline, or a spelling CommonMark's flanking rules cannot open or
 close where the run sits (`un**-real**istic`), or one CommonMark's matching pairs elsewhere — the
 intra-word `*` runs together with a neighbouring `**`, and the multiple-of-3 rule can leave the

src/adf/document.ts

@@ -1,6 +1,7 @@
 import type { ConvertFault } from '../result.ts'
 import { isJsonValue, overNested, type JsonValue } from '../json-value.ts'
 import { largestNesting } from '../nesting.ts'
+import { serializeCanonicalJson } from '../canonical-json.ts'
 
 export type AdfAttributes = { [key: string]: JsonValue }
 
@@ -17,6 +18,8 @@ export type AdfNode = {
   type: string
 }
 
+export type EmptyKey = 'attrs' | 'content' | 'marks'
+
 export type AdfDocument = {
   content?: AdfNode[]
   type: 'doc'
@@ -49,10 +52,45 @@ export function attributeNestingMessage(key: string, type: string): string {
 }
 
 export function carriesOnly(node: AdfNode, attributes: readonly string[]): boolean {
-  if (nodeMarks(node).length > 0 || node.text !== undefined) return false
+  if (nodeMarks(node).length > 0 || node.text !== undefined || emptyKeys(node).length > 0) return false
   return holdsOnly(nodeAttrs(node), attributes)
 }
 
+export function emptyKeys(held: { attrs?: AdfAttributes; content?: AdfNode[]; marks?: AdfMark[] }): EmptyKey[] {
+  const keys: EmptyKey[] = []
+  if (held.attrs !== undefined && Object.keys(held.attrs).length === 0) keys.push('attrs')
+  if (held.content?.length === 0) keys.push('content')
+  if (held.marks?.length === 0) keys.push('marks')
+  return keys
+}
+
+// A text node holding its text and marks alone: a format spells it as text, where anything more rides a carry.
+export function isPlainText(node: AdfNode): boolean {
+  return node.type === 'text' && node.attrs === undefined && node.content === undefined && node.marks?.length !== 0
+}
+
+export function identicalMark(left: AdfMark, right: AdfMark): boolean {
+  return marksKey([left]) === marksKey([right])
+}
+
+export function identicalMarks(left: readonly AdfMark[], right: readonly AdfMark[]): boolean {
+  return marksKey(left) === marksKey(right)
+}
+
+// Callers say when two nodes join.
+export function mergeAdjacentText(nodes: readonly AdfNode[], joins: (previous: AdfNode, node: AdfNode) => boolean): AdfNode[] {
+  const merged: AdfNode[] = []
+  for (const node of nodes) {
+    const previous = merged[merged.length - 1]
+    if (previous !== undefined && joins(previous, node)) {
+      merged[merged.length - 1] = { ...previous, text: `${previous.text ?? ''}${node.text ?? ''}` }
+      continue
+    }
+    merged.push(node)
+  }
+  return merged
+}
+
 // Depth is the walks' business, not the shape's: the guard waves a deep document through as blocks and marks do.
 export function isAdfDocument(value: unknown): value is AdfDocument {
   const fault = adfDocumentFault(value)
@@ -99,6 +137,13 @@ function isNodeArray(value: readonly unknown[]): value is readonly AdfNode[] {
   return true
 }
 
+function marksKey(marks: readonly AdfMark[]): string {
+  return serializeCanonicalJson(
+    marks.map((mark) => (mark.attrs === undefined ? [mark.type] : [mark.type, mark.attrs])),
+    'compact',
+  )
+}
+
 function nestingFault(nodes: readonly AdfNode[]): ConvertFault | undefined {
   const pending: AdfNode[] = [...nodes]
   while (pending.length > 0) {

src/adf/editor-normal.ts

@@ -1,34 +1,25 @@
 import type { AdfAttributes, AdfDocument, AdfMark, AdfNode } from './document.ts'
 import type { JsonValue } from '../json-value.ts'
-import { nodeAttrs, nodeContent, nodeMarks } from './document.ts'
+import { identicalMark, identicalMarks, mergeAdjacentText, nodeAttrs, nodeContent, nodeMarks } from './document.ts'
-import { serializeCanonicalJson } from '../canonical-json.ts'
 
 type JsonContainer = JsonValue[] | { [key: string]: JsonValue }
 
 type NodeHolder = { content?: AdfNode[] }
 
-export function sameMark(candidate: AdfMark, mark: AdfMark): boolean {
+export function sameMark(left: AdfMark, right: AdfMark): boolean {
-  return markKey(candidate) === markKey(mark)
+  return identicalMark(normalMark(left), normalMark(right))
 }
 
-export function mergeAdjacentText(nodes: readonly AdfNode[]): AdfNode[] {
+export function joinsNormally(previous: AdfNode, node: AdfNode): boolean {
-  const merged: AdfNode[] = []
+  if (!mergesText(previous) || !mergesText(node)) return false
-  for (const node of nodes) {
+  return identicalMarks(nodeMarks(previous).map(normalMark), nodeMarks(node).map(normalMark))
-    const previous = merged[merged.length - 1]
-    if (previous !== undefined && mergesText(previous) && mergesText(node) && sameMarks(previous, node)) {
-      merged[merged.length - 1] = { ...previous, text: `${previous.text ?? ''}${node.text ?? ''}` }
-      continue
-    }
-    merged.push(node)
-  }
-  return merged
 }
 
 export function toEditorNormal(document: AdfDocument): AdfDocument {
   const normal: AdfDocument = { type: document.type, version: Object.is(document.version, -0) ? 0 : document.version }
   const pending: { holder: NodeHolder; source: NodeHolder }[] = [{ holder: normal, source: document }]
   for (let entry = pending.pop(); entry !== undefined; entry = pending.pop()) {
-    const content = mergeAdjacentText(nodeContent(entry.source))
+    const content = mergeAdjacentText(nodeContent(entry.source), joinsNormally)
     if (content.length === 0) continue
     entry.holder.content = content.map((source) => {
       const holder = normalNode(source)
@@ -76,15 +67,3 @@ function normalValue(value: JsonValue, pending: JsonContainer[]): JsonValue {
 function mergesText(node: AdfNode): boolean {
   return node.type === 'text' && Object.keys(nodeAttrs(node)).length === 0
 }
-
-export function sameMarks(previous: AdfNode, node: AdfNode): boolean {
-  return marksKey(nodeMarks(previous)) === marksKey(nodeMarks(node))
-}
-
-function marksKey(marks: readonly AdfMark[]): string {
-  return marks.map(markKey).join('\n')
-}
-
-function markKey(mark: AdfMark): string {
-  return `${mark.type} ${serializeCanonicalJson(nodeAttrs(mark), 'compact')}`
-}

src/canonical-json.ts

@@ -18,7 +18,7 @@ export function serializeCanonicalJson(value: JsonValue, spelling: JsonSpelling)
     const { depth, value: held } = next
     if (Array.isArray(held)) schedule(pending, '[', held.map((item) => ({ label: '', value: item })), ']', indent, depth)
     else if (held !== null && typeof held === 'object') schedule(pending, '{', objectMembers(held, indent), '}', indent, depth)
-    else text.push(JSON.stringify(held))
+    else text.push(Object.is(held, -0) ? '-0' : JSON.stringify(held))
   }
   return text.join('')
 }

src/conformance/adf-property.test.ts

@@ -41,7 +41,7 @@ test('a generated document refuses to emit, or its markdown reads back to it', {
       if (!emitted.ok) return
       const read = markdownToAdf(emitted.value)
       assert.ok(read.ok, read.ok ? '' : `${read.error.code}: ${read.error.message} — reading ${JSON.stringify(emitted.value)}`)
-      assert.deepEqual(toEditorNormal(read.value), document, `reading ${JSON.stringify(emitted.value)}`)
+      assert.deepEqual(read.value, document, `reading ${JSON.stringify(emitted.value)}`)
     }),
     propertyRuns(gateRuns),
   )
@@ -62,3 +62,12 @@ test('a generated document writes plain markdown refusing only what the guard re
     propertyRuns(gateRuns),
   )
 })
+
+test('a generated document writes the plain markdown its editor-normal form writes', { timeout: propertyTimeout }, () => {
+  fc.assert(
+    fc.property(adfDocument, (document) => {
+      assert.deepEqual(adfToPlainMarkdown(document), adfToPlainMarkdown(toEditorNormal(document)))
+    }),
+    propertyRuns(gateRuns),
+  )
+})

src/conformance/corpus.test.ts

@@ -9,7 +9,6 @@ import { isAdfDocument } from '../adf/document.ts'
 import { isJsonValue } from '../json-value.ts'
 import { markdownToAdf } from '../markdown/parse/markdown-to-adf.ts'
 import { serializeCanonicalJson } from '../canonical-json.ts'
-import { toEditorNormal } from '../adf/editor-normal.ts'
 
 const corpusRoot = join(dirname(fileURLToPath(import.meta.url)), '..', '..', 'corpus')
 const errorsRoot = join(corpusRoot, 'errors')
@@ -97,7 +96,7 @@ for (const directory of roundTripDirectories) {
       assert.ok(isAdfDocument(expected), `${name}.json is not an ADF document`)
       const result = markdownToAdf(readFileSync(join(roundTripRoot, directory, `${name}.md`), 'utf8'))
       assert.ok(result.ok, result.ok ? '' : `${result.error.code}: ${result.error.message}`)
-      assert.deepEqual(toEditorNormal(result.value), expected)
+      assert.deepEqual(result.value, expected)
     })
   }
 }
@@ -125,12 +124,12 @@ for (const name of pairedNames(normalizationRoot, '.md', '.json')) {
     assert.ok(isAdfDocument(expected), `${name}.json is not an ADF document`)
     const result = markdownToAdf(readFileSync(join(normalizationRoot, `${name}.md`), 'utf8'))
     assert.ok(result.ok, result.ok ? '' : `${result.error.code}: ${result.error.message}`)
-    assert.deepEqual(toEditorNormal(result.value), expected)
+    assert.deepEqual(result.value, expected)
     const emitted = adfToMarkdown(result.value)
     assert.ok(emitted.ok, emitted.ok ? '' : `${emitted.error.code}: ${emitted.error.message}`)
     const again = markdownToAdf(emitted.value)
     assert.ok(again.ok, again.ok ? '' : `${again.error.code}: ${again.error.message}`)
-    assert.deepEqual(toEditorNormal(again.value), expected)
+    assert.deepEqual(again.value, expected)
   })
 }
 
@@ -146,7 +145,7 @@ for (const name of names(realPayloadsRoot, '.json')) {
     assert.ok(emitted.ok, emitted.ok ? '' : `${emitted.error.code}: ${emitted.error.message}`)
     const parsed = markdownToAdf(emitted.value)
     assert.ok(parsed.ok, parsed.ok ? '' : `${parsed.error.code}: ${parsed.error.message}`)
-    assert.deepEqual(toEditorNormal(parsed.value), payload)
+    assert.deepEqual(parsed.value, payload)
   })
 }
 

src/conformance/markdown-property.test.ts

@@ -31,7 +31,6 @@ import { markdownToAdf } from '../markdown/parse/markdown-to-adf.ts'
 import { nodeContent, nodeMarks } from '../adf/document.ts'
 import { serializeCanonicalJson } from '../canonical-json.ts'
 import { textDirectiveName } from '../markdown/text-directive.ts'
-import { toEditorNormal } from '../adf/editor-normal.ts'
 import { vocabularyPairs } from '../adf/attribute-vocabulary.ts'
 
 type Choice = { arbitrary: Arbitrary<string>; hostile?: true; weight: number }
@@ -433,7 +432,7 @@ test('generated markdown refuses, or what it parses to refuses to emit, or its s
       if (holdsDirectiveShape(parsed.value)) directiveShaped += 1
       const read = markdownToAdf(emitted.value)
       assert.ok(read.ok, read.ok ? '' : `${read.error.code}: ${read.error.message} — reading ${JSON.stringify(emitted.value)}`)
-      assert.deepEqual(toEditorNormal(read.value), toEditorNormal(parsed.value), `reading ${JSON.stringify(emitted.value)}`)
+      assert.deepEqual(read.value, parsed.value, `reading ${JSON.stringify(emitted.value)}`)
       const respelled = adfToMarkdown(read.value)
       assert.ok(respelled.ok, respelled.ok ? '' : `${respelled.error.code}: ${respelled.error.message} — spelling ${JSON.stringify(emitted.value)} again`)
       assert.equal(respelled.value, emitted.value)

src/conformance/property-harness.ts

@@ -2,16 +2,17 @@ import assert from 'node:assert/strict'
 import { env } from 'node:process'
 import fc from 'fast-check'
 
-import type { AdfAttributes, AdfDocument, AdfMark, AdfNode } from '../adf/document.ts'
+import type { AdfAttributes, AdfDocument, AdfMark, AdfNode, EmptyKey } from '../adf/document.ts'
 import type { Arbitrary } from 'fast-check'
 import type { AttributeKind, AttributeVocabulary } from '../adf/attribute-vocabulary.ts'
 import type { JsonValue } from '../json-value.ts'
 import { blockArgument } from '../markdown/block-directive.ts'
 import { blockNodes } from '../adf/block-nodes.ts'
 import { directivePrefix } from '../markdown/directive-syntax.ts'
+import { emptyKeys, mergeAdjacentText } from '../adf/document.ts'
 import { inlineNodes } from '../adf/inline-nodes.ts'
+import { joinsNormally } from '../adf/editor-normal.ts'
 import { markAttributes } from '../adf/mark-attributes.ts'
-import { toEditorNormal } from '../adf/editor-normal.ts'
 
 type Positions = { block: AdfNode; inline: AdfNode }
 
@@ -36,7 +37,11 @@ export function textOf(minLength: number): Arbitrary<string> {
 
 const text = textOf(1)
 const unknownType = fc.oneof(fc.stringMatching(/^[a-z][A-Za-z0-9]{0,7}$/), text).filter((type) => !spelledTypes.has(type))
-const numberValue = fc.oneof({ arbitrary: fc.integer({ max: 10, min: -1 }), weight: 3 }, { arbitrary: fc.double({ noDefaultInfinity: true, noNaN: true }), weight: 1 })
+const numberValue = fc.oneof(
+  { arbitrary: fc.integer({ max: 10, min: -1 }), weight: 6 },
+  { arbitrary: fc.double({ noDefaultInfinity: true, noNaN: true }), weight: 2 },
+  { arbitrary: fc.constant(-0), weight: 1 },
+)
 
 // V8's JSON.parse returns a wrong key after parsing a key holding an escaped backslash (https://issues.chromium.org/issues/521080746); Bun is unaffected.
 const keyPiece = fc
@@ -73,19 +78,35 @@ function heldAttributes(held: Readonly<Record<string, JsonValue | undefined>>):
   return attrs
 }
 
+// An empty attrs, content or marks key, and adjacent text a reader would merge, stay occasional: each takes a spelling outside CommonMark.
+// The copy gives fast-check's null-prototype records the prototype a parsed node has.
+function occasionallyEmpty<T extends AdfMark | AdfNode>(arbitrary: Arbitrary<T>): Arbitrary<T> {
+  return fc.tuple(arbitrary, fc.nat({ max: 9 })).map(([held, roll]) => (roll === 0 ? { ...held } : withoutEmptyKeys(held)))
+}
+
+function withoutEmptyKeys<T extends AdfMark | AdfNode>(held: T): T {
+  const kept: Partial<Record<EmptyKey, unknown>> & T = { ...held }
+  for (const key of emptyKeys(held)) delete kept[key]
+  return kept
+}
+
+function occasionallyApart(arbitrary: Arbitrary<AdfNode[]>): Arbitrary<AdfNode[]> {
+  return fc.tuple(arbitrary, fc.nat({ max: 5 })).map(([nodes, roll]) => (roll === 0 ? nodes : mergeAdjacentText(nodes, joinsNormally)))
+}
+
 function pipeTable({ body, header }: { body: AdfNode[][]; header: AdfNode[] }): AdfNode {
   const rows = [header, ...body.map((cells) => header.map((_, index) => cells[index] ?? emptyCell))]
   return { content: rows.map((content): AdfNode => ({ content, type: 'tableRow' })), type: 'table' }
 }
 
-const mark: Arbitrary<AdfMark> = fc.oneof(
+const mark: Arbitrary<AdfMark> = occasionallyEmpty(fc.oneof(
   { arbitrary: fc.oneof(...Object.entries(markAttributes).map(([type, vocabulary]) => attributes(vocabulary).map((attrs) => ({ attrs, type })))), weight: 9 },
   { arbitrary: attributes({ color: 'string' }).map((attrs) => ({ attrs, type: 'backgroundColor' })), weight: 2 },
   { arbitrary: fc.record({ attrs: fc.dictionary(jsonKey, jsonValue, { maxKeys: 2, noNullPrototype: true }), type: unknownType }), weight: 1 },
-)
+))
 const marks = fc.uniqueArray(mark, { maxLength: 3, selector: (held) => held.type })
 
-const textNode = fc.record({ marks, text }).map((held): AdfNode => ({ ...held, type: 'text' }))
+const textNode = occasionallyEmpty(fc.record({ marks, text }).map((held): AdfNode => ({ ...held, type: 'text' })))
 
 const backtickRunNode = fc
   .record({ marks: fc.oneof(fc.constant<AdfMark[]>([]), fc.constant<AdfMark[]>([{ type: 'code' }]), marks), text: fc.string({ maxLength: 6, minLength: 1, unit: fc.constantFrom('`', '``', ' ', 'a') }) })
@@ -96,7 +117,7 @@ const autolinkTextNode = fc
   .map(({ href, marks: held }): AdfNode => ({ marks: [...held.filter((outer) => outer.type !== 'link'), { attrs: { href }, type: 'link' }], text: href, type: 'text' }))
 
 const inlineArbitraries = Object.entries(inlineNodes).map(([type, model]) =>
-  fc.record({ attrs: attributes(model.attributes), marks }).map((held): AdfNode => ({ ...held, type })),
+  occasionallyEmpty(fc.record({ attrs: attributes(model.attributes), marks }).map((held): AdfNode => ({ ...held, type }))),
 )
 
 function weighted(arbitraries: readonly Arbitrary<AdfNode>[], weight: number): { arbitrary: Arbitrary<AdfNode>; weight: number }[] {
@@ -105,10 +126,10 @@ function weighted(arbitraries: readonly Arbitrary<AdfNode>[], weight: number): {
 
 const positions = fc.letrec<Positions>((tie) => {
   const blockContent = fc.array(tie('block'), { depthIdentifier, maxLength: 3 })
-  const inlineContent = fc.array(tie('inline'), { depthIdentifier, maxLength: 4 })
+  const inlineContent = occasionallyApart(fc.array(tie('inline'), { depthIdentifier, maxLength: 4 }))
   const contentByModel = {
     block: blockContent,
-    code: fc.array(text.map((held): AdfNode => ({ text: held, type: 'text' })), { maxLength: 2 }),
+    code: occasionallyApart(fc.array(text.map((held): AdfNode => ({ text: held, type: 'text' })), { maxLength: 2 })),
     inline: inlineContent,
     none: fc.constant<AdfNode[]>([]),
   }
@@ -116,35 +137,40 @@ const positions = fc.letrec<Positions>((tie) => {
   const blockArbitraries = Object.entries(blockNodes).map(([type, model]) => {
     const argument = blockArgument(type)
     const vocabulary: AttributeVocabulary = argument === undefined ? model.attributes : { ...model.attributes, [argument]: 'string' }
-    const node = fc.record({ attrs: attributes(vocabulary), content: contentByModel[model.contentModel], marks: blockMarks }).map((held): AdfNode => ({ ...held, type }))
+    const node = occasionallyEmpty(fc.record({ attrs: attributes(vocabulary), content: contentByModel[model.contentModel], marks: blockMarks }).map((held): AdfNode => ({ ...held, type })))
     return { leaf: model.contentModel === 'code' || model.contentModel === 'none', node }
   })
-  const unknownNode = fc
+  const unknownNode = occasionallyEmpty(
-    .record({ attrs: fc.dictionary(jsonKey, jsonValue, { maxKeys: 2, noNullPrototype: true }), content: fc.array(tie('inline'), { depthIdentifier, maxLength: 2 }), marks, type: unknownType })
+    fc.record({ attrs: fc.dictionary(jsonKey, jsonValue, { maxKeys: 2, noNullPrototype: true }), content: fc.array(tie('inline'), { depthIdentifier, maxLength: 2 }), marks, type: unknownType }),
-    .map((held): AdfNode => held)
+  )
   const leafBlocks = blockArbitraries.filter((entry) => entry.leaf).map((entry) => entry.node)
   const containerBlocks = blockArbitraries.filter((entry) => !entry.leaf).map((entry) => entry.node)
   const misplacedWeight = 7
-  const paragraph = fc.oneof({ arbitrary: inlineContent, weight: 3 }, { arbitrary: fc.array(backtickRunNode, { maxLength: 4, minLength: 2 }), weight: 1 }).map((content): AdfNode => ({ content, type: 'paragraph' }))
+  const paragraph = occasionallyEmpty(
+    fc.oneof({ arbitrary: inlineContent, weight: 3 }, { arbitrary: occasionallyApart(fc.array(backtickRunNode, { maxLength: 4, minLength: 2 })), weight: 1 }).map((content): AdfNode => ({ content, type: 'paragraph' })),
+  )
   const cell = (type: string) => paragraph.map((held): AdfNode => ({ content: [held], type }))
   const listItems = fc.array(
-    blockContent.map((content): AdfNode => ({ content, type: 'listItem' })),
+    occasionallyEmpty(blockContent.map((content): AdfNode => ({ content, type: 'listItem' }))),
     { depthIdentifier, maxLength: 3, minLength: 1 },
   )
   const flatCommonMarkShapes = [
-    fc.record({ content: inlineContent, level: fc.integer({ max: 6, min: 1 }) }).map(({ content, level }): AdfNode => ({ attrs: { level }, content, type: 'heading' })),
+    occasionallyEmpty(fc.record({ content: inlineContent, level: fc.integer({ max: 6, min: 1 }) }).map(({ content, level }): AdfNode => ({ attrs: { level }, content, type: 'heading' }))),
     paragraph,
     fc.record({ body: fc.array(fc.array(cell('tableCell'), { maxLength: 3 }), { maxLength: 2 }), header: fc.array(cell('tableHeader'), { maxLength: 3, minLength: 1 }) }).map(pipeTable),
   ]
   const task = (type: string, content: Arbitrary<AdfNode[]>) =>
-    fc.record({ content, state: fc.constantFrom('DONE', 'TODO') }).map(({ content: held, state }): AdfNode => ({ attrs: { state }, content: held, type }))
+    occasionallyEmpty(fc.record({ content, state: fc.constantFrom('DONE', 'TODO') }).map(({ content: held, state }): AdfNode => ({ attrs: { state }, content: held, type })))
   const taskItem = fc.oneof({ arbitrary: task('taskItem', inlineContent), weight: 3 }, { arbitrary: task('blockTaskItem', blockContent), weight: 1 })
   const nestingCommonMarkShapes = [
-    blockContent.map((content): AdfNode => ({ content, type: 'blockquote' })),
+    occasionallyEmpty(blockContent.map((content): AdfNode => ({ content, type: 'blockquote' }))),
     fc.array(fc.oneof({ arbitrary: taskItem, weight: 3 }, { arbitrary: tie('block'), weight: 1 }), { depthIdentifier, maxLength: 3, minLength: 1 }).map((content): AdfNode => ({ content, type: 'taskList' })),
     listItems.map((content): AdfNode => ({ content, type: 'bulletList' })),
     fc
-      .record({ content: listItems, order: fc.oneof({ arbitrary: fc.integer({ max: 3, min: 0 }), weight: 4 }, { arbitrary: fc.integer({ max: 999999999, min: 0 }), weight: 1 }) })
+      .record({
+        content: listItems,
+        order: fc.oneof({ arbitrary: fc.integer({ max: 3, min: 0 }), weight: 8 }, { arbitrary: fc.integer({ max: 999999999, min: 0 }), weight: 2 }, { arbitrary: fc.constant(-0), weight: 1 }),
+      })
       .map(({ content, order }): AdfNode => ({ attrs: { order }, content, type: 'orderedList' })),
   ]
   const flatBlocks = [...weighted(leafBlocks, 2), ...weighted(flatCommonMarkShapes, flatCommonMarkShapeWeight)]
@@ -167,7 +193,11 @@ const positions = fc.letrec<Positions>((tie) => {
   }
 })
 
-export const adfDocument = fc.array(positions.block, { depthIdentifier, maxLength: 4, minLength: 1 }).map((content): AdfDocument => toEditorNormal({ content, type: 'doc', version: 1 }))
+export const adfDocument = fc.oneof(
+  { arbitrary: fc.array(positions.block, { depthIdentifier, maxLength: 4, minLength: 1 }).map((content): AdfDocument => ({ content, type: 'doc', version: 1 })), weight: 30 },
+  { arbitrary: fc.constant<AdfDocument>({ content: [], type: 'doc', version: 1 }), weight: 1 },
+  { arbitrary: fc.constant<AdfDocument>({ type: 'doc', version: 1 }), weight: 1 },
+)
 
 export function propertyRuns(gateRuns: number): { gate: boolean; numRuns: number; seed?: number } {
   const deepRuns = env[deepRunsVariable]

src/markdown/block-directive.ts

@@ -2,9 +2,9 @@ import type { AdfMark } from '../adf/document.ts'
 import type { BlockType } from '../adf/block-nodes.ts'
 import type { JsonValue } from '../json-value.ts'
 import { blockNodeModel } from '../adf/block-nodes.ts'
-import { isAdfMark, nodeAttrs } from '../adf/document.ts'
+import { isAdfMark } from '../adf/document.ts'
 import { serializeCanonicalJson } from '../canonical-json.ts'
-import { spellDirectiveOpener } from './directive-syntax.ts'
+import { spellAttributes, spellDirectiveOpener } from './directive-syntax.ts'
 
 const argumentByType = new Map(
   Object.entries({
@@ -14,6 +14,13 @@ const argumentByType = new Map(
   } satisfies Partial<Record<BlockType, string>>),
 )
 
+export const documentName = 'doc'
+
+// spec/flavour.md, Directives: a document holding no content key, which the empty string cannot spell.
+export const documentAttribute = { key: 'content', value: 'none' } as const
+
+export const documentSpelling = spellDirectiveOpener(documentName, undefined, spellAttributes([[documentAttribute.key, documentAttribute.value]]))
+
 export const listBreakName = 'listBreak'
 
 export const listBreakSpelling = spellDirectiveOpener(listBreakName, undefined, '')
@@ -25,17 +32,14 @@ export function blockArgument(type: string): string | undefined {
 }
 
 export function blockDirectiveForm(name: string): 'container' | 'leaf' | undefined {
-  if (name === listBreakName) return 'leaf'
+  if (name === listBreakName || name === documentName) return 'leaf'
   const model = blockNodeModel(name)
   if (model === undefined) return undefined
   return model.contentModel === 'none' ? 'leaf' : 'container'
 }
 
 export function markValues(marks: readonly AdfMark[]): JsonValue {
-  return marks.map((mark) => {
+  return marks.map((mark) => (mark.attrs === undefined ? { type: mark.type } : { attrs: mark.attrs, type: mark.type }))
-    const attrs = nodeAttrs(mark)
-    return Object.keys(attrs).length === 0 ? { type: mark.type } : { attrs, type: mark.type }
-  })
 }
 
 export function readMarkValues(value: JsonValue): AdfMark[] | undefined {

src/markdown/code-language.ts

@@ -1,14 +1,12 @@
 import type { JsonValue } from '../json-value.ts'
-import { carryName } from './opaque-carry.ts'
+import { carryFenceType } from './opaque-carry.ts'
-import { holdsControlCharacter } from './commonmark/grammar.ts'
+import { infoStringCarries } from './commonmark/grammar.ts'
-import { holdsEntityReference } from './commonmark/entity-references.ts'
 
 export type LanguageSlot = { info: string; kind: 'fence' } | { kind: 'attribute' } | { kind: 'none' }
 
 // spec/flavour.md, The CommonMark blocks: the one slot a codeBlock's language rides.
 export function languageSlot(language: JsonValue | undefined): LanguageSlot {
   if (language === undefined) return { kind: 'none' }
-  if (typeof language !== 'string' || language === '' || language === carryName) return { kind: 'attribute' }
+  if (typeof language !== 'string' || carryFenceType(language) !== undefined || !infoStringCarries(language)) return { kind: 'attribute' }
-  if (/[`\\]/.test(language) || holdsControlCharacter(language) || language !== language.trim() || holdsEntityReference(language)) return { kind: 'attribute' }
   return { info: language, kind: 'fence' }
 }

src/markdown/commonmark/grammar.ts

@@ -1,4 +1,4 @@
-import { readEntityReference, replacementCharacter } from './entity-references.ts'
+import { holdsEntityReference, readEntityReference, replacementCharacter } from './entity-references.ts'
 
 export type LinePosition = 'first' | 'later'
 
@@ -133,6 +133,11 @@ export function holdsControlCharacter(text: string): boolean {
   return controlCharacter.test(text)
 }
 
+// What a backtick fence's info string reads back verbatim: escapes and entity references decode, and the edges trim.
+export function infoStringCarries(text: string): boolean {
+  return text !== '' && !/[`\\]/.test(text) && !holdsControlCharacter(text) && text === text.trim() && !holdsEntityReference(text)
+}
+
 export function holdsNullCharacter(text: string): boolean {
   return nullCharacter.test(text)
 }

src/markdown/directive-syntax.ts

@@ -139,7 +139,7 @@ export function spellStringAttribute(text: string): string {
 export function spellAttributeValue(value: VocabularyValue): string {
   if (value.kind === 'boolean') return `${value.value}`
   if (value.kind === 'json') return spellJsonAttribute(value.value)
-  if (value.kind === 'number') return spellStringAttribute(JSON.stringify(value.value))
+  if (value.kind === 'number') return spellStringAttribute(serializeCanonicalJson(value.value, 'compact'))
   return spellStringAttribute(value.value)
 }
 

src/markdown/emit/adf-to-markdown.test.ts

@@ -6,14 +6,13 @@ import type { JsonValue } from '../../json-value.ts'
 import type { Result } from '../../result.ts'
 import { adfToMarkdown, markdownToAdf } from '../../index.ts'
 import { largestNesting } from '../../nesting.ts'
-import { toEditorNormal } from '../../adf/editor-normal.ts'
 
 function document(...content: AdfNode[]): AdfDocument {
   return { content, type: 'doc', version: 1 }
 }
 
 function paragraph(...content: AdfNode[]): AdfNode {
-  return { content, type: 'paragraph' }
+  return content.length === 0 ? { type: 'paragraph' } : { content, type: 'paragraph' }
 }
 
 function code(result: Result<string>): string {
@@ -168,21 +167,24 @@ test('parts two adjacent lists of the same kind, the marker spelling being what
   const nested: AdfNode = { content: [{ content: [list, list], type: 'listItem' }], type: 'bulletList' }
   assert.equal(markdown(adfToMarkdown(document(nested))), '- - x\n\n  !adf:listBreak\n\n  - x\n')
   const carried: AdfNode = { ...list, attrs: { unknown: 'x' } }
-  assert.ok(markdown(adfToMarkdown(document(carried, carried))).includes('```\n\n```carry\n'))
+  assert.ok(markdown(adfToMarkdown(document(carried, carried))).includes('```\n\n```adf:bulletList\n'))
   assert.ok(markdown(adfToMarkdown(document(carried, list))).endsWith('```\n\n- x\n'))
-  assert.ok(markdown(adfToMarkdown(document(list, carried))).startsWith('- x\n\n```carry\n'))
+  assert.ok(markdown(adfToMarkdown(document(list, carried))).startsWith('- x\n\n```adf:bulletList\n'))
 })
 
-test('carries a node type no section spells', () => {
+test('carries a node type no section spells, the fence naming the type wherever an info string carries it', () => {
-  assert.equal(markdown(adfToMarkdown(document({ type: 'blockCard' }))), '```carry\n{\n  "type": "blockCard"\n}\n```\n')
+  assert.equal(markdown(adfToMarkdown(document({ type: 'blockCard' }))), '```adf:blockCard\n{}\n```\n')
-  assert.equal(markdown(adfToMarkdown(document({ type: 'toString' }))), '```carry\n{\n  "type": "toString"\n}\n```\n')
+  assert.equal(markdown(adfToMarkdown(document({ type: 'toString' }))), '```adf:toString\n{}\n```\n')
   assert.equal(markdown(adfToMarkdown(document(paragraph({ type: 'blockCard' })))), '!adf:carry{json="{\\"type\\":\\"blockCard\\"}"}\n')
-  assert.equal(markdown(adfToMarkdown(document({ text: 'x', type: 'text' }))), '```carry\n{\n  "text": "x",\n  "type": "text"\n}\n```\n')
+  assert.equal(markdown(adfToMarkdown(document({ text: 'x', type: 'text' }))), '```adf:text\n{\n  "text": "x"\n}\n```\n')
-  assert.equal(markdown(adfToMarkdown(document({ type: 'hardBreak' }))), '```carry\n{\n  "type": "hardBreak"\n}\n```\n')
+  assert.equal(markdown(adfToMarkdown(document({ type: 'hardBreak' }))), '```adf:hardBreak\n{}\n```\n')
+  assert.equal(markdown(adfToMarkdown(document({ type: 'a&amp;b' }))), '```adf:\n{\n  "type": "a&amp;b"\n}\n```\n')
+  assert.equal(markdown(adfToMarkdown(document({ type: 'a\\b' }))), '```adf:\n{\n  "type": "a\\\\b"\n}\n```\n')
 })
 
-test('spells the code block whose language is the reserved info string', () => {
+test('spells the code block whose language opens with the reserved info string prefix', () => {
-  assert.equal(markdown(adfToMarkdown(document({ attrs: { language: 'carry' }, type: 'codeBlock' }))), '!adf:codeBlock {language=carry}\n```\n```\n!adf:/codeBlock\n')
+  assert.equal(markdown(adfToMarkdown(document({ attrs: { language: 'adf:x' }, type: 'codeBlock' }))), '!adf:codeBlock {language="adf:x"}\n```\n```\n!adf:/codeBlock\n')
+  assert.equal(markdown(adfToMarkdown(document({ attrs: { language: 'carry' }, type: 'codeBlock' }))), '```carry\n```\n')
   assert.equal(markdown(adfToMarkdown(document({ attrs: { language: 'adf' }, type: 'codeBlock' }))), '```adf\n```\n')
 })
 
@@ -208,21 +210,17 @@ test('refuses a carried node nested deeper than the levels its position leaves',
   assert.equal(code(adfToMarkdown(document(quoted))), 'unsupported-nesting-depth')
 })
 
-test('refuses a node whose content model the canonical form cannot emit', () => {
+test('carries a code block holding a node no fence holds', () => {
-  assert.equal(
+  assert.equal(markdown(adfToMarkdown(document({ content: [paragraph()], type: 'codeBlock' }))), '```adf:codeBlock\n{\n  "content": [\n    {\n      "type": "paragraph"\n    }\n  ]\n}\n```\n')
-    markdown(adfToMarkdown(document({ content: [paragraph()], type: 'codeBlock' }))),
+  for (const child of [{ attrs: {}, text: 'x', type: 'text' }, { content: [], text: 'x', type: 'text' }, { marks: [], text: 'x', type: 'text' }, { content: [{ text: 'lost', type: 'text' }], text: 'x', type: 'text' }]) {
-    'unsupported-node-shape: a codeBlock holds plain text nodes only: this paragraph node is not one',
+    assert.ok(markdown(adfToMarkdown(document({ content: [child], type: 'codeBlock' }))).startsWith('```adf:codeBlock\n'), JSON.stringify(child))
-  )
+  }
-  assert.equal(
-    markdown(adfToMarkdown(document({ content: [{ content: [{ text: 'lost', type: 'text' }], text: 'x', type: 'text' }], type: 'codeBlock' }))),
-    'unsupported-node-shape: a codeBlock holds plain text nodes only: this text node is not one',
-  )
 })
 
 test('spells a list its own content shape cannot hold as a directive', () => {
   assert.equal(markdown(adfToMarkdown(document({ content: [paragraph()], type: 'bulletList' }))), '!adf:bulletList\n!adf:paragraph\n!adf:/paragraph\n!adf:/bulletList\n')
   assert.equal(markdown(adfToMarkdown(document({ type: 'bulletList' }))), '!adf:bulletList\n!adf:/bulletList\n')
-  assert.equal(markdown(adfToMarkdown(document({ attrs: { order: 2 }, content: [], type: 'orderedList' }))), '!adf:orderedList {order=2}\n!adf:/orderedList\n')
+  assert.equal(markdown(adfToMarkdown(document({ attrs: { order: 2 }, content: [], type: 'orderedList' }))), '!adf:orderedList {content=empty order=2}\n!adf:/orderedList\n')
 })
 
 test('spells an ordered list no marker fits as a directive', () => {
@@ -334,7 +332,7 @@ test('refuses a node carrying one mark type twice', () => {
 })
 
 test('parts a nested list the tight spelling would swallow from the block above it', () => {
-  const item = (...content: AdfNode[]): AdfNode => ({ content, type: 'listItem' })
+  const item = (...content: AdfNode[]): AdfNode => (content.length === 0 ? { type: 'listItem' } : { content, type: 'listItem' })
   const text = (value: string): AdfNode => ({ content: [{ text: value, type: 'text' }], type: 'paragraph' })
   const outer = (...content: AdfNode[]): AdfDocument => document({ content: [item(...content)], type: 'bulletList' })
   const ordered: AdfNode = { attrs: { order: 2 }, content: [item(text('b'))], type: 'orderedList' }
@@ -345,7 +343,7 @@ test('parts a nested list the tight spelling would swallow from the block above
   const list: AdfNode = { content: [item(text('b'))], type: 'bulletList' }
   const panel: AdfNode = { attrs: { panelType: 'info' }, content: [text('p')], type: 'panel' }
   assert.equal(markdown(adfToMarkdown(outer(panel, list))), '- !adf:panel info\n  p\n  !adf:/panel\n\n  - b\n')
-  assert.ok(markdown(adfToMarkdown(outer(text('a'), { ...list, attrs: { unknown: 'x' } }))).startsWith('- a\n\n  ```carry\n'))
+  assert.ok(markdown(adfToMarkdown(outer(text('a'), { ...list, attrs: { unknown: 'x' } }))).startsWith('- a\n\n  ```adf:bulletList\n'))
 })
 
 test('refuses marks and attributes nested deeper than the emitter carries', () => {
@@ -372,7 +370,7 @@ test('refuses marks and attributes nested deeper than the emitter carries', () =
     assert.ok(spelled.ok, spelled.ok ? '' : spelled.error.message)
     const read = markdownToAdf(spelled.value)
     assert.ok(read.ok, read.ok ? '' : read.error.message)
-    assert.deepEqual(toEditorNormal(read.value), document(node))
+    assert.deepEqual(read.value, document(node))
   }
 
   assert.equal(markdown(adfToMarkdown(document(paragraph({ marks: [{ attrs, type: 'em' }], text: 'x', type: 'text' })))), deeper('depth', 'em'))
@@ -442,7 +440,7 @@ test('escapes a hyphen underline a hard break would expose', () => {
 })
 
 test('spells a list item whose marker completes a thematic break as a directive', () => {
-  const item = (...content: AdfNode[]): AdfNode => ({ content, type: 'listItem' })
+  const item = (...content: AdfNode[]): AdfNode => (content.length === 0 ? { type: 'listItem' } : { content, type: 'listItem' })
   assert.equal(markdown(adfToMarkdown(document({ content: [item({ type: 'rule' })], type: 'bulletList' }))), '!adf:bulletList\n!adf:listItem\n---\n!adf:/listItem\n!adf:/bulletList\n')
   const nested: AdfNode = { content: [item({ content: [item()], type: 'bulletList' })], type: 'bulletList' }
   assert.equal(markdown(adfToMarkdown(document(nested))), '- -\n')
@@ -469,10 +467,7 @@ test('refuses the characters CommonMark rewrites', () => {
   )
   assert.equal(code(adfToMarkdown(document(paragraph({ text: 'a\u0000b', type: 'text' })))), 'unspellable-character')
   assert.equal(code(adfToMarkdown(document({ content: [{ text: 'a\u0000b', type: 'text' }], type: 'codeBlock' }))), 'unspellable-character')
-  assert.equal(
+  assert.equal(markdown(adfToMarkdown(document({ content: [{ text: '', type: 'text' }], type: 'codeBlock' }))), 'unsupported-node-shape: a text node holds text: this one has none')
-    markdown(adfToMarkdown(document({ content: [{ text: '', type: 'text' }], type: 'codeBlock' }))),
-    'unsupported-node-shape: a codeBlock holds plain text nodes only: this text node is not one',
-  )
 })
 
 test('refuses a text node the spelling would empty out', () => {
@@ -501,11 +496,11 @@ test('pads a code span whose edges CommonMark would strip', () => {
   assert.equal(markdown(adfToMarkdown(document(paragraph({ marks: [{ type: 'code' }], text: ' \t ', type: 'text' })))), '`  \t  `\n')
 })
 
-test('spells one code span over a run of code-marked nodes', () => {
+test('spells a code span per code-marked node, parted by the text break', () => {
   const code_ = { type: 'code' }
   assert.equal(
     markdown(adfToMarkdown(document(paragraph({ marks: [code_], text: 'a', type: 'text' }, { marks: [code_], text: 'b', type: 'text' })))),
-    '`ab`\n',
+    '`a`!adf:textBreak{}`b`\n',
   )
 })
 
@@ -545,7 +540,7 @@ test('emits an empty list item without trailing whitespace', () => {
 test('spells a block directive as its node type, arg and attributes', () => {
   const panel = (attrs: AdfAttributes): AdfDocument => document({ attrs, content: [paragraph({ text: 'x', type: 'text' })], type: 'panel' })
   assert.equal(markdown(adfToMarkdown(panel({ panelType: 'warning' }))), '!adf:panel warning\nx\n!adf:/panel\n')
-  assert.equal(markdown(adfToMarkdown(panel({}))), '!adf:panel\nx\n!adf:/panel\n')
+  assert.equal(markdown(adfToMarkdown(panel({}))), '!adf:panel {attrs=empty}\nx\n!adf:/panel\n')
   assert.equal(markdown(adfToMarkdown(document({ content: [{ text: 'x', type: 'text' }], type: 'caption' }))), '!adf:caption\nx\n!adf:/caption\n')
   assert.equal(markdown(adfToMarkdown(document({ type: 'caption' }))), '!adf:caption\n!adf:/caption\n')
   assert.equal(markdown(adfToMarkdown(document({ attrs: { localId: 'a' }, type: 'syncBlock' }))), '!adf:syncBlock {localId=a}\n')
@@ -553,20 +548,20 @@ test('spells a block directive as its node type, arg and attributes', () => {
 
 test('carries a directive attribute no section spells', () => {
   const carried = (node: AdfNode): string => markdown(adfToMarkdown(document(node)))
-  assert.equal(carried({ attrs: { rounded: true }, type: 'panel' }), '```carry\n{\n  "attrs": {\n    "rounded": true\n  },\n  "type": "panel"\n}\n```\n')
+  assert.equal(carried({ attrs: { rounded: true }, type: 'panel' }), '```adf:panel\n{\n  "attrs": {\n    "rounded": true\n  }\n}\n```\n')
-  assert.equal(carried({ attrs: { toString: 'x' }, type: 'panel' }), '```carry\n{\n  "attrs": {\n    "toString": "x"\n  },\n  "type": "panel"\n}\n```\n')
+  assert.equal(carried({ attrs: { toString: 'x' }, type: 'panel' }), '```adf:panel\n{\n  "attrs": {\n    "toString": "x"\n  }\n}\n```\n')
-  assert.equal(carried({ attrs: { localId: 4 }, type: 'panel' }), '```carry\n{\n  "attrs": {\n    "localId": 4\n  },\n  "type": "panel"\n}\n```\n')
+  assert.equal(carried({ attrs: { localId: 4 }, type: 'panel' }), '```adf:panel\n{\n  "attrs": {\n    "localId": 4\n  }\n}\n```\n')
   assert.equal(
     carried({ attrs: { width: '50' }, type: 'layoutColumn' }),
-    '```carry\n{\n  "attrs": {\n    "width": "50"\n  },\n  "type": "layoutColumn"\n}\n```\n',
+    '```adf:layoutColumn\n{\n  "attrs": {\n    "width": "50"\n  }\n}\n```\n',
   )
   assert.equal(
     carried({ attrs: { isNumberColumnEnabled: 'true' }, type: 'table' }),
-    '```carry\n{\n  "attrs": {\n    "isNumberColumnEnabled": "true"\n  },\n  "type": "table"\n}\n```\n',
+    '```adf:table\n{\n  "attrs": {\n    "isNumberColumnEnabled": "true"\n  }\n}\n```\n',
   )
   assert.equal(
     carried({ content: [{ attrs: { alt: 4 }, type: 'media' }], type: 'mediaGroup' }),
-    '!adf:mediaGroup\n```carry\n{\n  "attrs": {\n    "alt": 4\n  },\n  "type": "media"\n}\n```\n!adf:/mediaGroup\n',
+    '!adf:mediaGroup\n```adf:media\n{\n  "attrs": {\n    "alt": 4\n  }\n}\n```\n!adf:/mediaGroup\n',
   )
 })
 
@@ -574,15 +569,16 @@ test('carries an arg slot value no bare token spells', () => {
   const carried = (node: AdfNode): string => markdown(adfToMarkdown(document(node)))
   assert.equal(
     carried({ attrs: { panelType: 'extra info' }, type: 'panel' }),
-    '```carry\n{\n  "attrs": {\n    "panelType": "extra info"\n  },\n  "type": "panel"\n}\n```\n',
+    '```adf:panel\n{\n  "attrs": {\n    "panelType": "extra info"\n  }\n}\n```\n',
   )
-  assert.equal(carried({ attrs: { state: 2 }, type: 'taskItem' }), '```carry\n{\n  "attrs": {\n    "state": 2\n  },\n  "type": "taskItem"\n}\n```\n')
+  assert.equal(carried({ attrs: { state: 2 }, type: 'taskItem' }), '```adf:taskItem\n{\n  "attrs": {\n    "state": 2\n  }\n}\n```\n')
 })
 
 test('carries a block node mark in the reserved attribute', () => {
   const section = (...marks: AdfMark[]): AdfDocument => document({ marks, type: 'layoutSection' })
   assert.equal(markdown(adfToMarkdown(section({ type: 'breakout' }))), '!adf:layoutSection {marks="[{\\"type\\":\\"breakout\\"}]"}\n!adf:/layoutSection\n')
-  assert.equal(markdown(adfToMarkdown(section({ attrs: {}, type: 'breakout' }))), '!adf:layoutSection {marks="[{\\"type\\":\\"breakout\\"}]"}\n!adf:/layoutSection\n')
+  assert.equal(markdown(adfToMarkdown(section({ attrs: {}, type: 'breakout' }))), '!adf:layoutSection {marks="[{\\"attrs\\":{},\\"type\\":\\"breakout\\"}]"}\n!adf:/layoutSection\n')
+  assert.equal(markdown(adfToMarkdown(section())), '!adf:layoutSection {marks=empty}\n!adf:/layoutSection\n')
 })
 
 test('refuses the content a directive body has no room for', () => {
@@ -604,7 +600,7 @@ test('separates blocks in a container body by a blank line only where a directiv
 test('spells the image form for exactly the centered external media shape', () => {
   const url = 'https://example.com/moon.png'
   const single = (attrs: AdfAttributes, ...content: AdfNode[]): AdfDocument =>
-    document({ attrs: { layout: 'center' }, content: [{ attrs, content, type: 'media' }], type: 'mediaSingle' })
+    document({ attrs: { layout: 'center' }, content: [content.length === 0 ? { attrs, type: 'media' } : { attrs, content, type: 'media' }], type: 'mediaSingle' })
   assert.equal(markdown(adfToMarkdown(single({ alt: 'The moon', type: 'external', url }))), `![The moon](${url})\n`)
   assert.equal(markdown(adfToMarkdown(single({ type: 'external', url }))), `![](${url})\n`)
   assert.equal(markdown(adfToMarkdown(single({ alt: 'a [b] c', type: 'external', url }))), `![a \\[b\\] c](${url})\n`)
@@ -701,7 +697,8 @@ test('spells the directive marks around the longest run they cover', () => {
   const emitted = (...content: AdfNode[]): string => markdown(adfToMarkdown(document(paragraph(...content))))
   const underline: AdfMark = { type: 'underline' }
   assert.equal(emitted(marked('x', underline)), '!adf:underline[x]\n')
-  assert.equal(emitted(marked('a', underline), marked('b', underline)), '!adf:underline[ab]\n')
+  assert.equal(emitted(marked('a', underline), marked('b', underline)), '!adf:underline[a!adf:textBreak{}b]\n')
+  assert.equal(emitted(marked('a', { attrs: {}, type: 'underline' }), marked('b', underline)), '!adf:underline[a]{attrs=empty}!adf:underline[b]\n')
   assert.equal(emitted(marked('x', { attrs: { type: 'sub' }, type: 'subsup' })), '!adf:subsup[x]{type=sub}\n')
   assert.equal(emitted(marked('x', { attrs: { color: '#ae2e24' }, type: 'textColor' })), '!adf:textColor[x]{color="#ae2e24"}\n')
   assert.equal(emitted(marked('x', { attrs: { color: '#091e42', size: 2 }, type: 'border' })), '!adf:border[x]{color="#091e42" size=2}\n')
@@ -746,5 +743,5 @@ test('carries whitespace CommonMark strips in the reserved text directive', () =
 
 test("joins a mark run's segments as a walk rather than as one call's arguments", () => {
   const run = Array.from({ length: 200000 }, (): AdfNode => ({ marks: [{ type: 'strong' }], text: 'a', type: 'text' }))
-  assert.equal(markdown(adfToMarkdown(document({ content: run, type: 'paragraph' }))), `**${'a'.repeat(200000)}**\n`)
+  assert.equal(markdown(adfToMarkdown(document({ content: run, type: 'paragraph' }))), `**${Array.from({ length: 200000 }, () => 'a').join('!adf:textBreak{}')}**\n`)
 })

src/markdown/emit/adf-to-markdown.ts

@@ -1,16 +1,16 @@
 import type { AdfDocument, AdfNode } from '../../adf/document.ts'
 import type { BlockNodeModel } from '../../adf/block-nodes.ts'
 import type { Flavour } from '../plain-conventions.ts'
-import { adfDocumentFault, carriesOnly, nodeAttrs, nodeContent, nodeMarks } from '../../adf/document.ts'
+import { adfDocumentFault, carriesOnly, isPlainText, nodeAttrs, nodeContent } from '../../adf/document.ts'
 import { alertMarker, foldedAlertMarker, leadingMarker, readAlertMarker, readTaskMarker, taskMarker } from '../plain-conventions.ts'
-import { blockDirectiveForm, listBreakSpelling } from '../block-directive.ts'
+import { blockDirectiveForm, documentSpelling, listBreakSpelling } from '../block-directive.ts'
 import { blockNodeModel, blockNodes } from '../../adf/block-nodes.ts'
 import { carriedBlock } from '../opaque-carry.ts'
 import { emitInlineLine } from './inline-line.ts'
 import { failure, faulted, success, type ConvertErrorPath, type Result } from '../../result.ts'
 import { fencedCodeBlock } from '../commonmark/backtick-runs.ts'
 import { holdsNullCharacter, isBlankLine, isThematicBreak, markerInterruptsParagraph } from '../commonmark/grammar.ts'
-import { languageSlot } from '../code-language.ts'
+import { languageSlot, type LanguageSlot } from '../code-language.ts'
 import { largestNesting } from '../../nesting.ts'
 import { spellBlockDirectiveOpener } from './block-directive-spelling.ts'
 import { spellDirectiveCloser, spellDirectiveOpener } from '../directive-syntax.ts'
@@ -40,7 +40,8 @@ export function writeMarkdown(document: AdfDocument, flavour: Flavour): Result<s
   const fault = adfDocumentFault(document)
   if (fault !== undefined) return faulted(fault, [])
   if (document.version !== 1) return failure('unsupported-document-version', `no markdown spelling carries ADF version ${document.version}`, [])
-  const walk = walkBlocks(nodeContent(document), [], 0, { flavour, memo: undefined })
+  if (document.content === undefined) return success(`${documentSpelling}\n`)
+  const walk = walkBlocks(document.content, [], 0, { flavour, memo: undefined })
   if (!walk.ok) return walk
   const text = joinBlocks(walk.value.blocks, 'document')
   return success(text === '' ? '' : `${text}\n`)
@@ -254,40 +255,39 @@ function tryCodeBlock(node: AdfNode, path: ConvertErrorPath): Result<EmittedBloc
   if (!carriesOnly(node, ['language'])) return undefined
   const slot = languageSlot(nodeAttrs(node)['language'])
   if (slot.kind === 'attribute') return undefined
-  const text = codeBlockText(node, path)
+  const texts = fencedTexts(node, path)
-  if (!text.ok) return text
+  if (!texts.ok) return texts
-  return success(commonMarkText(fencedCodeBlock(slot.kind === 'fence' ? slot.info : '', text.value)))
+  const [only, ...others] = texts.value ?? []
+  if (only === undefined || others.length > 0) return undefined
+  return success(commonMarkText(fencedCodeBlock(slot.kind === 'fence' ? slot.info : '', only)))
 }
 
+// spec/flavour.md, The CommonMark blocks: one fence per text node, the language on each; with no fence to carry it, the attribute does.
 function emitCodeDirective(node: AdfNode, model: BlockNodeModel, path: ConvertErrorPath, depth: number): Result<EmittedBlock> {
-  const slot = languageSlot(nodeAttrs(node)['language'])
+  const texts = fencedTexts(node, path)
+  if (!texts.ok) return texts
+  if (texts.value === undefined) return commonMarkLine(carriedBlock(node, path, depth))
+  const slot: LanguageSlot = texts.value.length === 0 ? { kind: 'attribute' } : languageSlot(nodeAttrs(node)['language'])
   const opener = spellBlockDirectiveOpener(node, model, path, slot.kind === 'attribute' ? [] : ['language'])
   if (opener === undefined) return commonMarkLine(carriedBlock(node, path, depth))
   if (!opener.ok) return opener
-  const text = codeBlockText(node, path)
+  const info = slot.kind === 'fence' ? slot.info : ''
-  if (!text.ok) return text
+  return success(directivePair(node, opener.value, texts.value.map((text) => fencedCodeBlock(info, text)).join('\n')))
-  return success(directivePair(node, opener.value, fencedCodeBlock(slot.kind === 'fence' ? slot.info : '', text.value)))
 }
 
-function codeBlockText(node: AdfNode, path: ConvertErrorPath): Result<string> {
+// The text each fence holds, or `undefined` where a child is no plain text node, which the carry holds instead.
-  let text = ''
+function fencedTexts(node: AdfNode, path: ConvertErrorPath): Result<string[] | undefined> {
-  for (const [index, child] of nodeContent(node).entries()) {
+  if (node.content === undefined) return success([''])
+  if (node.content.some((child) => !isPlainText(child) || child.marks !== undefined)) return success(undefined)
+  const texts: string[] = []
+  for (const [index, child] of node.content.entries()) {
     const childPath = [...path, 'content', index]
-    if (
+    if (typeof child.text !== 'string' || child.text === '') return failure('unsupported-node-shape', 'a text node holds text: this one has none', childPath)
-      child.type !== 'text' ||
-      typeof child.text !== 'string' ||
-      child.text === '' ||
-      nodeContent(child).length > 0 ||
-      nodeMarks(child).length > 0 ||
-      Object.keys(nodeAttrs(child)).length > 0
-    ) {
-      return failure('unsupported-node-shape', `a codeBlock holds plain text nodes only: this ${child.type} node is not one`, childPath)
-    }
     if (/\r/.test(child.text)) return failure('unspellable-character', 'a codeBlock holds no carriage return CommonMark keeps: this text holds one', childPath)
     if (holdsNullCharacter(child.text)) return failure('unspellable-character', 'a codeBlock holds a null character CommonMark replaces', childPath)
-    text += child.text
+    texts.push(child.text)
   }
-  return success(text)
+  return success(texts)
 }
 
 function tryHeading(node: AdfNode, path: ConvertErrorPath, flavour: Flavour): Result<EmittedBlock> | undefined {
@@ -340,7 +340,7 @@ function directiveItems(items: readonly WalkedItem[]): PlacedBlock[] {
 function listStart(node: AdfNode, items: number): number | undefined {
   if (node.type !== 'orderedList') return 0
   const start = nodeAttrs(node)['order']
-  if (typeof start !== 'number' || !Number.isInteger(start) || start < 0 || start > largestListMarker) return undefined
+  if (typeof start !== 'number' || !Number.isInteger(start) || start < 0 || Object.is(start, -0) || start > largestListMarker) return undefined
   return start + items - 1 > largestListMarker ? undefined : start
 }
 

src/markdown/emit/block-directive-spelling.ts

@@ -5,6 +5,7 @@ import { blockArgument, markValues, marksAttribute } from '../block-directive.ts
 import { failure, success, type ConvertErrorPath, type Result } from '../../result.ts'
 import { isBareToken, spellAttributes, spellDirectiveOpener, spellJsonAttribute, spellVocabulary } from '../directive-syntax.ts'
 import { overNested } from '../../json-value.ts'
+import { spellEmptyKeys } from '../empty-keys.ts'
 import { vocabularyPairs } from '../../adf/attribute-vocabulary.ts'
 
 export function spellBlockDirectiveOpener(node: AdfNode, model: BlockNodeModel, path: ConvertErrorPath, spelledByBody: readonly string[] = []): Result<string> | undefined {
@@ -14,7 +15,7 @@ export function spellBlockDirectiveOpener(node: AdfNode, model: BlockNodeModel,
   const spelled = argumentAttribute === undefined ? spelledByBody : [argumentAttribute, ...spelledByBody]
   const pairs = vocabularyPairs(nodeAttrs(node), model.attributes, spelled)
   if (pairs === undefined) return undefined
-  const spelledPairs = spellVocabulary(pairs)
+  const spelledPairs = [...spellVocabulary(pairs), ...spellEmptyKeys(node)]
   const marks = nodeMarks(node)
   if (marks.length > 0) {
     const values = markValues(marks)

src/markdown/emit/inline-directive-spelling.ts

@@ -2,9 +2,10 @@ import type { AdfNode } from '../../adf/document.ts'
 import type { InlineNodeModel } from '../../adf/inline-nodes.ts'
 import { nodeAttrs } from '../../adf/document.ts'
 import { spellAttributes, spellVocabulary } from '../directive-syntax.ts'
+import { spellEmptyKeys } from '../empty-keys.ts'
 import { vocabularyPairs } from '../../adf/attribute-vocabulary.ts'
 
 export function spellInlineNodeAttributes(node: AdfNode, model: InlineNodeModel): string | undefined {
   const pairs = vocabularyPairs(nodeAttrs(node), model.attributes, model.textAttribute === undefined ? [] : [model.textAttribute])
-  return pairs === undefined ? undefined : spellAttributes(spellVocabulary(pairs))
+  return pairs === undefined ? undefined : spellAttributes([...spellVocabulary(pairs), ...spellEmptyKeys(node)])
 }

src/markdown/emit/inline-line.ts

@@ -6,14 +6,14 @@ import { assembleInlineLine, isSyntax, type InlineEscaping, type InlineSegment,
 import { carriedInline } from '../opaque-carry.ts'
 import { claimsLine, holdsNullCharacter, trimTrailingSpace } from '../commonmark/grammar.ts'
 import { commonMarkLink, linkHref, markSpelling, spellMarkAttributes } from '../mark-spellings.ts'
+import { identicalMark, isPlainText, nodeAttrs, nodeContent, nodeMarks } from '../../adf/document.ts'
 import { escapeUnbalanced, spellDestination } from '../commonmark/link-syntax.ts'
 import { failure, faulted, success, type ConvertErrorPath, type Result } from '../../result.ts'
 import { highlightDelimiter } from '../plain-conventions.ts'
 import { inlineNodeModel } from '../../adf/inline-nodes.ts'
 import { largestNesting } from '../../nesting.ts'
 import { longestBacktickRun } from '../commonmark/backtick-runs.ts'
-import { nodeAttrs, nodeContent, nodeMarks } from '../../adf/document.ts'
+import { readsAsOne, textBreakSpelling } from '../text-break.ts'
-import { sameMark } from '../../adf/editor-normal.ts'
 import { slotLineEndingFault, spellInlineDirectiveOpener, spellInlineLeafDirective } from '../directive-syntax.ts'
 import { spellInlineNodeAttributes } from './inline-directive-spelling.ts'
 import { spellTextDirective } from '../text-directive.ts'
@@ -186,6 +186,7 @@ function emitRun(nodes: readonly AdfNode[], depth: number, firstIndex: number, c
   const runs = inlineRuns(nodes, depth, firstIndex, context.carried)
   const segments: InlineSegment[] = []
   for (const [offset, run] of runs.entries()) {
+    if (partsText(runs[offset - 1], run, context.carried)) segments.push(syntax(textBreakSpelling))
     const runContext = { ...context, atBlockEnd: context.atBlockEnd && offset === runs.length - 1 }
     const emitted = run.kind === 'plain' ? emitLeaf(run.node, runContext, run.index) : emitMarkedRun(run.nodes, run.mark, depth, run.index, runContext)
     if (!emitted.ok) return emitted
@@ -206,12 +207,18 @@ function inlineRuns(nodes: readonly AdfNode[], depth: number, firstIndex: number
       continue
     }
     const previous = runs[runs.length - 1]
-    if (previous?.kind === 'marked' && sameMark(previous.mark, mark)) previous.nodes.push(node)
+    if (previous?.kind === 'marked' && identicalMark(previous.mark, mark)) previous.nodes.push(node)
     else runs.push({ index, kind: 'marked', mark, nodes: [node] })
   }
   return runs
 }
 
+function partsText(previous: InlineRun | undefined, run: InlineRun, carried: ReadonlySet<number>): boolean {
+  if (previous?.kind !== 'plain' || run.kind !== 'plain') return false
+  if (carries(previous.node, carried, previous.index) || carries(run.node, carried, run.index)) return false
+  return readsAsOne(previous.node, run.node)
+}
+
 function nodePath(context: InlineContext, index: number): ConvertErrorPath {
   return [...context.path, 'content', index]
 }
@@ -261,9 +268,9 @@ function emitInlineDirective(node: AdfNode, model: InlineNodeModel, index: numbe
 }
 
 function emitText(node: AdfNode, context: InlineContext, index: number, path: ConvertErrorPath): Result<Emission> {
-  if (Object.keys(nodeAttrs(node)).length > 0) return success({ carry: { first: index, last: index } })
-  if (typeof node.text !== 'string' || node.text === '') return failure('unsupported-node-shape', 'a text node holds text: this one has none', path)
   if (nodeContent(node).length > 0) return failure('unsupported-node-shape', 'a text node holds no content: this one holds some', path)
+  if (!isPlainText(node)) return success({ carry: { first: index, last: index } })
+  if (typeof node.text !== 'string' || node.text === '') return failure('unsupported-node-shape', 'a text node holds text: this one has none', path)
   if (/\r/.test(node.text)) return failure('unspellable-character', 'a text node holds a carriage return CommonMark rewrites', path)
   if (holdsNullCharacter(node.text)) return failure('unspellable-character', 'a text node holds a null character CommonMark replaces', path)
   const escaping: InlineEscaping = context.bracketed ? 'bracketed' : 'backslash'
@@ -277,7 +284,7 @@ function emitMarkedRun(nodes: readonly AdfNode[], mark: AdfMark, depth: number,
   if (mark.type === 'backgroundColor' && context.flavour === 'plain') return emitHighlight(nodes, depth, range, context)
   const spelling = markSpelling(mark.type)
   if (spelling === undefined) return success({ carry: range })
-  const attributes = spellMarkAttributes(mark, spelling.attributes)
+  const attributes = spellMarkAttributes(mark, spelling)
   if (attributes === undefined) return success({ carry: range })
   if (spelling.kind === 'code') return emitCodeSpan(nodes, depth, range, path)
   if (spelling.kind === 'emphasis') return emitEmphasis(nodes, spelling.spelling, depth, range, context)
@@ -312,19 +319,20 @@ function emitHighlight(nodes: readonly AdfNode[], depth: number, range: NodeRang
   })
 }
 
+// Each node its own span: CommonMark reads two adjacent text nodes in one span back as one.
 function emitCodeSpan(nodes: readonly AdfNode[], depth: number, range: NodeRange, path: ConvertErrorPath): Result<Emission> {
-  let text = ''
+  const spans: string[] = []
   for (const node of nodes) {
-    if (node.type !== 'text' || nodeMarks(node).length !== depth + 1) return success({ carry: range })
+    if (node.type === 'text' && nodeContent(node).length > 0) return failure('unsupported-node-shape', 'a text node holds no content: this one holds some', path)
-    if (typeof node.text !== 'string' || node.text === '') return failure('unsupported-node-shape', 'a text node holds text: this one has none', path)
+    if (!isPlainText(node) || nodeMarks(node).length !== depth + 1) return success({ carry: range })
-    if (nodeContent(node).length > 0) return failure('unsupported-node-shape', 'a text node holds no content: this one holds some', path)
+    const { text } = node
-    text += node.text
+    if (typeof text !== 'string' || text === '') return failure('unsupported-node-shape', 'a text node holds text: this one has none', path)
+    if (/[\n\r]/.test(text)) return success({ carry: range })
+    if (holdsNullCharacter(text)) return failure('unspellable-character', 'a code span holds a null character CommonMark replaces', path)
+    const fence = '`'.repeat(longestBacktickRun(text) + 1)
+    spans.push(`${fence}${needsPadding(text) ? ` ${text} ` : text}${fence}`)
   }
-  if (/[\n\r]/.test(text)) return success({ carry: range })
+  return success({ segments: [syntax(spans.join(textBreakSpelling))] })
-  if (holdsNullCharacter(text)) return failure('unspellable-character', 'a code span holds a null character CommonMark replaces', path)
-  const fence = '`'.repeat(longestBacktickRun(text) + 1)
-  const padded = needsPadding(text) ? ` ${text} ` : text
-  return success({ segments: [syntax(`${fence}${padded}${fence}`)] })
 }
 
 function needsPadding(text: string): boolean {

src/markdown/emit/plain-inline.ts

@@ -3,9 +3,9 @@ import type { LineContainer } from '../line-container.ts'
 import type { MarkRun } from './line-escaping.ts'
 import { blockNodeModel } from '../../adf/block-nodes.ts'
 import { failure, success, type ConvertErrorPath, type Result } from '../../result.ts'
+import { joinsNormally, sameMark } from '../../adf/editor-normal.ts'
 import { largestNesting } from '../../nesting.ts'
-import { mergeAdjacentText, sameMark } from '../../adf/editor-normal.ts'
+import { mergeAdjacentText, nodeAttrs, nodeContent, nodeMarks } from '../../adf/document.ts'
-import { nodeAttrs, nodeContent, nodeMarks } from '../../adf/document.ts'
 import { plainLineFallback, type PlainLineFallback } from './inline-line.ts'
 import { spellDestination, spellLinkTarget } from '../commonmark/link-syntax.ts'
 
@@ -193,7 +193,7 @@ function withMarks(leaf: AdfNode, marks: readonly AdfMark[]): AdfNode {
 
 function trimmedEdges(leaves: readonly AdfNode[]): AdfNode[] {
   for (let current = leaves; ; ) {
-    const merged = withoutEdgeBreaks(mergeAdjacentText(current))
+    const merged = withoutEdgeBreaks(mergeAdjacentText(current, joinsNormally))
     let changed = false
     const trimmed: AdfNode[] = []
     for (const [index, leaf] of merged.entries()) {

src/markdown/emit/plain-reduction.test.ts

@@ -64,6 +64,7 @@ test('refuses what the document guard refuses, and nothing else', () => {
   let deep: AdfNode = said('x')
   for (let level = 0; level <= largestNesting; level += 1) deep = { content: [deep], type: 'layoutColumn' }
   assert.match(plainDocument(document(deep)), /^unsupported-nesting-depth at \/content\/0(\/content\/0)+$/)
+  assert.match(plainDocument(document(paragraph(text('a'), text('b')), text('c'), text('d'), deep)), /^unsupported-nesting-depth at \/content\/3(\/content\/0)+$/)
   let deepInline: AdfNode = text('x')
   for (let level = 0; level <= largestNesting; level += 1) deepInline = { content: [deepInline], type: 'unknownInline' }
   assert.match(plainDocument(document(paragraph(deepInline))), /^unsupported-nesting-depth at /)
@@ -177,7 +178,7 @@ test('keeps the CommonMark blocks in their spelling and drops their attributes a
   assert.equal(plain(node('paragraph', localId, text('x')), node('heading', { level: 2, localId: '01a0d99b-1f57-7fec-94ae-50c2ee25c9de' }, text('h'))), 'x\n\n## h\n')
   assert.equal(plain({ attrs: localId, content: [said('q')], marks: [{ type: 'breakout' }], type: 'blockquote' }), '> q\n')
   assert.equal(plain(node('codeBlock', { language: 'ts', wrap: true }, text('a\r\nb\u0000'))), '```ts\na\nb\n```\n')
-  assert.equal(plain(node('codeBlock', { language: 'carry' }, text('a'), { type: 'hardBreak' }, text('b', strong))), '```\na\nb\n```\n')
+  assert.equal(plain(node('codeBlock', { language: 'adf:x' }, text('a'), { type: 'hardBreak' }, text('b', strong))), '```\na\nb\n```\n')
   assert.equal(plain(node('codeBlock', {})), '```\n```\n')
   assert.equal(plain(node('rule', { color: '#000' })), '---\n')
   assert.equal(plain(node('orderedList', { localId: '01a0d99b-1f58-7b95-829b-6f9860371d54', order: 3 }, item(said('c')))), '3. c\n')
@@ -337,3 +338,9 @@ test('keeps the nodes the plain flavour spells and degrades only what it cannot'
   const reduced = reduceToPlain(document(node('panel', { localId: '01a0d99b-1f56-7a50-889a-f4375f09ee05', panelType: 'info' }, said('x')), tasks))
   assert.deepEqual(reduced.ok ? reduced.value : undefined, document(node('panel', { panelType: 'info' }, said('x')), { content: [node('taskItem', { state: 'DONE' }, text('t'))], type: 'taskList' }))
 })
+
+test('writes what the editor-normal form of the document writes', () => {
+  assert.equal(plain({ attrs: { order: -0 }, content: [{ content: [], type: 'listItem' }], type: 'orderedList' }), '0.\n')
+  assert.equal(plain({ content: [], text: 'hi', type: 'futureInline' }), 'hi\n')
+  assert.equal(plain(paragraph({ marks: [{ attrs: {}, type: 'code' }], text: 'a', type: 'text' }, { marks: [{ type: 'code' }], text: '|b', type: 'text' })), '`a|b`\n')
+})

src/markdown/emit/plain-reduction.ts

@@ -8,6 +8,7 @@ import { inlineNodeModel } from '../../adf/inline-nodes.ts'
 import { languageSlot } from '../code-language.ts'
 import { largestNesting } from '../../nesting.ts'
 import { taskMarker } from '../plain-conventions.ts'
+import { toEditorNormal } from '../../adf/editor-normal.ts'
 
 // depth: the level the node reduced stands at, counted as the emitter counts it.
 type Reduction = { depth: number; memo: SpellingMemo; path: ConvertErrorPath }
@@ -50,8 +51,14 @@ export function reduceToPlain(document: AdfDocument): Result<AdfDocument> {
   const fault = adfDocumentFault(document)
   if (fault !== undefined) return faulted(fault, [])
   if (document.version !== 1) return failure('unsupported-document-version', `no markdown spelling carries ADF version ${document.version}`, [])
-  const blocks = reduceBlocks(nodeContent(document), { depth: 0, memo: new Map(), path: [] })
+  // docs/decisions.md §Equality is deep: the reduction reads and writes editor-normal ADF.
-  return blocks.ok ? success({ content: blocks.value, type: 'doc', version: 1 }) : blocks
+  const blocks = reduceBlocks(nodeContent(toEditorNormal(document)), { depth: 0, memo: new Map(), path: [] })
+  if (!blocks.ok) {
+    // Merging text renumbers siblings, so the caller's document names the path of the refusal the editor-normal one meets.
+    const raw = reduceBlocks(nodeContent(document), { depth: 0, memo: new Map(), path: [] })
+    return !raw.ok && raw.error.code === blocks.error.code ? raw : blocks
+  }
+  return success({ content: nodeContent(toEditorNormal({ content: blocks.value, type: 'doc', version: 1 })).slice(), type: 'doc', version: 1 })
 }
 
 function reduceBlocks(nodes: readonly AdfNode[], reduction: Reduction): Result<AdfNode[]> {
@@ -84,7 +91,7 @@ function reduceStanding(node: AdfNode, reduction: Reduction): Result<AdfNode[]>
 
 function standsInline(node: AdfNode): boolean {
   if (node.type === 'text' || inlineNodeModel(node.type) !== undefined) return true
-  return !isBlockNodeType(node.type) && node.content === undefined
+  return !isBlockNodeType(node.type) && nodeContent(node).length === 0
 }
 
 function reduceBody(node: AdfNode, reduction: Reduction): Result<AdfNode[]> {
@@ -262,7 +269,8 @@ function listItem(blocks: Result<AdfNode[]>): Result<AdfNode[]> {
 // A list item's first line reads as no rule and holds no line of spaces alone: the rule and the spaces give way.
 function itemOf(blocks: readonly AdfNode[]): AdfNode {
   const rules = blocks.findIndex((block) => block.type !== 'rule')
-  return { content: blankedCode(blocks.slice(rules === -1 ? blocks.length : rules)), type: 'listItem' }
+  const content = blankedCode(blocks.slice(rules === -1 ? blocks.length : rules))
+  return content.length === 0 ? { type: 'listItem' } : { content, type: 'listItem' }
 }
 
 function blankedCode(blocks: readonly AdfNode[]): AdfNode[] {

src/markdown/empty-keys.ts

@@ -0,0 +1,32 @@
+import type { DirectiveAttributes, DirectiveValue, Read } from './directive-syntax.ts'
+import type { EmptyKey } from '../adf/document.ts'
+import { emptyKeys } from '../adf/document.ts'
+import { unsupportedNodeShape } from './directive-syntax.ts'
+
+type EmptyKeysRead = { empty: ReadonlySet<EmptyKey>; rest: Map<string, DirectiveValue> }
+
+const emptyValue = 'empty'
+
+// spec/flavour.md, Attributes: no attribute value spells an empty object or array.
+export function spellEmptyKeys(held: Parameters<typeof emptyKeys>[0]): [string, string][] {
+  return emptyKeys(held).map((key) => [key, emptyValue])
+}
+
+export function spellsEmpty(value: DirectiveValue | undefined): boolean {
+  return value?.spelling === emptyValue
+}
+
+// `held` is whether the directive spells an attribute outside {attrs}: an argument or a content slot.
+export function readEmptyKeys(type: string, attributes: DirectiveAttributes, keys: readonly EmptyKey[], held: boolean): Read<EmptyKeysRead> {
+  const rest = new Map(attributes)
+  const empty = new Set<EmptyKey>()
+  for (const key of keys) {
+    const spelled = rest.get(key)
+    if (spelled === undefined) continue
+    if (!spellsEmpty(spelled)) return { fault: unsupportedNodeShape(`the reserved key ${key} reads ${key}=${emptyValue} alone: this one spells ${key}=${spelled.spelling}`) }
+    empty.add(key)
+    rest.delete(key)
+  }
+  if (empty.has('attrs') && (held || [...rest.keys()].some((key) => key !== 'marks'))) return { fault: unsupportedNodeShape(`${type} spells attrs=empty beside an attribute it holds`) }
+  return { value: { empty, rest } }
+}

src/markdown/mark-spellings.ts

@@ -7,6 +7,7 @@ import { holdsEntityReference } from './commonmark/entity-references.ts'
 import { isAutolink } from './commonmark/grammar.ts'
 import { isMarkType, markAttributes } from '../adf/mark-attributes.ts'
 import { nodeAttrs, nodeMarks } from '../adf/document.ts'
+import { spellEmptyKeys } from './empty-keys.ts'
 import { vocabularyPairs } from '../adf/attribute-vocabulary.ts'
 
 type Spelling = { kind: 'code' | 'directive' | 'link'; spelling?: undefined } | { kind: 'emphasis'; spelling: string }
@@ -55,7 +56,10 @@ function isBareLink(nodes: readonly AdfNode[], href: string, marksInside: number
   return nodes.length === 1 && node !== undefined && node.type === 'text' && node.text === href && nodeMarks(node).length === marksInside
 }
 
-export function spellMarkAttributes(mark: AdfMark, vocabulary: AttributeVocabulary): string | undefined {
+// `undefined` where the spelling holds no such attrs: only a directive spells attrs=empty.
-  const pairs = vocabularyPairs(nodeAttrs(mark), vocabulary, [])
+export function spellMarkAttributes(mark: AdfMark, spelling: MarkSpelling): string | undefined {
-  return pairs === undefined ? undefined : spellAttributes(spellVocabulary(pairs))
+  const pairs = vocabularyPairs(nodeAttrs(mark), spelling.attributes, [])
+  const empty = spellEmptyKeys(mark)
+  if (pairs === undefined || (empty.length > 0 && spelling.kind !== 'directive')) return undefined
+  return spellAttributes([...spellVocabulary(pairs), ...empty])
 }

src/markdown/opaque-carry.ts

@@ -1,32 +1,49 @@
 import type { AdfNode } from '../adf/document.ts'
 import type { DirectiveSpan, Read } from './directive-syntax.ts'
 import type { JsonSpelling } from '../canonical-json.ts'
+import type { JsonValue } from '../json-value.ts'
 import { failure, success, type ConvertErrorPath, type Result } from '../result.ts'
 import { fencedCodeBlock } from './commonmark/backtick-runs.ts'
+import { infoStringCarries } from './commonmark/grammar.ts'
 import { isAdfNode } from '../adf/document.ts'
 import { isJsonValue, nestingDepth, overNested } from '../json-value.ts'
 import { largestNesting } from '../nesting.ts'
-import { malformedDirective, readSoleStringAttribute, spellAttributes, spellInlineLeafDirective, spellStringAttribute, unsupportedNodeShape } from './directive-syntax.ts'
+import { directiveEscape, malformedDirective, readSoleStringAttribute, spellAttributes, spellDirectiveOpener, spellInlineLeafDirective, spellStringAttribute, unsupportedNodeShape } from './directive-syntax.ts'
 import { serializeCanonicalJson } from '../canonical-json.ts'
 
+export const carryFencePrefix = 'adf:'
+
 export const carryName = 'carry'
 
 const jsonAttribute = 'json'
 
+// spec/flavour.md, The opaque carry: the info string names the type wherever it carries the type back.
 export function carriedBlock(node: AdfNode, path: ConvertErrorPath, depth: number): Result<{ headroom: number; text: string }> {
-  const json = carriedJson(node, 'two-space', path, largestNesting - depth)
+  const { type, ...untyped } = node
+  const named = infoStringCarries(type)
+  const json = carriedJson(node, named ? untyped : node, 'two-space', path, largestNesting - depth)
   if (!json.ok) return json
-  return success({ headroom: json.value.headroom, text: fencedCodeBlock(carryName, json.value.json) })
+  return success({ headroom: json.value.headroom, text: fencedCodeBlock(`${carryFencePrefix}${named ? type : ''}`, json.value.json) })
 }
 
 export function carriedInline(node: AdfNode, path: ConvertErrorPath): Result<string> {
-  const json = carriedJson(node, 'compact', path, largestNesting)
+  const json = carriedJson(node, node, 'compact', path, largestNesting)
   if (!json.ok) return json
   return success(spellInlineLeafDirective(carryName, spellAttributes([[jsonAttribute, spellStringAttribute(json.value.json)]])))
 }
 
-export function readCarriedBlock(body: string, depth: number): Read<AdfNode> {
+// The type a fence's info string names, `undefined` where the info string opens no carry.
-  return readCarriedJson(body, 'two-space', largestNesting - depth)
+export function carryFenceType(info: string): string | undefined {
+  return info.startsWith(carryFencePrefix) ? info.slice(carryFencePrefix.length) : undefined
+}
+
+export function readCarriedBlock(type: string, body: string, depth: number): Read<AdfNode> {
+  if (type !== '' && !infoStringCarries(type)) {
+    return { fault: unsupportedNodeShape(`the carry fence names a type no info string carries back: spell it ${carryFencePrefix} with the type in the JSON`) }
+  }
+  const read = readCarriedJson(body, 'two-space', largestNesting - depth, type)
+  if (read.fault !== undefined || type !== '' || !infoStringCarries(read.value.type)) return read
+  return { fault: unsupportedNodeShape(`the carry fence names a type its info string carries: spell it ${carryFencePrefix}${read.value.type}`) }
 }
 
 export function readCarriedInline(span: DirectiveSpan): Read<AdfNode> | undefined {
@@ -36,28 +53,44 @@ export function readCarriedInline(span: DirectiveSpan): Read<AdfNode> | undefine
   return readCarriedJson(spelled.value, 'compact', largestNesting)
 }
 
-function carriedJson(node: AdfNode, spelling: JsonSpelling, path: ConvertErrorPath, levels: number): Result<{ headroom: number; json: string }> {
+function carriedJson(node: AdfNode, spelled: object, spelling: JsonSpelling, path: ConvertErrorPath, levels: number): Result<{ headroom: number; json: string }> {
   const headroom = levels - nestingDepth(node)
-  if (!isJsonValue(node) || headroom < 0) {
+  if (!isJsonValue(spelled) || headroom < 0) {
     return failure('unsupported-nesting-depth', `a carried node's JSON nests deeper than the ${levels} levels its position leaves`, path)
   }
-  return success({ headroom, json: serializeCanonicalJson(node, spelling) })
+  return success({ headroom, json: serializeCanonicalJson(spelled, spelling) })
 }
 
-function readCarriedJson(raw: string, spelling: JsonSpelling, levels: number): Read<AdfNode> {
+// `type` is the one the fence's info string names, `undefined` for the inline carry.
+function readCarriedJson(raw: string, spelling: JsonSpelling, levels: number, type?: string): Read<AdfNode> {
+  const wayOut = type === undefined ? directiveEscape : fenceEscape(type)
   const parsed = parseJsonText(raw)
-  if (parsed === undefined) return { fault: malformedDirective('the opaque carry holds invalid JSON') }
+  if (parsed === undefined) return { fault: malformedDirective(`the opaque carry holds invalid JSON; ${wayOut}`) }
   const { value } = parsed
   if (!isJsonValue(value)) return { fault: unsupportedNodeShape('the opaque carry holds a number JSON cannot spell') }
-  if (overNested(value, levels)) {
+  const typed = type === undefined || type === '' ? { value } : typedValue(value, type)
+  if (typed.fault !== undefined) return typed
+  const held = typed.value
+  if (overNested(held, levels)) {
     return { fault: { code: 'unsupported-nesting-depth', message: `a carried node's JSON nests deeper than the ${levels} levels its position leaves` } }
   }
   if (serializeCanonicalJson(value, spelling) !== raw) {
     const shape = spelling === 'compact' ? 'compact, keys sorted' : 'two-space indent, keys sorted'
-    return { fault: unsupportedNodeShape(`the opaque carry spells its node's JSON canonically: ${shape}`) }
+    return { fault: unsupportedNodeShape(`the opaque carry spells its node's JSON canonically: ${shape}; ${wayOut}`) }
   }
-  if (!isAdfNode(value)) return { fault: unsupportedNodeShape("the opaque carry holds one ADF node's JSON: this JSON is no ADF node") }
+  if (!isAdfNode(held)) return { fault: unsupportedNodeShape(`the opaque carry holds one ADF node's JSON: this JSON is no ADF node; ${wayOut}`) }
-  return { value }
+  return { value: held }
+}
+
+// A code fence the carry claims stays code under the directive, whose language rides the attribute.
+function fenceEscape(type: string): string {
+  return `${spellDirectiveOpener('codeBlock', undefined, spellAttributes([['language', spellStringAttribute(`${carryFencePrefix}${type}`)]]))} around a bare fence keeps it a code block`
+}
+
+function typedValue(value: JsonValue, type: string): Read<JsonValue> {
+  if (value === null || typeof value !== 'object' || Array.isArray(value)) return { value }
+  if ('type' in value) return { fault: unsupportedNodeShape(`the ${carryFencePrefix}${type} fence names its node's type: this JSON holds a type as well`) }
+  return { value: { ...value, type } }
 }
 
 function parseJsonText(raw: string): { value: unknown } | undefined {

src/markdown/parse/directive-marks.ts

@@ -3,8 +3,9 @@ import type { ConvertFault } from '../../result.ts'
 import type { DirectiveAttributes } from '../directive-syntax.ts'
 import type { MarkSpelling } from '../mark-spellings.ts'
 import { directivePrefix } from '../directive-syntax.ts'
-import { failure, success, type ConvertErrorPath, type Result } from '../../result.ts'
+import { failure, faulted, success, type ConvertErrorPath, type Result } from '../../result.ts'
 import { markSpelling } from '../mark-spellings.ts'
+import { readEmptyKeys } from '../empty-keys.ts'
 import { readVocabulary } from './directive-attributes.ts'
 
 export function readDirectiveMark(name: string, attributes: DirectiveAttributes, path: ConvertErrorPath): Result<AdfMark> | undefined {
@@ -12,8 +13,11 @@ export function readDirectiveMark(name: string, attributes: DirectiveAttributes,
   if (spelling === undefined) return undefined
   const markdown = markdownForm(spelling)
   if (markdown !== undefined) return failure('unsupported-node-shape', `${name} is spelled ${markdown}, never as a directive`, path)
-  const attrs = readVocabulary(name, attributes, spelling.attributes, undefined, path)
+  const empty = readEmptyKeys(name, attributes, ['attrs'], false)
+  if (empty.fault !== undefined) return faulted(empty.fault, path)
+  const attrs = readVocabulary(name, empty.value.rest, spelling.attributes, undefined, path)
   if (!attrs.ok) return attrs
+  if (empty.value.empty.has('attrs')) return success({ attrs: {}, type: name })
   return success(Object.keys(attrs.value).length === 0 ? { type: name } : { attrs: attrs.value, type: name })
 }
 

src/markdown/parse/directive-nodes.ts

@@ -1,18 +1,20 @@
-import type { AdfAttributes, AdfMark, AdfNode } from '../../adf/document.ts'
+import type { AdfAttributes, AdfMark, AdfNode, EmptyKey } from '../../adf/document.ts'
 import type { BlockNodeModel } from '../../adf/block-nodes.ts'
 import type { ConvertFault } from '../../result.ts'
 import type { DirectiveAttributes, DirectiveValue } from '../directive-syntax.ts'
 import type { Elsewhere } from './directive-attributes.ts'
-import { attributeNestingMessage, nodeAttrs, nodeContent, nodeMarks } from '../../adf/document.ts'
+import { attributeNestingMessage, isPlainText } from '../../adf/document.ts'
 import { attributeValue, directivePrefix, spellAttributeValue, unknownDirectiveFault } from '../directive-syntax.ts'
 import { blockArgument, blockDirectiveForm, marksAttribute, readMarkValues } from '../block-directive.ts'
 import { blockNodeModel } from '../../adf/block-nodes.ts'
-import { carryName } from '../opaque-carry.ts'
+import { carryFencePrefix, carryName } from '../opaque-carry.ts'
 import { failure, faulted, success, type ConvertErrorPath, type Result } from '../../result.ts'
 import { inlineMarkSpellingFault } from './directive-marks.ts'
 import { inlineNodeModel } from '../../adf/inline-nodes.ts'
+import { readEmptyKeys, spellsEmpty } from '../empty-keys.ts'
 import { readVocabulary } from './directive-attributes.ts'
 import { slotLineEndingFault } from '../directive-syntax.ts'
+import { textBreakName } from '../text-break.ts'
 import { textDirectiveName } from '../text-directive.ts'
 
 export type BlockDirectiveNode = { contentModel: BlockNodeModel['contentModel']; node: AdfNode }
@@ -24,12 +26,15 @@ export function readBlockDirectiveNode(
   path: ConvertErrorPath,
 ): Result<BlockDirectiveNode> {
   if (name === carryName) {
-    return failure('malformed-directive', `the name ${carryName} is reserved for the opaque carry, whose block form is the ${carryName} fence`, path)
+    return failure('malformed-directive', `the name ${carryName} is reserved for the opaque carry, whose block form is the ${carryFencePrefix} fence`, path)
   }
   const model = blockNodeModel(name)
   if (model === undefined) return faulted(inlineSpellingFault(name) ?? unknownDirectiveFault(name), path)
   const argumentKey = blockArgument(name)
-  const rest = new Map(attributes)
+  const spelled = attributes.get(marksAttribute)
+  const empty = readEmptyKeys(name, attributes, spellsEmpty(spelled) ? ['attrs', 'content', 'marks'] : ['attrs', 'content'], argument !== undefined)
+  if (empty.fault !== undefined) return faulted(empty.fault, path)
+  const { rest } = empty.value
   rest.delete(marksAttribute)
   const elsewhere: Elsewhere | undefined = argumentKey === undefined ? undefined : { key: argumentKey, slot: 'argument' }
   const attrs = readVocabulary(name, rest, model.attributes, elsewhere, path)
@@ -38,10 +43,9 @@ export function readBlockDirectiveNode(
     if (argumentKey === undefined) return failure('unsupported-node-shape', `${name} takes no argument: this one spells one`, path)
     attrs.value[argumentKey] = argument
   }
-  const spelled = attributes.get(marksAttribute)
+  const marks: Result<AdfMark[] | undefined> = spelled === undefined || spellsEmpty(spelled) ? success(undefined) : readMarks(name, spelled, path)
-  const marks: Result<AdfMark[] | undefined> = spelled === undefined ? success(undefined) : readMarks(name, spelled, path)
   if (!marks.ok) return marks
-  return success({ contentModel: model.contentModel, node: namedNode(name, attrs.value, marks.value) })
+  return success({ contentModel: model.contentModel, node: namedNode(name, attrs.value, marks.value, empty.value.empty) })
 }
 
 export function readInlineDirectiveNode(
@@ -55,7 +59,9 @@ export function readInlineDirectiveNode(
   const slot = model.textAttribute
   if (slot === undefined && content !== undefined) return failure('unsupported-node-shape', `${name} takes no content: this one holds some`, path)
   const elsewhere: Elsewhere | undefined = slot === undefined ? undefined : { key: slot, slot: 'content' }
-  const attrs = readVocabulary(name, attributes, model.attributes, elsewhere, path)
+  const empty = readEmptyKeys(name, attributes, ['attrs', 'content', 'marks'], slot !== undefined && content !== undefined)
+  if (empty.fault !== undefined) return faulted(empty.fault, path)
+  const attrs = readVocabulary(name, empty.value.rest, model.attributes, elsewhere, path)
   if (!attrs.ok) return attrs
   if (slot !== undefined && content !== undefined) {
     const text = slotText(content)
@@ -66,14 +72,14 @@ export function readInlineDirectiveNode(
     if (spans !== undefined) return faulted(spans, path)
     attrs.value[slot] = text
   }
-  return success(namedNode(name, attrs.value, undefined))
+  return success(namedNode(name, attrs.value, undefined, empty.value.empty))
 }
 
 // A name the other position spells names that spelling, never the code a later MINOR may fill (docs/decisions.md §Which code a cause takes).
 function inlineSpellingFault(name: string): ConvertFault | undefined {
   const mark = inlineMarkSpellingFault(name)
   if (mark !== undefined) return mark
-  if (inlineNodeModel(name) === undefined && name !== textDirectiveName) return undefined
+  if (inlineNodeModel(name) === undefined && name !== textDirectiveName && name !== textBreakName) return undefined
   return { code: 'unsupported-node-shape', message: `${name} takes the inline form, ${directivePrefix}${name}{…}, never the block form` }
 }
 
@@ -86,7 +92,7 @@ function blockSpellingFault(name: string): ConvertFault | undefined {
 function slotText(content: readonly AdfNode[]): string | undefined {
   if (content.length === 0) return ''
   const only = content.length === 1 ? content[0] : undefined
-  if (only?.type !== 'text' || nodeMarks(only).length > 0 || Object.keys(nodeAttrs(only)).length > 0 || nodeContent(only).length > 0 || typeof only.text !== 'string') return undefined
+  if (only === undefined || !isPlainText(only) || only.marks !== undefined || typeof only.text !== 'string') return undefined
   return only.text
 }
 
@@ -100,7 +106,9 @@ function readMarks(type: string, spelled: DirectiveValue, path: ConvertErrorPath
   return success(marks)
 }
 
-function namedNode(type: string, attrs: AdfAttributes, marks: readonly AdfMark[] | undefined): AdfNode {
+function namedNode(type: string, attrs: AdfAttributes, marks: readonly AdfMark[] | undefined, empty: ReadonlySet<EmptyKey>): AdfNode {
-  const named = Object.keys(attrs).length === 0 ? { type } : { attrs, type }
+  const node: AdfNode = Object.keys(attrs).length > 0 || empty.has('attrs') ? { attrs, type } : { type }
-  return marks === undefined ? named : { ...named, marks: [...marks] }
+  if (empty.has('content')) node.content = []
+  if (marks !== undefined || empty.has('marks')) node.marks = [...(marks ?? [])]
+  return node
 }

src/markdown/parse/inline-content.ts

@@ -10,19 +10,29 @@ import { commonMarkLink, linkHref } from '../mark-spellings.ts'
 import { delimiterFlags, matchEmphasis, runLength } from '../commonmark/emphasis-matching.ts'
 import { failure, faulted, success, type ConvertErrorPath, type Result } from '../../result.ts'
 import { highlightDelimiter, highlightFlanking } from '../plain-conventions.ts'
+import { identicalMarks, mergeAdjacentText, nodeAttrs, nodeMarks } from '../../adf/document.ts'
 import { inlineNodeModel } from '../../adf/inline-nodes.ts'
-import { mergeAdjacentText, sameMarks } from '../../adf/editor-normal.ts'
 import { noSpans, readInlineDirective } from '../directive-syntax.ts'
-import { nodeAttrs, nodeMarks } from '../../adf/document.ts'
 import { normalizeLabel, readInlineTarget, readLabel } from '../commonmark/link-syntax.ts'
 import { openingLinkTakesDirective } from '../emit/inline-line.ts'
 import { readCarriedInline } from '../opaque-carry.ts'
 import { readDirectiveMark } from './directive-marks.ts'
 import { readInlineDirectiveNode } from './directive-nodes.ts'
 import { readTextDirective } from '../text-directive.ts'
+import { readsAsOne, textBreakName, textBreakSpelling } from '../text-break.ts'
 
 export type InlineContent = { carry?: undefined; image: AdfNode; nodes?: undefined } | { carry: boolean; image?: undefined; nodes: AdfNode[] }
 
+// The text break leaf: it holds no marks, and stands between the nodes it parts until the outermost content checks and drops it.
+type TextBreak = { kind: 'textBreak' }
+
+// A node whose marks are its own — an opaque carry, or an inline node spelling marks=empty: no mark spelling wraps it, and it joins no neighbour.
+type Carried = { kind: 'carry'; node: AdfNode }
+
+type Inline = AdfNode | Carried | TextBreak
+
+type Scanned = { carry?: undefined; image: AdfNode; nodes?: undefined } | { carry: boolean; image?: undefined; nodes: Inline[] }
+
 export type LinkDefinitions = ReadonlyMap<string, LinkDefinition>
 
 type Bracket = { active: boolean; image: boolean; kind: 'open'; start: number }
@@ -33,9 +43,10 @@ type Pairing = EmphasisPairing<Run>
 
 type Piece =
   | Bracket
-  | { kind: 'carry'; node: AdfNode }
+  | Carried
   | { alt: string; kind: 'image'; node: AdfNode }
-  | { kind: 'nodes'; nodes: AdfNode[] }
+  | { kind: 'nodes'; nodes: Inline[] }
+  | TextBreak
   | { canClose: boolean; canOpen: boolean; character: string; kind: 'run'; length: number }
   | { closes: boolean; kind: 'highlight'; opens: boolean }
 
@@ -56,21 +67,41 @@ type Scan = {
   spans: NestedSpans
 }
 
-type SlotContent = { carry: boolean; nodes: AdfNode[] }
+// What a directive's content slot inherits from the content holding it.
+type SharedScan = Pick<Scan, 'definitions' | 'path'>
 
-const carriedInMark = 'no mark spelling wraps an opaque carry: the carried node restores exactly, marks included'
+type SlotContent = { carry: boolean; nodes: Inline[] }
+
+const carriedInMark = 'no mark spelling wraps an opaque carry or an inline node spelling marks=empty: its marks are its own'
 const editorHighlight: AdfMark = { attrs: { color: '#f8e6a0' }, type: 'backgroundColor' }
 const hreflessLink = 'the link mark spells its href: this one spells none'
 const imageAlone = 'an image fits only as a paragraph of its own: this one sits inside other content'
 const linkInLink = 'no link wraps a link: the [content] this one marks already holds one'
 const spellableLink = 'link takes the directive form only where CommonMark cannot spell it: this one it can, as [text](url "title") or <url>'
+const textBreak: TextBreak = { kind: 'textBreak' }
 
+// The outermost content: only here does a text break see both its neighbours.
 export function parseInlineContent(source: string, definitions: LinkDefinitions, path: ConvertErrorPath, container: LineContainer, flavour: Flavour): Result<InlineContent> {
-  return parseInline(source, definitions, path, container, noSpans, flavour === 'plain')
+  const scan = freshScan(source, { definitions, path }, { container, highlights: flavour === 'plain', spans: noSpans })
+  const scanned = scanInline(scan)
+  if (!scanned.ok) return scanned
+  if (scanned.value.image !== undefined) return success({ image: scanned.value.image })
+  const nodes = partText(scanned.value.nodes, scan)
+  if (!nodes.ok) return nodes
+  if (scan.openingSpellableLink) {
+    const takesDirective = openingLinkTakesDirective(nodes.value, path)
+    if (!takesDirective.ok) return takesDirective
+    if (!takesDirective.value) return failure('unsupported-node-shape', spellableLink, path)
+  }
+  return success({ carry: scanned.value.carry, nodes: nodes.value })
 }
 
-function parseInline(source: string, definitions: LinkDefinitions, path: ConvertErrorPath, container: LineContainer | undefined, spans: NestedSpans, highlights: boolean): Result<InlineContent> {
+function freshScan(source: string, shared: SharedScan, own: Pick<Scan, 'container' | 'highlights' | 'spans'>): Scan {
-  const scan: Scan = { container, deactivatedBefore: 0, definitions, highlights, openingSpellableLink: false, path, pending: '', pieces: [], source, spans }
+  return { ...shared, ...own, deactivatedBefore: 0, openingSpellableLink: false, pending: '', pieces: [], source }
+}
+
+function scanInline(scan: Scan): Result<Scanned> {
+  const { source } = scan
   let index = 0
   while (index < source.length) {
     switch (source.charAt(index)) {
@@ -103,7 +134,7 @@ function parseInline(source: string, definitions: LinkDefinitions, path: Convert
         index = readCharacter(scan, index)
     }
   }
-  flush(scan, container !== undefined)
+  flush(scan, scan.container !== undefined)
   return assemble(scan)
 }
 
@@ -207,6 +238,7 @@ function directivePiece(scan: Scan, span: DirectiveSpan, index: number): Result<
     if (carried.fault !== undefined) return faulted(carried.fault, scan.path)
     return success({ kind: 'carry', node: carried.value })
   }
+  if (span.name === textBreakName) return span.content === undefined && span.attributes.size === 0 ? success(textBreak) : failure('unsupported-node-shape', `${textBreakName} spells the bare leaf form, ${textBreakSpelling}: this one spells more`, scan.path)
   const text = readTextDirective(span)
   if (text?.fault !== undefined) return faulted(text.fault, scan.path)
   if (text !== undefined) return success({ kind: 'nodes', nodes: [{ text: text.value, type: 'text' }] })
@@ -214,9 +246,11 @@ function directivePiece(scan: Scan, span: DirectiveSpan, index: number): Result<
   if (!slot.ok) return slot
   const mark = readDirectiveMark(span.name, span.attributes, scan.path)
   if (mark !== undefined) return mark.ok ? directiveMarkPiece(scan, span.name, mark.value, slot.value, index) : mark
-  const node = readInlineDirectiveNode(span.name, span.attributes, slot.value?.nodes, scan.path)
+  const content = slot.value?.nodes
+  if (content?.some(isTextBreak) === true) return failure('unsupported-node-shape', `${textBreakName} parts two text nodes: a content slot holds plain text`, scan.path)
+  const node = readInlineDirectiveNode(span.name, span.attributes, content === undefined ? undefined : adfNodes(content), scan.path)
   if (!node.ok) return node
-  return success({ kind: 'nodes', nodes: [node.value] })
+  return success(node.value.marks === undefined ? { kind: 'nodes', nodes: [node.value] } : { kind: 'carry', node: node.value })
 }
 
 function directiveMarkPiece(scan: Scan, name: string, mark: AdfMark, slot: SlotContent | undefined, index: number): Result<Piece> {
@@ -230,11 +264,11 @@ function directiveMarkPiece(scan: Scan, name: string, mark: AdfMark, slot: SlotC
 }
 
 // spec/flavour.md, Marks. A link opening a paragraph may still need the directive form for the line it opens, which `assemble` asks the emitter.
-function refuseLinkDirective(scan: Scan, mark: AdfMark, nodes: readonly AdfNode[], index: number): Result<Piece> | undefined {
+function refuseLinkDirective(scan: Scan, mark: AdfMark, nodes: readonly Inline[], index: number): Result<Piece> | undefined {
   const href = linkHref(nodeAttrs(mark))
   if (href === undefined) return failure('unsupported-node-shape', hreflessLink, scan.path)
   if (marksLink(nodes)) return failure('unsupported-node-shape', linkInLink, scan.path)
-  if (commonMarkLink(nodeAttrs(mark), href, nodes, 0, scan.container === undefined) === undefined) return undefined
+  if (commonMarkLink(nodeAttrs(mark), href, adfNodes(nodes), 0, scan.container === undefined) === undefined) return undefined
   if (index !== 0 || scan.container !== 'paragraph') return failure('unsupported-node-shape', spellableLink, scan.path)
   scan.openingSpellableLink = true
   return undefined
@@ -242,7 +276,8 @@ function refuseLinkDirective(scan: Scan, mark: AdfMark, nodes: readonly AdfNode[
 
 function slotContent(scan: Scan, span: DirectiveSpan): Result<SlotContent | undefined> {
   if (span.content === undefined) return success(undefined)
-  const parsed = parseInline(span.content, scan.definitions, scan.path, undefined, span.spans, false)
+  const { definitions, path } = scan
+  const parsed = scanInline(freshScan(span.content, { definitions, path }, { container: undefined, highlights: false, spans: span.spans }))
   if (!parsed.ok) return parsed
   if (parsed.value.image !== undefined) return failure('unmappable-image', imageAlone, scan.path)
   return success(parsed.value)
@@ -258,20 +293,50 @@ function pushNode(scan: Scan, node: AdfNode): void {
   scan.pieces.push({ kind: 'nodes', nodes: [node] })
 }
 
-function assemble(scan: Scan): Result<InlineContent> {
+function assemble(scan: Scan): Result<Scanned> {
   const only = scan.pieces[0]
   if (scan.pieces.length === 1 && only?.kind === 'image') return success({ image: only.node })
   if (holdsImage(scan.pieces)) return failure('unmappable-image', imageAlone, scan.path)
-  const nodes = resolveNodes(scan.pieces, scan.path, true)
+  const nodes = resolveNodes(scan.pieces, scan, true)
   if (!nodes.ok) return nodes
-  if (scan.openingSpellableLink) {
-    const takesDirective = openingLinkTakesDirective(nodes.value, scan.path)
-    if (!takesDirective.ok) return takesDirective
-    if (!takesDirective.value) return failure('unsupported-node-shape', spellableLink, scan.path)
-  }
   return success({ carry: holdsCarry(scan.pieces), nodes: nodes.value })
 }
 
+// spec/flavour.md, Inline nodes: the leaf builds no node, so only the pair it parts spells it.
+function partText(items: readonly Inline[], scan: Scan): Result<AdfNode[]> {
+  const parted: AdfNode[] = []
+  for (const [index, item] of items.entries()) {
+    if (!isTextBreak(item)) {
+      parted.push(isNode(item) ? item : item.node)
+      continue
+    }
+    const previous = items[index - 1]
+    const next = items[index + 1]
+    if (previous === undefined || next === undefined || !isNode(previous) || !isNode(next) || !readsAsOne(previous, next)) {
+      return failure('unsupported-node-shape', `${textBreakName} parts two text nodes CommonMark reads back as one: this one parts something else`, scan.path)
+    }
+  }
+  return success(parted)
+}
+
+function isNode(item: Inline): item is AdfNode {
+  return !('kind' in item)
+}
+
+function isTextBreak(item: Inline): item is TextBreak {
+  return 'kind' in item && item.kind === 'textBreak'
+}
+
+// The nodes the items hold, carried ones unwrapped and text breaks dropped.
+function adfNodes(items: readonly Inline[]): AdfNode[] {
+  const nodes: AdfNode[] = []
+  for (const item of items) {
+    if (isNode(item)) nodes.push(item)
+    else if (!isTextBreak(item)) nodes.push(item.node)
+  }
+  return nodes
+}
+
 function holdsCarry(pieces: readonly Piece[]): boolean {
   return pieces.some((piece) => piece.kind === 'carry')
 }
@@ -285,8 +350,8 @@ function holdsLink(pieces: readonly Piece[]): boolean {
   return pieces.some((piece) => piece.kind === 'nodes' && marksLink(piece.nodes))
 }
 
-function marksLink(nodes: readonly AdfNode[]): boolean {
+function marksLink(nodes: readonly Inline[]): boolean {
-  return nodes.some((node) => nodeMarks(node).some((mark) => mark.type === 'link'))
+  return adfNodes(nodes).some((node) => nodeMarks(node).some((mark) => mark.type === 'link'))
 }
 
 function readDelimiterRun(scan: Scan, index: number): number {
@@ -383,7 +448,7 @@ function closeLink(scan: Scan, at: number, inner: readonly Piece[], definition:
   }
   if (holdsImage(inner)) return failure('unmappable-image', imageAlone, scan.path)
   if (holdsCarry(inner)) return failure('unsupported-node-shape', carriedInMark, scan.path)
-  const resolved = resolveNodes(inner, scan.path, true)
+  const resolved = resolveNodes(inner, scan, true)
   if (!resolved.ok) return resolved
   const nodes = resolved.value
   // An empty link text gives the mark no node to ride, so the brackets stay text.
@@ -411,7 +476,7 @@ function truncatePieces(scan: Scan, to: number): void {
 
 function closeImage(scan: Scan, at: number, inner: readonly Piece[], definition: LinkDefinition): Result<null> {
   if (definition.title !== undefined) return failure('unmappable-image', 'no media node carries a link title', scan.path)
-  const resolved = imageAlt(inner, scan.path)
+  const resolved = imageAlt(inner, scan)
   if (!resolved.ok) return resolved
   const alt = resolved.value
   const attrs = alt === '' ? { type: 'external', url: definition.destination } : { alt, type: 'external', url: definition.destination }
@@ -420,10 +485,11 @@ function closeImage(scan: Scan, at: number, inner: readonly Piece[], definition:
   return success(null)
 }
 
-function imageAlt(inner: readonly Piece[], path: ConvertErrorPath): Result<string> {
+function imageAlt(inner: readonly Piece[], scan: Scan): Result<string> {
-  const nodes = resolveNodes(inner, path, false)
+  const nodes = resolveNodes(inner, scan, false)
   if (!nodes.ok) return nodes
-  return success(nodes.value.map(altText).join(''))
+  if (nodes.value.some(isTextBreak)) return failure('unsupported-node-shape', `${textBreakName} parts two text nodes: an image description holds plain text`, scan.path)
+  return success(adfNodes(nodes.value).map(altText).join(''))
 }
 
 // spec/flavour.md, The CommonMark image: the description's plain text, the content slot included.
@@ -435,22 +501,39 @@ function altText(node: AdfNode): string {
 }
 
 // An image's alt text is plain, so `highlights` is off there and every `==` stays text.
-function resolveNodes(pieces: readonly Piece[], path: ConvertErrorPath, highlights: boolean): Result<AdfNode[]> {
+function resolveNodes(pieces: readonly Piece[], scan: Scan, highlights: boolean): Result<Inline[]> {
   const nodes = pieces.map(pieceNodes)
   const runs = delimiterRuns(pieces)
   const pairings = matchEmphasis(runs)
   writeUnpaired(nodes, runs, pairings)
-  if (!markPairings(pieces, nodes, pairings)) return failure('unsupported-node-shape', carriedInMark, path)
+  if (!markPairings(pieces, nodes, pairings)) return failure('unsupported-node-shape', carriedInMark, scan.path)
   markHighlights(pieces, nodes, highlights ? pairedHighlights(pieces, nodes) : [])
-  return success(mergeAdjacentText(nodes.flat()))
+  return success(mergeText(nodes.flat()))
+}
+
+// A text break and a carried node are walls: the text on either side joins only its own side.
+function mergeText(items: readonly Inline[]): Inline[] {
+  const merged: Inline[] = []
+  let run: AdfNode[] = []
+  for (const item of items) {
+    if (isNode(item)) {
+      run.push(item)
+      continue
+    }
+    for (const node of mergeAdjacentText(run, readsAsOne)) merged.push(node)
+    merged.push(item)
+    run = []
+  }
+  for (const node of mergeAdjacentText(run, readsAsOne)) merged.push(node)
+  return merged
 }
 
 // Only `imageAlt` reaches the image arm: everywhere else an image amid other content is refused first.
 // A highlight delimiter holds an empty text node until it pairs, so the emphasis around it marks it.
-function pieceNodes(piece: Piece): AdfNode[] {
+function pieceNodes(piece: Piece): Inline[] {
   switch (piece.kind) {
     case 'carry':
-      return [piece.node]
+      return [piece]
     case 'highlight':
       return [{ text: '', type: 'text' }]
     case 'image':
@@ -461,6 +544,8 @@ function pieceNodes(piece: Piece): AdfNode[] {
       return bracketNodes(piece)
     case 'run':
       return []
+    case 'textBreak':
+      return [piece]
   }
 }
 
@@ -473,7 +558,7 @@ function delimiterRuns(pieces: readonly Piece[]): Run[] {
 }
 
 // A run gives its delimiters up from the head closing and the tail opening; what is left between them is text.
-function writeUnpaired(nodes: AdfNode[][], runs: readonly Run[], pairings: readonly Pairing[]): void {
+function writeUnpaired(nodes: Inline[][], runs: readonly Run[], pairings: readonly Pairing[]): void {
   const heads = new Map<Run, number>()
   const tails = new Map<Run, number>()
   for (const pairing of pairings) {
@@ -488,7 +573,7 @@ function writeUnpaired(nodes: AdfNode[][], runs: readonly Run[], pairings: reado
 }
 
 // Innermost pairing first, so prepending leaves the marks array outermost first (spec/flavour.md, Marks).
-function markPairings(pieces: readonly Piece[], nodes: AdfNode[][], pairings: readonly Pairing[]): boolean {
+function markPairings(pieces: readonly Piece[], nodes: Inline[][], pairings: readonly Pairing[]): boolean {
   for (const pairing of pairings) {
     const mark: AdfMark = { type: markType(pairing.opener.character, pairing.used) }
     for (let index = pairing.opener.index + 1; index < pairing.closer.index; index += 1) {
@@ -499,12 +584,12 @@ function markPairings(pieces: readonly Piece[], nodes: AdfNode[][], pairings: re
   return true
 }
 
-function highlightDelimiters(pieces: readonly Piece[], nodes: readonly AdfNode[][]): HighlightDelimiter[] {
+function highlightDelimiters(pieces: readonly Piece[], nodes: readonly Inline[][]): HighlightDelimiter[] {
   const found: HighlightDelimiter[] = []
   let line = 0
   let position = 0
   for (const [index, piece] of pieces.entries()) {
-    const held = nodes[index] ?? []
+    const held = adfNodes(nodes[index] ?? [])
     const [holder] = held
     if (piece.kind === 'highlight' && holder !== undefined) {
       found.push({ closes: piece.closes, holder, index, line, opens: piece.opens, position })
@@ -520,7 +605,7 @@ function highlightDelimiters(pieces: readonly Piece[], nodes: readonly AdfNode[]
 }
 
 // Each opener takes the next closer holding at least one character after it, both in one line and under the same marks.
-function pairedHighlights(pieces: readonly Piece[], nodes: readonly AdfNode[][]): { closer: number; opener: number }[] {
+function pairedHighlights(pieces: readonly Piece[], nodes: readonly Inline[][]): { closer: number; opener: number }[] {
   const found = highlightDelimiters(pieces, nodes)
   const paired: { closer: number; opener: number }[] = []
   let closer = 0
@@ -531,7 +616,7 @@ function pairedHighlights(pieces: readonly Piece[], nodes: readonly AdfNode[][])
     let candidate = found[closer]
     while (candidate !== undefined && (!candidate.closes || candidate.position < earliest)) candidate = found[(closer += 1)]
     if (candidate === undefined) break
-    if (candidate.line !== opener.line || !sameMarks(opener.holder, candidate.holder)) continue
+    if (candidate.line !== opener.line || !identicalMarks(nodeMarks(opener.holder), nodeMarks(candidate.holder))) continue
     paired.push({ closer: candidate.index, opener: opener.index })
     resume = candidate.position + highlightDelimiter.length
   }
@@ -539,9 +624,9 @@ function pairedHighlights(pieces: readonly Piece[], nodes: readonly AdfNode[][])
 }
 
 // Atlassian's schema refuses a highlight on code, a node holds one highlight, and a rebuilt node would lose its attributes.
-function markHighlights(pieces: readonly Piece[], nodes: AdfNode[][], paired: readonly { closer: number; opener: number }[]): void {
+function markHighlights(pieces: readonly Piece[], nodes: Inline[][], paired: readonly { closer: number; opener: number }[]): void {
   for (const [index, piece] of pieces.entries()) {
-    if (piece.kind === 'highlight') nodes[index] = (nodes[index] ?? []).map((holder) => ({ ...holder, text: highlightDelimiter }))
+    if (piece.kind === 'highlight') nodes[index] = adfNodes(nodes[index] ?? []).map((holder) => ({ ...holder, text: highlightDelimiter }))
   }
   for (const { closer, opener } of paired) {
     nodes[opener] = []
@@ -550,7 +635,8 @@ function markHighlights(pieces: readonly Piece[], nodes: AdfNode[][], paired: re
   }
 }
 
-function highlighted(node: AdfNode): AdfNode {
+function highlighted(node: Inline): Inline {
+  if (!isNode(node)) return node
   const marks = nodeMarks(node)
   if (node.type !== 'text' || Object.keys(nodeAttrs(node)).length > 0 || marks.some((mark) => mark.type === 'code' || mark.type === 'backgroundColor')) return node
   return { ...node, marks: [editorHighlight, ...marks] }
@@ -562,8 +648,9 @@ function markType(character: string, used: number): string {
 }
 
 // A node cannot carry one mark type twice (docs/decisions.md §No schema validation).
-function applyMark(nodes: readonly AdfNode[], mark: AdfMark): AdfNode[] {
+function applyMark(nodes: readonly Inline[], mark: AdfMark): Inline[] {
   return nodes.map((node) => {
+    if (!isNode(node)) return node
     const marks = nodeMarks(node)
     return marks.some((carried) => carried.type === mark.type) ? node : { ...node, marks: [mark, ...marks] }
   })

src/markdown/parse/markdown-to-adf.test.ts

@@ -84,9 +84,21 @@ function table(...rows: AdfNode[]): AdfNode {
   return { content: rows, type: 'table' }
 }
 
-test('builds an empty document from input holding no block', () => {
+test('builds an empty document from input holding no block, and one holding no content key from the doc directive alone', () => {
-  assert.deepEqual(markdownToAdf(''), { ok: true, value: { type: 'doc', version: 1 } })
+  assert.deepEqual(markdownToAdf(''), { ok: true, value: { content: [], type: 'doc', version: 1 } })
-  assert.deepEqual(content(markdownToAdf('\n \n\t\n')), [])
+  assert.deepEqual(markdownToAdf('\n \n\t\n'), { ok: true, value: { content: [], type: 'doc', version: 1 } })
+  assert.deepEqual(markdownToAdf('!adf:doc {content=none}\n'), { ok: true, value: { type: 'doc', version: 1 } })
+  assert.deepEqual(markdownToAdf('\n!adf:doc {content=none}\n\n'), { ok: true, value: { type: 'doc', version: 1 } })
+  const form = 'unsupported-node-shape: doc spells the one form !adf:doc {content=none}: this one spells another'
+  assert.equal(content(markdownToAdf('!adf:doc\n')), form)
+  assert.equal(content(markdownToAdf('!adf:doc x {content=none}\n')), form)
+  assert.equal(content(markdownToAdf('!adf:doc {content=empty}\n')), form)
+  assert.equal(content(markdownToAdf('!adf:doc {content="none"}\n')), form)
+  const alone = 'unsupported-node-shape: !adf:doc {content=none} spells a whole document holding no content key, alone: this one stands among other blocks'
+  assert.equal(content(markdownToAdf('x\n\n!adf:doc {content=none}\n')), alone)
+  assert.equal(content(markdownToAdf('!adf:panel info\n!adf:doc {content=none}\n!adf:/panel\n')), alone)
+  assert.equal(content(markdownToAdf('!adf:doc\n!adf:/doc\n')), 'malformed-directive: doc takes no body, so no !adf:/doc closes it; \\!adf: keeps the prefix literal')
+  assert.equal(content(markdownToAdf('a !adf:doc{content=none}\n')), 'unsupported-node-shape: doc takes the block form, !adf:doc, never the inline form')
 })
 
 test('builds one paragraph from the lines a blank line does not part', () => {
@@ -153,7 +165,7 @@ test('names the slot a codeBlock spells its language outside of', () => {
   const slot = 'unsupported-node-shape: codeBlock spells its language in the fence info string, or in the attribute where no info string carries it back'
   assert.equal(content(markdownToAdf('!adf:codeBlock {language=rust wrap=true}\n```\nx\n```\n!adf:/codeBlock\n')), slot)
   assert.equal(content(markdownToAdf('!adf:codeBlock {language=rust}\n```sql\nx\n```\n!adf:/codeBlock\n')), slot)
-  assert.equal(content(markdownToAdf('!adf:codeBlock {wrap=true}\n```carry\nx\n```\n!adf:/codeBlock\n')), slot)
+  assert.equal(content(markdownToAdf('!adf:codeBlock {wrap=true}\n```adf:x\nx\n```\n!adf:/codeBlock\n')), slot)
   assert.equal(content(markdownToAdf('!adf:codeBlock {wrap=true}\n```a\\b\nx\n```\n!adf:/codeBlock\n')), slot)
 })
 
@@ -343,18 +355,28 @@ test('names the position a directive name the other one spells belongs to', () =
 })
 
 test('names the reserved carry name a block directive spells', () => {
-  const reserved = 'malformed-directive: the name carry is reserved for the opaque carry, whose block form is the carry fence'
+  const reserved = 'malformed-directive: the name carry is reserved for the opaque carry, whose block form is the adf: fence'
   assert.equal(content(markdownToAdf('!adf:carry\n')), reserved)
   assert.equal(content(markdownToAdf('!adf:carry\nx\n!adf:/carry\n')), reserved)
   assert.deepEqual(content(markdownToAdf('```adf\nx\n```\n')), [{ attrs: { language: 'adf' }, content: [text('x')], type: 'codeBlock' }])
+  assert.deepEqual(content(markdownToAdf('```carry\nx\n```\n')), [{ attrs: { language: 'carry' }, content: [text('x')], type: 'codeBlock' }])
 })
 
 const carried = '!adf:carry{json="{\\"type\\":\\"placeholder\\"}"}'
 
-test('reads the carry fence back to the node its JSON holds', () => {
+test('reads the carry fence back to the node its info string names and its JSON holds', () => {
-  assert.deepEqual(content(markdownToAdf('```carry\n{\n  "attrs": {\n    "url": "https://example.com/x"\n  },\n  "type": "blockCard"\n}\n```\n')), [
+  assert.deepEqual(content(markdownToAdf('```adf:blockCard\n{\n  "attrs": {\n    "url": "https://example.com/x"\n  }\n}\n```\n')), [
     { attrs: { url: 'https://example.com/x' }, type: 'blockCard' },
   ])
+  assert.deepEqual(content(markdownToAdf('```adf:\n{\n  "type": "a`b"\n}\n```\n')), [{ type: 'a`b' }])
+  assert.equal(
+    content(markdownToAdf('```adf:blockCard\n{\n  "type": "blockCard"\n}\n```\n')),
+    "unsupported-node-shape: the adf:blockCard fence names its node's type: this JSON holds a type as well",
+  )
+  assert.equal(content(markdownToAdf('```adf:\n{\n  "type": "blockCard"\n}\n```\n')), 'unsupported-node-shape: the carry fence names a type its info string carries: spell it adf:blockCard')
+  assert.equal(content(markdownToAdf('```adf:\n{}\n```\n')), "unsupported-node-shape: the opaque carry holds one ADF node's JSON: this JSON is no ADF node; !adf:codeBlock {language=\"adf:\"} around a bare fence keeps it a code block")
+  const unnamed = 'unsupported-node-shape: the carry fence names a type no info string carries back: spell it adf: with the type in the JSON'
+  assert.equal(content(markdownToAdf('```adf:\\\\\n{}\n```\n')), unnamed)
 })
 
 test('reads the inline carry back to the node its json attribute holds', () => {
@@ -363,26 +385,31 @@ test('reads the inline carry back to the node its json attribute holds', () => {
   ])
 })
 
-test('names the invalid JSON no opaque carry holds', () => {
+test('names the invalid JSON no opaque carry holds, and the spelling that keeps it literal', () => {
-  const invalid = 'malformed-directive: the opaque carry holds invalid JSON'
+  const invalid = 'malformed-directive: the opaque carry holds invalid JSON; '
-  assert.equal(content(markdownToAdf('```carry\n{"type":\n```\n')), invalid)
+  const fence = `${invalid}!adf:codeBlock {language="adf:x"} around a bare fence keeps it a code block`
-  assert.equal(content(markdownToAdf('```carry\n```\n')), invalid)
+  assert.equal(content(markdownToAdf('```adf:x\n{"attrs":\n```\n')), fence)
-  assert.equal(content(markdownToAdf('!adf:carry{json="{"}\n')), invalid)
+  assert.equal(content(markdownToAdf('```adf:x\n```\n')), fence)
-  assert.equal(content(markdownToAdf('!adf:carry{json=abc}\n')), invalid)
+  assert.equal(content(markdownToAdf('!adf:carry{json="{"}\n')), `${invalid}\\!adf: keeps the prefix literal`)
+  assert.equal(content(markdownToAdf('!adf:carry{json=abc}\n')), `${invalid}\\!adf: keeps the prefix literal`)
 })
 
 test('names the canonical spelling a carried JSON reads alone', () => {
   const canonically = "unsupported-node-shape: the opaque carry spells its node's JSON canonically: "
-  assert.equal(content(markdownToAdf('```carry\n{"type":"blockCard"}\n```\n')), `${canonically}two-space indent, keys sorted`)
+  const literal = '; \\!adf: keeps the prefix literal'
-  assert.equal(content(markdownToAdf('!adf:carry{json="{\\"type\\": \\"blockCard\\"}"}\n')), `${canonically}compact, keys sorted`)
+  assert.equal(
-  assert.equal(content(markdownToAdf('!adf:carry{json="{\\"type\\":\\"blockCard\\",\\"attrs\\":{}}"}\n')), `${canonically}compact, keys sorted`)
+    content(markdownToAdf('```adf:blockCard\n{"attrs":{}}\n```\n')),
+    `${canonically}two-space indent, keys sorted; !adf:codeBlock {language="adf:blockCard"} around a bare fence keeps it a code block`,
+  )
+  assert.equal(content(markdownToAdf('!adf:carry{json="{\\"type\\": \\"blockCard\\"}"}\n')), `${canonically}compact, keys sorted${literal}`)
+  assert.equal(content(markdownToAdf('!adf:carry{json="{\\"type\\":\\"blockCard\\",\\"attrs\\":{}}"}\n')), `${canonically}compact, keys sorted${literal}`)
 })
 
 test('names the node JSON an opaque carry restores alone', () => {
-  const node = "unsupported-node-shape: the opaque carry holds one ADF node's JSON: this JSON is no ADF node"
+  const node = "unsupported-node-shape: the opaque carry holds one ADF node's JSON: this JSON is no ADF node; "
-  assert.equal(content(markdownToAdf('```carry\n[]\n```\n')), node)
+  assert.equal(content(markdownToAdf('```adf:x\n[]\n```\n')), `${node}!adf:codeBlock {language="adf:x"} around a bare fence keeps it a code block`)
-  assert.equal(content(markdownToAdf('!adf:carry{json=null}\n')), node)
+  assert.equal(content(markdownToAdf('!adf:carry{json=null}\n')), `${node}\\!adf: keeps the prefix literal`)
-  assert.equal(content(markdownToAdf('!adf:carry{json="{\\"kind\\":\\"x\\"}"}\n')), node)
+  assert.equal(content(markdownToAdf('!adf:carry{json="{\\"kind\\":\\"x\\"}"}\n')), `${node}\\!adf: keeps the prefix literal`)
 })
 
 test('names the shape the inline carry reads alone', () => {
@@ -394,12 +421,12 @@ test('names the shape the inline carry reads alone', () => {
 
 test('holds a carried JSON value to the nesting its position leaves', () => {
   const nested = (levels: number): string => `${'['.repeat(levels)}${']'.repeat(levels)}`
-  const fence = (prefix: string, levels: number): string => `${prefix}\`\`\`carry\n${prefix}${nested(levels)}\n${prefix}\`\`\`\n`
+  const fence = (prefix: string, levels: number): string => `${prefix}\`\`\`adf:x\n${prefix}${nested(levels)}\n${prefix}\`\`\`\n`
   const deeper = (levels: number): string => `unsupported-nesting-depth: a carried node's JSON nests deeper than the ${levels} levels its position leaves`
   assert.equal(content(markdownToAdf(`!adf:carry{json="${nested(largestNesting + 2)}"}\n`)), deeper(largestNesting))
   assert.equal(
     content(markdownToAdf(fence('', largestNesting + 1))),
-    "unsupported-node-shape: the opaque carry spells its node's JSON canonically: two-space indent, keys sorted",
+    'unsupported-node-shape: the opaque carry spells its node\'s JSON canonically: two-space indent, keys sorted; !adf:codeBlock {language="adf:x"} around a bare fence keeps it a code block',
   )
   assert.equal(content(markdownToAdf(fence('> ', largestNesting + 1))), deeper(largestNesting - 1))
 })
@@ -407,11 +434,11 @@ test('holds a carried JSON value to the nesting its position leaves', () => {
 test('names the number no JSON spelling carries in an opaque carry', () => {
   const named = 'unsupported-node-shape: the opaque carry holds a number JSON cannot spell'
   assert.equal(content(markdownToAdf('!adf:carry{json="{\\"attrs\\":{\\"width\\":1e999},\\"type\\":\\"blockCard\\"}"}\n')), named)
-  assert.equal(content(markdownToAdf('```carry\n1e999\n```\n')), named)
+  assert.equal(content(markdownToAdf('```adf:x\n1e999\n```\n')), named)
 })
 
 test('names the mark spelling no opaque carry sits inside', () => {
-  const named = 'unsupported-node-shape: no mark spelling wraps an opaque carry: the carried node restores exactly, marks included'
+  const named = 'unsupported-node-shape: no mark spelling wraps an opaque carry or an inline node spelling marks=empty: its marks are its own'
   assert.equal(content(markdownToAdf(`_a ${carried} b_\n`)), named)
   assert.equal(content(markdownToAdf(`**${carried}**\n`)), named)
   assert.equal(content(markdownToAdf(`~~a ${carried}~~\n`)), named)
@@ -454,7 +481,9 @@ test('names the marks key no marks array reads back from', () => {
   assert.equal(content(markdownToAdf('!adf:rule {marks="[1]"}\n')), named)
   assert.equal(content(markdownToAdf('!adf:rule {marks="{}"}\n')), named)
   assert.equal(content(markdownToAdf('!adf:rule {marks=x}\n')), named)
-  assert.equal(content(markdownToAdf('!adf:rule {marks="[{\\"attrs\\":{},\\"type\\":\\"em\\"}]"}\n')), named)
+  assert.equal(content(markdownToAdf('!adf:rule {marks="[{\\"type\\":\\"em\\",\\"attrs\\":{}}]"}\n')), named)
+  assert.deepEqual(content(markdownToAdf('!adf:rule {marks="[{\\"attrs\\":{},\\"type\\":\\"em\\"}]"}\n')), [{ marks: [{ attrs: {}, type: 'em' }], type: 'rule' }])
+  assert.deepEqual(content(markdownToAdf('!adf:rule {marks=empty}\n')), [{ marks: [], type: 'rule' }])
 })
 
 test('names the attribute a node holds no reading for', () => {
@@ -487,7 +516,8 @@ test('names the argument and the body a node takes no reading for', () => {
   assert.equal(content(markdownToAdf('!adf:rule x\n')), 'unsupported-node-shape: rule takes no argument: this one spells one')
   assert.equal(content(markdownToAdf('!adf:paragraph\nOne.\n\nTwo.\n!adf:/paragraph\n')), 'unsupported-node-shape: paragraph takes one paragraph as its body: this body is not one')
   assert.equal(content(markdownToAdf('!adf:paragraph\n---\n!adf:/paragraph\n')), 'unsupported-node-shape: paragraph takes one paragraph as its body: this body is not one')
-  assert.equal(content(markdownToAdf('!adf:codeBlock {wrap=true}\nx\n!adf:/codeBlock\n')), 'unsupported-node-shape: codeBlock takes one code block as its body: this body is not one')
+  assert.equal(content(markdownToAdf('!adf:codeBlock {wrap=true}\nx\n!adf:/codeBlock\n')), 'unsupported-node-shape: codeBlock takes code blocks as its body: this body holds another block')
+  assert.equal(content(markdownToAdf('!adf:codeBlock {wrap=true}\n!adf:/codeBlock\n')), 'unsupported-node-shape: codeBlock takes code blocks as its body: this body holds none')
   assert.equal(content(markdownToAdf('!adf:paragraph\n![a](/u)\n!adf:/paragraph\n')), 'unmappable-image: no ADF node carries an image inside a paragraph')
   assert.equal(content(markdownToAdf('Part !adf:date[now]{timestamp=1}.\n')), 'unsupported-node-shape: date takes no content: this one holds some')
 })
@@ -545,7 +575,7 @@ test('names the line and the offset in the input a refusal sits at, the innermos
   assert.deepEqual(position(markdownToAdf('- Part.\n- a <span>b</span>\n')), { line: 2, offset: 8 })
   assert.deepEqual(position(markdownToAdf('Part.\n\n!adf:panel info\nMore.\n')), { line: 3, offset: 7 })
   assert.deepEqual(position(markdownToAdf('x\n\na <span>b</span>\n===\n')), { line: 3, offset: 3 })
-  assert.deepEqual(position(markdownToAdf('x\n\n```carry\n{\n```\n')), { line: 3, offset: 3 })
+  assert.deepEqual(position(markdownToAdf('x\n\n```adf:x\n{\n```\n')), { line: 3, offset: 3 })
   assert.deepEqual(position(markdownToAdf('x\n\n| a |\n')), { line: 3, offset: 3 })
   assert.deepEqual(position(markdownToAdf('a\nb <span>c</span>\n')), { line: 1, offset: 0 })
   assert.deepEqual(position(markdownToAdf('Part.\r\n\r\n<div>\r\n')), { line: 3, offset: 9 })
@@ -1053,3 +1083,17 @@ test('names the directive mark left without the content it wraps', () => {
   assert.equal(content(markdownToAdf('!adf:underline[]\n')), named)
   assert.equal(content(markdownToAdf('!adf:underline{}\n')), named)
 })
+
+test('reads the text break only between two text nodes CommonMark joins, building no node', () => {
+  assert.deepEqual(content(markdownToAdf('a!adf:textBreak{}b\n')), [{ content: [text('a'), text('b')], type: 'paragraph' }])
+  assert.deepEqual(content(markdownToAdf('==a!adf:textBreak{}b==\n')), [{ content: [text('==a'), text('b==')], type: 'paragraph' }])
+  const parts = 'unsupported-node-shape: textBreak parts two text nodes CommonMark reads back as one: this one parts something else'
+  const carried = '!adf:carry{json="{\\"text\\":\\"b\\",\\"type\\":\\"text\\"}"}'
+  for (const markdown of ['!adf:textBreak{}a\n', 'a!adf:textBreak{}\n', 'a!adf:textBreak{}!adf:textBreak{}b\n', '**a**!adf:textBreak{}b\n', '[!adf:textBreak{}](/u)\n', `a!adf:textBreak{}${carried}\n`]) {
+    assert.equal(content(markdownToAdf(markdown)), parts, markdown)
+  }
+  assert.equal(content(markdownToAdf('!adf:status[a!adf:textBreak{}b]\n')), 'unsupported-node-shape: textBreak parts two text nodes: a content slot holds plain text')
+  assert.equal(content(markdownToAdf('![a!adf:textBreak{}b](/i)\n')), 'unsupported-node-shape: textBreak parts two text nodes: an image description holds plain text')
+  assert.equal(content(markdownToAdf('a!adf:textBreak{x=y}b\n')), 'unsupported-node-shape: textBreak spells the bare leaf form, !adf:textBreak{}: this one spells more')
+  assert.deepEqual(content(markdownToAdf('!adf:carry{json="{\\"type\\":\\"textBreak\\"}"}\n')), [{ content: [{ type: 'textBreak' }], type: 'paragraph' }])
+})

src/markdown/parse/markdown-to-adf.ts

@@ -5,14 +5,14 @@ import type { ConvertFault } from '../../result.ts'
 import type { Flavour } from '../plain-conventions.ts'
 import type { LineContainer } from '../line-container.ts'
 import type { LinkDefinitions } from './inline-content.ts'
-import { carryName, readCarriedBlock } from '../opaque-carry.ts'
+import { carryFenceType, readCarriedBlock } from '../opaque-carry.ts'
 import { commonMarkSpelling, type SpellingMemo } from '../emit/adf-to-markdown.ts'
+import { documentAttribute, documentName, documentSpelling, listBreakName, listBreakSpelling } from '../block-directive.ts'
 import { failure, faulted, positioned, success, type ConvertErrorPath, type ParseError, type Result, type SourcePosition } from '../../result.ts'
 import { inlineLeaves } from '../emit/plain-inline.ts'
 import { languageSlot } from '../code-language.ts'
 import { largestNesting } from '../../nesting.ts'
 import { leadingMarker, readAlertMarker, readTaskMarker } from '../plain-conventions.ts'
-import { listBreakName, listBreakSpelling } from '../block-directive.ts'
 import { mintTaskIds } from './task-ids.ts'
 import { nodeAttrs, nodeContent } from '../../adf/document.ts'
 import { parseBlocks } from './blocks.ts'
@@ -39,10 +39,15 @@ export function plainMarkdownToAdf(markdown: string): Result<AdfDocument, ParseE
 
 function readDocument(markdown: string, flavour: Flavour): Result<AdfDocument, ParseError> {
   const parsed = parseBlocks(markdown)
+  const [only] = parsed.blocks
+  if (parsed.blocks.length === 1 && only?.kind === 'directive' && only.name === documentName) {
+    const fault = documentFault(only)
+    return fault === undefined ? success({ type: 'doc', version: 1 }) : positioned(faulted(fault, []), only.position)
+  }
   const reading: Reading = { carried: new Set(), definitions: parsed.definitions, flavour, inExpand: false, memo: new Map() }
   const content = positioned(readBlocks(parsed.blocks, reading, [], 0), documentStart)
   if (!content.ok) return content
-  const document: AdfDocument = content.value.length === 0 ? { type: 'doc', version: 1 } : { content: content.value, type: 'doc', version: 1 }
+  const document: AdfDocument = { content: content.value, type: 'doc', version: 1 }
   if (flavour === 'plain') mintTaskIds(document, markdown, reading.carried)
   return success(document)
 }
@@ -52,6 +57,10 @@ function readBlocks(blocks: readonly Block[], reading: Reading, path: ConvertErr
   const content: AdfNode[] = []
   for (const [index, block] of blocks.entries()) {
     const nodePath = [...path, 'content', content.length]
+    if (block.kind === 'directive' && block.name === documentName) {
+      const fault = documentFault(block) ?? unsupportedNodeShape(`${documentSpelling} spells a whole document holding no content key, alone: this one stands among other blocks`)
+      return positioned(faulted(fault, nodePath), block.position)
+    }
     if (block.kind === 'directive' && block.name === listBreakName) {
       const fault = listBreakFault(block, blocks[index - 1], blocks[index + 1])
       if (fault !== undefined) return positioned(faulted(fault, nodePath), block.position)
@@ -73,6 +82,12 @@ function listBreakFault(block: DirectiveBlock, previous: Block | undefined, next
   return previous.kind === next?.kind ? undefined : partsFault()
 }
 
+function documentFault(block: DirectiveBlock): ConvertFault | undefined {
+  const spelled = block.attributes.get(documentAttribute.key)
+  if (block.argument === undefined && block.attributes.size === 1 && spelled?.spelling === documentAttribute.value) return undefined
+  return unsupportedNodeShape(`${documentName} spells the one form ${documentSpelling}: this one spells another`)
+}
+
 function partsFault(): ConvertFault {
   return unsupportedNodeShape(`${listBreakName} parts two adjacent lists of one type: this one parts something else`)
 }
@@ -216,22 +231,32 @@ function directiveNode(block: DirectiveBlock, reading: Reading, path: ConvertErr
 function directiveBody(read: BlockDirectiveNode, blocks: Block[] | undefined, reading: Reading, path: ConvertErrorPath, depth: number): Result<AdfNode> {
   const { contentModel, node } = read
   if (blocks === undefined) return success(node)
+  // A directive builds a content key only from content=empty, which holds no body.
+  if (node.content !== undefined) return blocks.length === 0 ? success(node) : failure('unsupported-node-shape', `${node.type} spells content=empty, which holds no body: this one holds one`, path)
   if (contentModel === 'code') return codeDirectiveNode(node, blocks, path)
   if (contentModel === 'inline') return inlineBodyNode(node, blocks, reading, path)
   return containerNode(node, blocks, reading, path, depth)
 }
 
+// spec/flavour.md, The CommonMark blocks: one fence per text node, every fence carrying the one language.
 function codeDirectiveNode(node: AdfNode, blocks: readonly Block[], path: ConvertErrorPath): Result<AdfNode> {
-  const only = blocks.length === 1 ? blocks[0] : undefined
+  const fences: Extract<Block, { kind: 'code' }>[] = []
-  if (only?.kind !== 'code') return failure('unsupported-node-shape', `${node.type} takes one code block as its body: this body is not one`, path)
+  for (const block of blocks) {
+    if (block.kind !== 'code') return failure('unsupported-node-shape', `${node.type} takes code blocks as its body: this body holds another block`, path)
+    fences.push(block)
+  }
+  const [first] = fences
+  if (first === undefined) return failure('unsupported-node-shape', `${node.type} takes code blocks as its body: this body holds none`, path)
+  if (fences.some((fence) => fence.language !== first.language)) return failure('unsupported-node-shape', `${node.type} holds one language, so its fences carry one info string: these differ`, path)
+  if (fences.length > 1 && fences.some((fence) => fence.text === '')) return failure('unsupported-node-shape', `a fence beside another spells a text node, which holds text: this one is empty`, path)
   const attribute = nodeAttrs(node)['language']
-  const fromFence = only.language !== ''
+  const fromFence = first.language !== ''
-  const slot = languageSlot(fromFence ? only.language : attribute)
+  const slot = languageSlot(fromFence ? first.language : attribute)
   if ((slot.kind === 'fence') !== fromFence || (fromFence && attribute !== undefined)) {
     return failure('unsupported-node-shape', `${node.type} spells its language in the fence info string, or in the attribute where no info string carries it back`, path)
   }
-  const spelled = fromFence ? { ...node, attrs: { ...node.attrs, language: only.language } } : node
+  const spelled = fromFence ? { ...node, attrs: { ...node.attrs, language: first.language } } : node
-  return success(withContent(spelled, only.text === '' ? [] : [{ text: only.text, type: 'text' }]))
+  return success(withContent(spelled, first.text === '' ? [] : fences.map((fence): AdfNode => ({ text: fence.text, type: 'text' }))))
 }
 
 function tableNode(rows: readonly string[][], reading: Reading, path: ConvertErrorPath): Result<AdfNode> {
@@ -278,8 +303,9 @@ function listNode(node: AdfNode, items: readonly Block[][], reading: Reading, pa
 }
 
 function codeBlockNode(language: string, text: string, reading: Reading, path: ConvertErrorPath, depth: number): Result<AdfNode> {
-  if (language === carryName) {
+  const carriedType = carryFenceType(language)
-    const carried = readCarriedBlock(text, depth)
+  if (carriedType !== undefined) {
+    const carried = readCarriedBlock(carriedType, text, depth)
     if (carried.fault !== undefined) return faulted(carried.fault, path)
     reading.carried.add(carried.value)
     return success(carried.value)

src/markdown/parse/plain-markdown-to-adf.test.ts

@@ -6,7 +6,6 @@ import { adfToMarkdown } from '../emit/adf-to-markdown.ts'
 import { adfToPlainMarkdown } from '../emit/plain-reduction.ts'
 import { largestNesting } from '../../nesting.ts'
 import { markdownToAdf, plainMarkdownToAdf } from './markdown-to-adf.ts'
-import { toEditorNormal } from '../../adf/editor-normal.ts'
 
 const code: AdfMark = { type: 'code' }
 const em: AdfMark = { type: 'em' }
@@ -18,7 +17,7 @@ const taskTypes = ['blockTaskItem', 'taskItem', 'taskList']
 function read(markdown: string): readonly AdfNode[] | string {
   const parsed = plainMarkdownToAdf(markdown)
   if (!parsed.ok) return parsed.error.code
-  const blocks = toEditorNormal(parsed.value).content ?? []
+  const blocks = parsed.value.content ?? []
   const pending = [...blocks]
   for (let block = pending.pop(); block !== undefined; block = pending.pop()) {
     for (const child of block.content ?? []) pending.push(child)
@@ -31,8 +30,8 @@ function read(markdown: string): readonly AdfNode[] | string {
   return blocks
 }
 
-function normal(...blocks: AdfNode[]): readonly AdfNode[] {
+function blocks(...content: AdfNode[]): readonly AdfNode[] {
-  return toEditorNormal(document(...blocks)).content ?? []
+  return content
 }
 
 function document(...content: AdfNode[]): AdfDocument {
@@ -57,7 +56,7 @@ function bare(type: string, ...content: AdfNode[]): AdfNode {
 }
 
 function paragraph(...content: AdfNode[]): AdfNode {
-  return bare('paragraph', ...content)
+  return content.length === 0 ? { type: 'paragraph' } : bare('paragraph', ...content)
 }
 
 function said(value: string): AdfNode {
@@ -101,7 +100,7 @@ test('reads the rest of an alert marker line as the panel first body paragraph,
   assert.deepEqual(read('> [!tip] Title\n> body\n'), [panel('tip', said('Title'), said('body'))])
   assert.deepEqual(read('> [!tip] Title\\\n> body\n'), [panel('tip', said('Title'), said('body'))])
   assert.deepEqual(read('> [!NOTE]\\\n> Broken.\n'), [panel('info', said('Broken.'))])
-  assert.deepEqual(read('> [!NOTE]\n'), normal(panel('info', paragraph())))
+  assert.deepEqual(read('> [!NOTE]\n'), blocks(panel('info', paragraph())))
   assert.deepEqual(read('> > [!WARNING]\n> > Inner.\n'), [bare('blockquote', panel('warning', said('Inner.')))])
 })
 
@@ -120,27 +119,27 @@ test('reads a folded callout to an expand titled by the rest of its marker line,
     node('expand', { title: 'Why?' }, paragraph(text('See '), text('the docs', link), text(', '), text('now', strong), text('.'))),
   ])
   assert.deepEqual(read('> [!NOTE]- Two\\\n> lines\n'), [node('expand', { title: 'Two' }, said('lines'))])
-  assert.deepEqual(read('> [!faq]- See [x](http://y)\n'), normal(node('expand', { title: 'See x (http://y)' }, paragraph())))
+  assert.deepEqual(read('> [!faq]- See [x](http://y)\n'), blocks(node('expand', { title: 'See x (http://y)' }, paragraph())))
-  assert.deepEqual(read('> [!faq]- [a **b**](u "t")[c](u) and [d](v)\n'), normal(node('expand', { title: 'a bc (u) and d (v)' }, paragraph())))
+  assert.deepEqual(read('> [!faq]- [a **b**](u "t")[c](u) and [d](v)\n'), blocks(node('expand', { title: 'a bc (u) and d (v)' }, paragraph())))
-  assert.deepEqual(read('> [!faq]- <http://y> or <a@b.c> or <http://a\\b>\n'), normal(node('expand', { title: 'http://y or a@b.c or http://a\\b' }, paragraph())))
+  assert.deepEqual(read('> [!faq]- <http://y> or <a@b.c> or <http://a\\b>\n'), blocks(node('expand', { title: 'http://y or a@b.c or http://a\\b' }, paragraph())))
-  assert.deepEqual(read('> [!faq]- [a&#10;b](u)\n'), normal(node('expand', { title: 'a\nb (u)' }, paragraph())))
+  assert.deepEqual(read('> [!faq]- [a&#10;b](u)\n'), blocks(node('expand', { title: 'a\nb (u)' }, paragraph())))
-  assert.deepEqual(read('> [!faq]- [!adf:mention[@M]{id=5}](u) !adf:inlineCard{url="http://y"}\n'), normal(node('expand', { title: '@M (u) http://y' }, paragraph())))
+  assert.deepEqual(read('> [!faq]- [!adf:mention[@M]{id=5}](u) !adf:inlineCard{url="http://y"}\n'), blocks(node('expand', { title: '@M (u) http://y' }, paragraph())))
-  assert.deepEqual(read('> [!NOTE]- Set ==x== here\n'), normal(node('expand', { title: 'Set ==x== here' }, paragraph())))
+  assert.deepEqual(read('> [!NOTE]- Set ==x== here\n'), blocks(node('expand', { title: 'Set ==x== here' }, paragraph())))
   assert.deepEqual(read('> [!NOTE]-\n>\n> Line.\n'), [bare('expand', said('Line.'))])
 })
 
 test('reads a folded callout inside an expand to a nested expand', () => {
   const markdown = '> [!NOTE]- Outer\n>\n> > [!NOTE]- Inner\n> >\n> > Deep.\n>\n> > [!TIP]\n> >\n> > > [!NOTE]-\n'
-  assert.deepEqual(read(markdown), normal(node('expand', { title: 'Outer' }, node('nestedExpand', { title: 'Inner' }, said('Deep.')), panel('tip', bare('nestedExpand', paragraph())))))
+  assert.deepEqual(read(markdown), blocks(node('expand', { title: 'Outer' }, node('nestedExpand', { title: 'Inner' }, said('Deep.')), panel('tip', bare('nestedExpand', paragraph())))))
-  assert.deepEqual(read('- > [!NOTE]-\n'), normal(bare('bulletList', bare('listItem', bare('expand', paragraph())))))
+  assert.deepEqual(read('- > [!NOTE]-\n'), blocks(bare('bulletList', bare('listItem', bare('expand', paragraph())))))
-  assert.deepEqual(read('!adf:expand\n> [!NOTE]- Inner\n!adf:/expand\n'), normal(bare('expand', node('nestedExpand', { title: 'Inner' }, paragraph()))))
+  assert.deepEqual(read('!adf:expand\n> [!NOTE]- Inner\n!adf:/expand\n'), blocks(bare('expand', node('nestedExpand', { title: 'Inner' }, paragraph()))))
 })
 
 test('reads a bullet list whose every item leads with a task marker to a task list', () => {
   assert.deepEqual(read('- [x] Write the spec\n- [ ] Ship **it**\n- [X] Tell\n'), [
     bare('taskList', task('DONE', text('Write the spec')), task('TODO', text('Ship '), text('it', strong)), task('DONE', text('Tell'))),
   ])
-  assert.deepEqual(read('- [x]\n- [ ]\\\n  after\n'), normal(bare('taskList', task('DONE'), task('TODO', text('after')))))
+  assert.deepEqual(read('- [x]\n- [ ]\\\n  after\n'), blocks(bare('taskList', task('DONE'), task('TODO', text('after')))))
   const minted = plainMarkdownToAdf('- [x] Parent\n  - [ ] Child\n')
   assert.deepEqual(minted.ok ? minted.value.content : minted.error.code, [
     node(
@@ -202,7 +201,7 @@ test('reads what the reduction wrote back to the node it reduced, less the attri
     assert.deepEqual(roundTripped(node('panel', { localId, panelType }, said('Check.'))), [panel(panelType, said('Check.'))], panelType)
   }
   const expand = node('expand', { localId, title: 'Log' }, said('Line.'), node('nestedExpand', { title: 'Inner' }, said('Deep.')))
-  assert.deepEqual(roundTripped(node('panel', { panelType: 'tip' }, paragraph()), node('expand', { title: 'Empty' }, paragraph())), normal(panel('tip', paragraph()), node('expand', { title: 'Empty' }, paragraph())))
+  assert.deepEqual(roundTripped(node('panel', { panelType: 'tip' }, paragraph()), node('expand', { title: 'Empty' }, paragraph())), blocks(panel('tip', paragraph()), node('expand', { title: 'Empty' }, paragraph())))
   assert.deepEqual(roundTripped(expand), [node('expand', { title: 'Log' }, said('Line.'), node('nestedExpand', { title: 'Inner' }, said('Deep.')))])
   const tasks = bare(
     'taskList',
@@ -266,3 +265,8 @@ test('keeps what markdownToAdf reads that no row reads, and refuses only what it
   for (let level = 0; level < largestNesting; level += 1) deep = `> ${deep}`
   assert.equal(typeof read(deep), 'object')
 })
+
+test('joins a carried text node to no neighbour, inside a highlight too', () => {
+  const carried = '!adf:carry{json="{\\"marks\\":[],\\"text\\":\\"b\\",\\"type\\":\\"text\\"}"}'
+  assert.deepEqual(read(`==a${carried}==\n`), blocks(paragraph(text('a', highlight), { marks: [], text: 'b', type: 'text' })))
+})

src/markdown/text-break.ts

@@ -0,0 +1,12 @@
+import type { AdfNode } from '../adf/document.ts'
+import { identicalMarks, isPlainText, nodeMarks } from '../adf/document.ts'
+import { spellInlineLeafDirective } from './directive-syntax.ts'
+
+export const textBreakName = 'textBreak'
+
+export const textBreakSpelling = spellInlineLeafDirective(textBreakName, '')
+
+// Whether CommonMark reads the pair back as one text node, where neither rides the carry (spec/flavour.md, Inline nodes).
+export function readsAsOne(previous: AdfNode, node: AdfNode): boolean {
+  return isPlainText(previous) && isPlainText(node) && identicalMarks(nodeMarks(previous), nodeMarks(node))
+}

todo.md

@@ -6,7 +6,7 @@
 
 `Bar = 9`
 
-`Next ID = 52`
+`Next ID = 59`
 
 | Goal | W |
 |---|---|
@@ -24,13 +24,16 @@
 
 | ID | Release | Exempt | Item | R | S | A | G | Goals | Score |
 |---|---|---|---|---|---|---|---|---|---|
-| 40 | 0.2.0 | decision | **Make `markdownToAdf(adfToMarkdown(doc))` deep-equal `doc` for every document `adfToMarkdown` takes.** | 6 | 7 | 8 | 9 | 1 | 26.2 |
 | 7 | 0.2.0 |  | **Ship HTML: `adfToHtml`, `htmlToAdf`, and `markdownToHtml` / `htmlToMarkdown` composed through ADF.** | 6 | 9 | 9 | 9 | 2, 3 | 25.8 |
+| 55 | 0.2.0 | defect | **Spell through the carry every document the emitter refuses today for a carriage return, a NUL, a code-span line start or a newline inside an emoji, mention or status.** | 5 | 5 | 7 | 9 | 1 | 25.8 |
 | 45 | 0.2.0 |  | **Replace `isAdfDocument` with a reader returning `Result<AdfDocument>`.** | 2 | 3 | 6 | 8 | 1 | 25.2 |
 | 6 | 0.2.0 | decision | **Specify the HTML dialect.** | 2 | 6 | 7 | 8 | 2, 3 | 24.7 |
-| 43 | 0.2.0 |  | **Give each markdown input its own reader, strict to its own standard.** | 6 | 7 | 8 | 9 | 3, 4 | 22.3 |
+| 43 | 0.2.0 | decision | **Give each markdown input its own reader, strict to its own standard.** | 6 | 7 | 8 | 9 | 3, 4 | 22.3 |
 | 49 | 0.2.0 |  | **Read a list whose bullet or ordered delimiter changes as two lists in the CommonMark reader.** | 4 | 5 | 5 | 8 | 3, 4 | 17.2 |
+| 52 | 0.2.0 |  | **Spell `colwidth` as a comma list, `colwidth="340,420"`.** | 3 | 3 | 5 | 6 | 5 | 13.0 |
 | 51 | 0.2.0 |  | **Match a reference label to its definition under Unicode case folding.** | 2 | 2 | 2 | 7 | 3, 4 | 12.4 |
+| 54 | 0.2.0 |  | **Make `docs/decisions.md` §The source parts by ADF and format name every import `parse/` takes from `emit/`.** | 2 | 2 | 2 | 5 | 1 | 11.5 |
+| 53 | 0.2.0 |  | **Bench a block's `marks` spelling with the writer panel and adopt its pick.** | 4 | 5 | 5 | 6 | 5 | 11.5 |
 | 50 | 0.2.0 |  | **Read `[](/url)` and `[]()` as CommonMark's empty link.** | 4 | 4 | 3 | 6 | 3, 4, 6 | 10.4 |
 | 38 | 0.3.0 |  | **Spell a lone surrogate in a text node so it survives a UTF-8 encode.** | 2 | 2 | 4 | 7 | 1 | 19.5 |
 | 47 | 0.3.0 |  | **Open the README with what the package is, what it does and for whom.** | 1 | 4 | 7 | 9 | 9 | 14.0 |
@@ -41,24 +44,29 @@
 | 33 | 0.3.0 |  | **Emit a line in time linear in its mark runs, in `adfToMarkdown` and `adfToPlainMarkdown`.** | 4 | 5 | 6 | 9 | 8 | 10.7 |
 | 46 | 0.3.0 |  | **Publish the bundle size in the README, failing the release pipeline when it drifts.** | 2 | 4 | 5 | 5 | 7, 9 | 10.3 |
 | 9 | 0.3.0 |  | **Ship an online sandbox: a web page with two textboxes converting between ADF and markdown on the library's browser build.** | 2 | 6 | 6 | 8 | 9 | 10.3 |
+| 56 | 0.3.0 | principle | **Give each piece of `blocks.ts`'s block-walk state one owner that returns what it changes.** | 4 | 5 | 2 | 4 | 1 | 6.8 |
+| 57 | 0.3.0 | principle | **Make each `ci.sh` leg build what it reads, so one leg run alone tests the current tree.** | 2 | 3 | 1 | 3 | 7 | 1.2 |
+| 58 | 0.3.0 | principle | **Port `ci.sh`, `publish.sh` and `docker-runner.sh` to standalone Python scripts.** | 4 | 6 | 1 | 2 | 7 | -2.2 |
 | 8 | 0.4.0 |  | **Ship a CLI.** | 3 | 7 | 7 | 6 | 9 | 10.6 |
 
 ## Details
 
-### 40. Make `markdownToAdf(adfToMarkdown(doc))` deep-equal `doc` for every document `adfToMarkdown` takes.
-
-Today it holds for editor-normal documents only: two adjacent text nodes with the same marks merge,
-an empty `attrs`, `marks` or `content` drops, and `-0` reads back `0` — shapes pipelines and bots
-build. Spell each so it reads back as written; CommonMark's spelling stays wherever the document
-holds none of these shapes. The spellings are part of the chunk. `docs/decisions.md` §Equality is
-editor-normal, `spec/flavour.md` and `corpus/README.md` follow, and the tests drop `toEditorNormal`.
-
 ### 7. Ship HTML: `adfToHtml`, `htmlToAdf`, and `markdownToHtml` / `htmlToMarkdown` composed through ADF.
 
 Lands after item 6. The CommonMark spec suite also runs against `markdownToHtml`. The README
 documents HTML as it documents markdown, and its tagline and `package.json`'s `description` regain
 HTML.
 
+### 55. Spell through the carry every document the emitter refuses today for a carriage return, a NUL, a code-span line start or a newline inside an emoji, mention or status.
+
+`inline-line.ts` refuses a carriage return or NUL in text (`unspellable-character`) and a paragraph
+opening with a code-span run (`unspellable-line-start`); `fencedTexts` refuses the same characters
+in a code block; `directive-syntax.ts` refuses a newline in an emoji, mention or status
+(`unspellable-whitespace`). The inline carry's JSON escapes all of them, so a lossless spelling
+exists, and §The code list says a cause the carry answers gets no code. Fixing it revises §Which
+code a cause takes and §Markdown in is a canonical fixpoint, which the maintainer decides. Found by
+the README-goals audit, 2026-10-03.
+
 ### 45. Replace `isAdfDocument` with a reader returning `Result<AdfDocument>`.
 
 Goal 1 has every call return a result; the boolean guard is the one export that does not, and it
@@ -74,7 +82,8 @@ HTML in input). The set sorts per `docs/decisions.md` §Foreign HTML sorts three
 ### 43. Give each markdown input its own reader, strict to its own standard.
 
 Today `markdownToAdf` reads CommonMark and the lossless flavour as one input: text shaped like a
-directive, a pipe table or a `~~` pair becomes a flavour node where CommonMark reads plain text. A
+directive, a pipe table or a `~~` pair becomes a flavour node where CommonMark reads plain text, and
+a code fence whose info string opens `adf:` becomes the block carry where CommonMark reads code. A
 caller names the markdown it hands in: CommonMark, read as its spec says, or the lossless flavour,
 read as `spec/flavour.md` says. Breaking: `MIGRATION.md` says which call a caller takes.
 
@@ -90,6 +99,11 @@ Readings table gains its row, and its Spellings table one if `!adf:listBreak` re
 and 302 lose their `pending` exceptions, and the spelling leaves the README's "Four CommonMark
 spellings" bullet, which counts one fewer.
 
+### 52. Spell `colwidth` as a comma list, `colwidth="340,420"`.
+
+A writer panel chose it on 2026-10-03, 5 of 7, over today's `colwidth="[340,420]"`. Breaking:
+`MIGRATION.md`'s Spellings table gains its row.
+
 ### 51. Match a reference label to its definition under Unicode case folding.
 
 `link-syntax.ts` normalizes a label with `toLowerCase`, so `[ẞ]` misses its `[SS]` definition (spec
@@ -97,6 +111,18 @@ example 540); lowercasing and then uppercasing folds it. Breaking, so it ships b
 `MIGRATION.md`'s Readings table gains its row. Its `pending` exceptions go, and its spelling leaves
 the README's "Four CommonMark spellings" bullet, which counts one fewer.
 
+### 54. Make `docs/decisions.md` §The source parts by ADF and format name every import `parse/` takes from `emit/`.
+
+The entry says `parse/` imports only `commonMarkSpelling` and `openingLinkTakesDirective` from
+`emit/`; `parse/markdown-to-adf.ts` also imports `inlineLeaves` from `emit/plain-inline.ts`. Either
+the entry names it with its reason, or the reader stops importing it.
+
+### 53. Bench a block's `marks` spelling with the writer panel and adopt its pick.
+
+Today `marks="[{\"attrs\":{\"mode\":\"wide\"},\"type\":\"breakout\"}]"`, the marks array as
+escaped JSON. Breaking where the panel picks another spelling: `MIGRATION.md`'s Spellings table
+gains its row.
+
 ### 50. Read `[](/url)` and `[]()` as CommonMark's empty link.
 
 Both stay literal text today (spec examples 484 and 487). ADF holds no empty text node to carry a
@@ -173,6 +199,23 @@ or report the unminified gzip). The figure lands in README §The package beside
 dependencies" claim. Measured today, unminified: tarball 60.4 kB, unpacked 221.5 kB, JS gzipped 45.6
 kB.
 
+### 56. Give each piece of `blocks.ts`'s block-walk state one owner that returns what it changes.
+
+Technical principle "One owner per value": the `Walk` record passes through twelve functions that
+mutate it and return `void`; `walk.leaf` alone is written in seven places. `ContainerStack` in the
+same file shows the shape to follow.
+
+### 57. Make each `ci.sh` leg build what it reads, so one leg run alone tests the current tree.
+
+Technical principle "Compose, do not entangle": Deno, Bun and the tests read the install leg's
+`node_modules`, the floor and consumer legs the pack leg's, and the browser leg whatever `dist/` the
+build leg left, so a leg run alone can pass on stale output.
+
+### 58. Port `ci.sh`, `publish.sh` and `docker-runner.sh` to standalone Python scripts.
+
+Technical principle "Prefer a standalone Python script over a shell script": `AGENTS.md` §3 teaches
+bash footguns (`&&` chaining under `||`, `tee /dev/stderr`) the port removes.
+
 ### 8. Ship a CLI.
 
 The Goals and G cells are provisional: no README goal or persona covers a CLI yet. The chunk