adf-codec/todo.md

todo

Scoring

Score = -R - S/4 + 2*A + 2*G*W

Bar = 9

Next ID = 74

Goal	W
1	1.00
2	0.89
3	0.78
4	0.67
5	0.56
6	0.44
7	0.33
8	0.22
9	0.11

Items

ID	Release	Exempt	Item	R	S	A	G	Goals	Score
7	0.2.0		Ship HTML: adfToHtml, htmlToAdf, and markdownToHtml / htmlToMarkdown composed through ADF.	6	9	9	9	2, 4	25.8
45	0.2.0		Replace isAdfDocument with a reader returning Result<AdfDocument>.	2	3	6	8	1	25.2
67	0.2.0	defect	Push an expand title's leaves one at a time, so a title carrying many text nodes returns a result.	1	1	4	9	1	24.8
6	0.2.0	decision	Specify the HTML dialect.	2	6	7	8	2, 4	24.7
64	0.2.0	defect	Read back unchanged on V8 every JSON key the emitter writes, with a fixture whose key holds \, " or a control character.	4	4	6	8	1, 8	23.0
66	0.2.0	defect	Refuse an attribute value that is not a plain object, array or JSON primitive.	3	2	5	8	1	22.5
71	0.2.0		Compare the package with the other ADF converters in the README, each claim checked against that package's own docs or behaviour.	1	3	6	7	3	21.2
43	0.2.0	decision	Give each markdown input its own reader, strict to its own standard.	6	7	8	9	4, 5	20.3
68	0.2.0		Carry a non-text inline node holding content or text, a block holding text, and a leaf block holding content.	3	3	4	7	1	18.2
65	0.2.0	defect	Read an image not alone in its paragraph as its alt text linked to its URL, and a titled image alone in its paragraph as the image captioned with its title.	5	5	7	7	4, 5, 7	17.1
70	0.2.0		Give package.json the keywords someone searching npm for an ADF converter types.	1	1	4	6	3	16.1
72	0.2.0		Make GitHub the canonical repository, with gitea.larvit.se/larvit/adf-codec a read-only pull mirror of it.	5	6	5	8	3	16.0
49	0.2.0		Read a list whose bullet or ordered delimiter changes as two lists, in both readers, and retire !adf:listBreak.	4	5	5	8	4, 5	15.5
61	0.2.0	decision	Have the emitter ask the inline reader how a line reads back, in place of line-escaping.ts predicting it.	6	8	3	8	1	14.0
52	0.2.0		Spell colwidth as a comma list, colwidth="340,420".	3	3	5	6	6	11.5
51	0.2.0		Match a reference label to its definition under Unicode case folding.	2	2	2	7	4, 5	10.9
60	0.2.0	decision	Collect the questions parse/ asks emit/ into one named module.	3	4	2	6	2	10.7
59	0.2.0	decision	Group the directive grammar into src/markdown/directive/, move Read<T> to result.ts, and move Flavour to markdown/flavour.ts.	3	5	2	6	2	10.4
53	0.2.0		Put a block's marks spelling to the writer panel and adopt its pick.	4	5	5	6	6	10.0
62	0.2.0	decision	Move the plain flavour's reading out of parse/ and its writing out of emit/ into markdown/plain/, so emit/ no longer imports plain/.	4	5	2	6	2	9.4
50	0.2.0		Read [](/url) and []() as CommonMark's empty link.	4	4	3	6	4, 5, 7	9.0
47	0.3.0		Open the README with what the package is, what it does and for whom.	1	4	7	9	3	26.0
73	0.3.0		Announce the package where someone needing an ADF converter already reads: JRACLOUD-77436, the Atlassian developer community and Stack Overflow's ADF-to-markdown questions.	1	2	6	8	3	23.0
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	3	21.0
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
48	0.3.0		Keep the release path publishing past npm's bypass-2FA token retirement.	4	4	8	3	3	15.7
46	0.3.0		Publish the bundle size in the README, failing the release pipeline when it drifts.	2	4	5	5	3, 8	14.8
34	0.3.0		Read emphasis flanking by the whole character beside an astral symbol.	2	3	3	6	4, 5	11.3
31	0.3.0		Make the branch-coverage figure repeat across runs of an unchanged tree.	2	3	3	4	1	11.2
42	0.3.0	defect	Trim a text leaf's trailing blanks in linear time.	1	2	5	9	9	10.5
33	0.3.0	defect	Emit a line in time linear in its mark runs, in adfToMarkdown and adfToPlainMarkdown.	4	5	6	9	9	8.7
56	0.3.0	principle	Give each piece of blocks.ts's block-walk state and inline-content.ts's Scan 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	8	0.6
58	0.3.0	principle	Port ci.sh, publish.sh and docker-runner.sh to standalone Python scripts.	4	6	1	2	8	-2.6
8	0.4.0		Ship a CLI.	3	7	7	6	3	18.6

Details

7. Ship HTML: adfToHtml, htmlToAdf, and markdownToHtml / htmlToMarkdown composed through ADF.

Lands after items 6, 59, 60, 61 and 62. The CommonMark spec suite also runs against markdownToHtml. The README documents HTML as it documents markdown.

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 cannot say which branch refused, where not-an-adf-document's message already does. Breaking: MIGRATION.md shows the guard's replacement.

67. Push an expand title's leaves one at a time, so a title carrying many text nodes returns a result.

quoteNode's title loop (parse/markdown-to-adf.ts) spreads inlineLeaves's result into push, against docs/decisions.md §Nothing spreads an unbounded array: an inline carry in a folded callout's title holding an unknown node with ~125k text children throws a RangeError from plainMarkdownToAdf. Found by the README-goals audit, 2026-10-04.

6. Specify the HTML dialect.

Element-by-element mapping, the data-* fidelity scheme, the opaque-carry form, and the documented foreign-element set htmlToAdf accepts — the set markdownToAdf shares (spec/flavour.md §Raw HTML in input). The set sorts per docs/decisions.md §Foreign HTML sorts three ways.

64. Read back unchanged on V8 every JSON key the emitter writes, with a fixture whose key holds \, " or a control character.

property-harness.ts strips \, " and control characters from generated JSON keys, citing V8's JSON.parse returning a wrong key for an escaped backslash. The library reads json attributes (directive-syntax.ts) and both carries (opaque-carry.ts) through that JSON.parse, so on Node, Deno and Chrome the round-trip would refuse its own output or read back a different key; the fixture tells which. Confirm with a fixture first; then read JSON with our own parser or record the gap with an ending item. Found by the README-goals audit, 2026-10-03.

66. Refuse an attribute value that is not a plain object, array or JSON primitive.

isJsonValue (json-value.ts) walks any object by Object.values, so a Date, Map, Set or typed array in attrs, as structuredClone keeps it, passes the guard and is written as {} or a key per element: ok: true with content lost. Accept an object whose prototype is Object.prototype or null, and an array. Found by the stability review of item 63, 2026-10-04.

71. Compare the package with the other ADF converters in the README, each claim checked against that package's own docs or behaviour.

Lands after item 7, whose HTML the comparison names. A table: which directions each converts, whether the round-trip holds, what an unknown node does, the formats, runtime dependencies and size. Candidates are the packages an npm search for ADF and markdown returns, marklassian among them. Every cell cites that package's README or a run against its published version, since a reader relies on it. Found while planning discoverability, 2026-10-04.

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, 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.

68. Carry a non-text inline node holding content or text, a block holding text, and a leaf block holding content.

refuseContentAndText (emit/inline-line.ts) and emitDirectiveBlock (emit/adf-to-markdown.ts) refuse these with unsupported-node-shape, though isAdfDocument accepts them and the carry round-trips them; docs/decisions.md §The code list gives a cause the carry answers no code. Drop them from the emitting half of the README's unsupported-node-shape row. Found by the README-goals audit, 2026-10-04.

65. Read an image not alone in its paragraph as its alt text linked to its URL, and a titled image alone in its paragraph as the image captioned with its title.

Both are refused today with unmappable-image: 14 examples in corpus/commonmark-spec/refusals.json, a bot's See ![diagram](url) here among them. mediaInline takes only Atlassian media ids, never a URL. An image not alone in its paragraph reads as its alt text, linked to its URL. Its title becomes the link's title. An empty alt text reads as the URL, as item 50 reads an empty link. A titled image alone in its paragraph reads as the image, with its title as a caption (the maintainer, 2026-10-04). By Goal 5, the lossless flavour reads both as CommonMark does. An image inside a link: [![moon](moon.jpg)](/uri) (spec example 517). Alone in its paragraph, it reads as the image with a link mark to /uri, which ADF's media takes. Not alone in its paragraph, the image reads as its alt text linked to /uri, and its URL drops. A writer panel chose this reading 3 of 3 (2026-10-04), by Goal 6, which outranks Goal 7.

70. Give package.json the keywords someone searching npm for an ADF converter types.

package.json has none, and npm's search ranks on them. Take the terms the other ADF converters on npm list and the names the README uses — ADF, Atlassian Document Format, markdown, CommonMark, HTML — so a search for any of them can reach the package. Found while planning discoverability, 2026-10-04.

72. Make GitHub the canonical repository, with gitea.larvit.se/larvit/adf-codec a read-only pull mirror of it.

The maintainer's call, 2026-10-04: GitHub is where an ADF converter is searched for, and the Gitea copy keeps the move back open. The workflows move to .github/workflows/ with ci.sh and publish.sh unchanged; package.json's repository, bugs and homepage, the README's, CHANGELOG's and MIGRATION's links, and AGENTS.md's forge commands and merge style follow. docs/decisions.md gains the entry "GitHub is canonical, and Gitea a pull mirror of it" (Goal 3), and every entry naming Gitea Actions is revised. Creating the GitHub repository, its NPM_TOKEN secret, and re-creating the Gitea copy as a mirror — Gitea converts a repository to a mirror only that way — are the maintainer's.

49. Read a list whose bullet or ordered delimiter changes as two lists, in both readers, and retire !adf:listBreak.

Lands after item 43. Today - a then + b, or 1. then 1), reads as one list; CommonMark reads two (spec examples 301 and 302). Goal 5 settles it for the lossless flavour too: CommonMark spells adjacent lists by changing the marker, so adfToMarkdown and adfToPlainMarkdown alternate - and * between adjacent bullet lists and . and ) between adjacent ordered lists, and !adf:listBreak retires. * is the maintainer's pick (2026-10-04). Breaking, so it ships beside item 43: MIGRATION.md's Readings and Spellings tables gain their rows. Examples 301 and 302 lose their pending exceptions, and the spelling leaves the README's "Four CommonMark spellings" bullet, which counts one fewer.

61. Have the emitter ask the inline reader how a line reads back, in place of line-escaping.ts predicting it.

The comprehension panel's worst place: escapeClaims and escapeClosedRuns re-implement the reader's view — flanking, code-span closers, link-definition openings, highlight flanking — and only the property tests catch drift. It caps the panel's Locality score. The replacement runs in time linear in the line: today mergesWithSyntax, touchesSyntax and closesHeading rescan a run of one character from each of its characters.

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 example 540); lowercasing and then uppercasing folds it. Breaking, so it ships beside item 43: 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.

60. Collect the questions parse/ asks emit/ into one named module.

parse/ asks emit/ through commonMarkSpelling and openingLinkTakesDirective, each imported from where it happens to live. One module naming the questions keeps docs/decisions.md §The source parts by ADF and format true as they grow.

59. Group the directive grammar into src/markdown/directive/, move Read<T> to result.ts, and move Flavour to markdown/flavour.ts.

Eight directive files sit across three directories, and the markdown/ root holds 14 entries. HTML needs Read<T> and Flavour out of the plain flavour and markdown/; it needs the carry and the mark spellings too, which item 7 moves where it learns what HTML shares.

53. Put a block's marks spelling to 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.

62. Move the plain flavour's reading out of parse/ and its writing out of emit/ into markdown/plain/, so emit/ no longer imports plain/.

Reading: the alert and task-marker reads in parse/markdown-to-adf.ts, the mintTaskIds call and the inlineLeaves use. Writing: spellPlainBlock, quotedUnder, tryTaskList and taskBlocks in emit/adf-to-markdown.ts. And highlightDelimiter and highlightFlanking move out of plain/conventions.ts, which emit/ imports them from; item 59 moves Flavour. Every comprehension reader on 2026-10-03 named the plain flavour's spread across three directories.

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 link mark, so a link whose text is empty takes its URL as its text (the maintainer, 2026-10-04): [](/url) reads as /url linked to /url. []() has no URL to show; the chunk settles it by Goals 4 and 7. Breaking, so it ships beside item 43: 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.

47. Open the README with what the package is, what it does and for whom.

It opens with the pre-launch rationale — Atlassian's REST APIs, pf-editor-service/convert being decommissioned, a link to JRACLOUD-77436. The background goes entirely, no endpoint, ticket or "why" note left. The badges are npm's version and the Gitea Actions status. The README names the lossy pair, adfToPlainMarkdown and plainMarkdownToAdf, and the flavours it writes and reads — GitHub Flavored Markdown's alerts and task lists, Obsidian Flavored Markdown's callouts — so a search for any of these names finds the package.

73. Announce the package where someone needing an ADF converter already reads: JRACLOUD-77436, the Atlassian developer community and Stack Overflow's ADF-to-markdown questions.

Lands after 0.2.0 is published and items 71 and 72 land, so each post links the GitHub repository and the comparison. We draft each post for the place it goes; the maintainer posts them. Found while planning discoverability, 2026-10-04.

38. Spell a lone surrogate in a text node so it survives a UTF-8 encode.

adfToMarkdown emits it verbatim, so markdown stored as UTF-8 reads back U+FFFD; attribute values already escape it.

48. Keep the release path publishing past npm's bypass-2FA token retirement.

Lands after 2027-01-01, or after a release run fails on the token, whichever comes first: the maintainer chose on 2026-10-02 to wait and see whether the retirement bites. It holds back no release: when the rest of its release is done, it moves to the next. 0.1.0 published only once the npm token carried Bypass 2FA: the account requiring no 2FA on writes was not enough, and npm answered EOTP until the token itself bypassed. npm retires bypass-2FA tokens for direct publishing around January 2027, leaving them npm stage publish, which a maintainer approves with 2FA; its replacement — trusted publishing over OIDC — supports GitHub-hosted Actions, GitLab.com's shared runners and CircleCI's cloud, self-hosted runners planned without a date. Revisit: whether npm has added Gitea or self-hosted OIDC, and otherwise whether the release moves to the staged publish — which fits badly with publish-on-merge, and is the maintainer's trade to weigh. The Goals and G cells are provisional: no README goal covers the release path.

46. Publish the bundle size in the README, failing the release pipeline when it drifts.

Lands after item 7, which changes it. The quantity is what a consumer downloads and loads: the tarball npm pack produces, its unpacked dist, and the built JavaScript minified + gzipped — the figure the competitors advertise (marklassian's "12kb") and the only apples-to-apples one, since ours ships tsc's unminified output and no minifier yet (decide here whether to minify for the build or report the unminified gzip). The figure lands in README §The package beside the "no runtime dependencies" claim. Measured today, unminified: tarball 60.4 kB, unpacked 221.5 kB, JS gzipped 45.6 kB.

34. Read emphasis flanking by the whole character beside an astral symbol.

Check whether line-escaping.ts's charAt and the parser's flanking read one UTF-16 unit beside an astral symbol — a lone surrogate is neither punctuation nor symbol, where CommonMark reads 😀 as punctuation — and, where they do, read the code point, with a fixture per direction.

31. Make the branch-coverage figure repeat across runs of an unchanged tree.

Three Node test legs over one unchanged tree reported emit/inline-line.ts at 95.83%, 96.23% and 96.23%, and the total at 98.80%, 98.84% and 98.84% (2026-09-21). --experimental-test-coverage counts branches off V8's own coverage, which the runner's parallel files and V8's optimization make run-dependent, so the number the floor is read against is not the code's alone. The floor of 98 holds today on 0.8 points of slack and docs/decisions.md §The coverage floors says it only ever moves upward, so the first raise to the measured figure reddens a run that changed nothing. Make the measurement repeatable, or state the number the floor may be raised to and why it is not the measured one.

42. Trim a text leaf's trailing blanks in linear time.

plain/inline-reduction.ts's leafEdges finds the trail with an unanchored /[ \t]*$/, quadratic in a run of blanks inside one leaf: a paragraph of a, 80 000 spaces, b takes 6.5 s in adfToPlainMarkdown. Scan backward, as the expand title's trim does.

33. Emit a line in time linear in its mark runs, in adfToMarkdown and adfToPlainMarkdown.

adfToMarkdown spends 23 s on one paragraph of 2000 × un plus **-r**: each run its flanking cannot spell re-emits the whole line before riding the carry, quadratic in the runs, and the plain reduction's spellableLine drops one mark per re-emit the same way. An inline node whose attributes no spelling writes re-emits the line the same way before riding the carry. Make all three linear.

56. Give each piece of blocks.ts's block-walk state and inline-content.ts's Scan 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. inline-content.ts's Scan has the same shape: about fifteen functions write pending, pieces, deactivatedBefore and openingSpellableLink and return void, and parseInlineContent reads a flag scanInline leaves on it. The same shape recurs in mintTaskIds (plain/task-ids.ts), which writes attrs on the document readDocument built; takeFallback (emit/inline-line.ts), which writes the record the next lineSegments pass reads; writeUnpaired, markPairings and markHighlights (inline-content.ts), whose order changes the output; and nestIn (plain/adf-to-plain-markdown.ts), which pushes into its caller's array.

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 proposes both (the maintainer, 2026-10-02), and they land in the README's ## Goals and ## Audience with the CLI.