The browser hit-tests a behind-text frame last of all — the page and every
paragraph over it come first — so a cover picture could not be clicked at all:
the caret landed in the text as if it were not there. A mousedown over one now
goes to the frame's own node view, which selects and drags it as any other
frame; text painted over the point still wins, as the caret does in both
products. A box's node view mirrors renderHTML's data-wrap for it, as an
image's already did.
Claude-Session: https://claude.ai/code/session_01Wzs2qG6Y5qm7mKmJHfu44G
Word's positionV relativeFrom="page" and ODF's style:vertical-rel="page" — how
a cover page's own blocks are placed — were dropped by both importers, so every
such frame sat at its anchor paragraph instead: a title page's coloured blocks
landed wherever the text happened to end. The offset now rides `wrapFromPage`
on the frame, the node view resolves the page the anchor lands on from the page
grid, and both exports write the relation back.
A shape behind the text also goes under a picture behind it (z-index -2, the
sheet moving to -3): probed against LibreOffice, which paints a cover picture
over a shape whatever stacking order the file gives them.
Claude-Session: https://claude.ai/code/session_01Wzs2qG6Y5qm7mKmJHfu44G
A divider line is written as a frame of zero height, and the node view handed
that straight to the SVG's viewBox and viewport — a zero-high viewport turns
rendering off altogether, so the line was simply absent. The viewport is now
at least the stroke, which it overflows by half as in any flat frame.
Claude-Session: https://claude.ai/code/session_01Wzs2qG6Y5qm7mKmJHfu44G
Only <a:srgbClr> was read, so every shape naming a theme slot — which is what
a Word template's own colours are — came back unfilled and its rules black or
gone: a cover page lost both its coloured blocks and its divider lines. The
scheme is read from theme1.xml by slot, with the text/background aliases
Word's default mapping gives them, and the lumMod/lumOff/tint/shade modifiers
applied per channel.
Claude-Session: https://claude.ai/code/session_01Wzs2qG6Y5qm7mKmJHfu44G
Both importers matched a file's format against the seven date and four time
formats the picker offers and kept anything else as the text the producer had
cached — a Word template's `DATE \@ "MMMM d"` then froze at the day it was
written. An unlisted format now rides its own DOCX picture as the key:
`findFormat` parses it, so display, ODF number style and DOCX picture are
derived from it as they are for a catalog entry, and it re-exports unchanged.
Claude-Session: https://claude.ai/code/session_01Wzs2qG6Y5qm7mKmJHfu44G
A .docx zone holding a table lost everything in it: hfParagraphs walked only
the direct w:p children, so the one-cell table every Word template draws its
rule line with left the zone empty. Its cells' paragraphs now become lines of
the zone, as they already do on the ODF side, and the cell's own borders and
shading (its w:tcBorders over the table's w:tblBorders) become the collapsed
paragraph's box — which is what draws the rule.
Claude-Session: https://claude.ai/code/session_01Wzs2qG6Y5qm7mKmJHfu44G
Everything since 0.2.0 on 2026-09-06: the run of load and typing latency work,
each tab editing its own document, the format prompt on a first save, the
unsaved-close warning, and two fixes.
The check was queued in idle time, which on a long document falls between two
pagination passes: its decoration swap redraws the view, and the pass behind it
pays for the redraw. It now waits until no pass is running or queued
(`isPaginating`), so an opened file paginates first and is checked after.
Measured on a 460-page file: settled 5.99 s → 5.10 s (interleaved A/B, median
of seven; a second set of nine gave 5.76 → 5.34). Spacer, index and text-box
geometry byte-identical on five files.
A long document placed a zone box on every page, and the browser laid all of
them out again on every pagination pass. Only a window around the current page
carries its zones now; a capture or a printout opens it to the whole document
first, so the PDF export and Cmd+P are unchanged.
Measured on a 460-page file: settled 6.17 s → 5.27 s (interleaved A/B, median
of seven). Spacer, index and text-box geometry byte-identical on five files.
Both indexes and the cross-references each read it for themselves, twice per round in
the index's case: 59 readings of it over a load where 37 do. The round hands one to
every reader instead, so they all resolve their pages against the same grid, and the
lazy holders they kept go with it. No measurable time — the style pass a reading forces
is owed to the geometry read that follows it either way.
Reading an added node's position computes the style of the whole document, and the band
rebuilds its galleries whenever a document is opened: 300 ms of that during a load, for
40 gallery buttons and no panel at all. A click or a key in the band still classifies its
records at once; anything else is looked at a task later, where the style is there for
free. A 460-page file is on screen 270 ms earlier, every panel pinned as before.
Checking every word of a 460-page file costs half a second, and at mount it was paid
before the document was on screen. The pass now waits for idle time (2 s timeout), as
does the one a language or dictionary change asks for; the check after an edit is
unchanged. The file is named 200 ms earlier and settles 700 ms earlier, with the same
page count, spacers, page numbers and references.
coordsAtPos called straight from the selection handler lays the whole document out
before anything is painted — half a second on a 460-page file, and the writes of the
load that follow it throw the result away. The read is queued as its own task and
coalesced, so a run of selection changes costs one.
Every field resolving against the settled layout — each index's page numbers, every
cross-reference — registers with scheduleFieldRound instead of taking a frame of its
own. A round measures for all of them before any of them writes, since a write between
two measurements lays the whole document out again, and their cached results ride one
transaction: a dispatch of its own costs a view update over the document, 55 ms per
index on a 460-page file. That file settles about 220 ms earlier, with every spacer,
page number and reference exactly where it was on five documents.
A single ResizeObserver refits the frames, reading each rotor before any wrapper is
written: a per-view one arrives in a callback of its own, so the browser laid the
document out again between two frames. pm-pagecount is announced only where the layout
differs from the last announcement — every reader re-reads the whole document from it and
writes back, which is another pass. Both node views also leave the DOM alone when update()
brings the same attrs object. A 460-page file holding 450 frames and two indexes settles
2.4 s earlier and is named 200 ms earlier, with every spacer and page number where it was.
The format pre-pass and the importer read the same inflated archive, where each
unzipped the file for itself, and a picture is encoded through the browser's
Uint8Array.toBase64, the chunked loop standing in where there is none.
On a 14 MB file holding 600 pictures that takes 250 ms of inflating and 285 ms of
encoding off the open: it lands 480 ms earlier and settles 240 ms earlier, and a
small illustrated file lands 170 ms earlier.
Cross-references resolve in one batch per pass, the index reads every leader gap
before filling one, and a zone's tabs lay out with the other zones' — and only
where a field asks for the page count or the chapter map.
List marker and list style decorations are rebuilt only for a transaction
reaching into a list; the unsaved dot compares documents structurally, and the
autosave serializes when it writes rather than per keystroke.
The columns flow applies every op it can judge in a pass, so a section pages in
one pass. Two long files that took 33 s and 20 s now settle in 9 s and 2 s.
The image NodeView reads the page grid from the view ProseMirror hands it at
construction; editor.view is a throwing proxy until the view is built, and a
saved document with such a frame is passed in exactly then.
The dom leg loads its band document through the autosave, the path a reload
takes. The layout baseline is re-recorded for the two showcase files whose
second page was reworked after the baseline was taken.
A `:has()` left of `+` in editor.css made Chromium restyle every element
under .tiptap on each inserted child, so the band-clearing rule keys on a
data-wrap-band attribute the imageLine plugin writes. getLineRects reuses
one Range instead of leaving hundreds alive. TipTap's Placeholder, which
hit-tested the viewport on every keystroke, is replaced by placeholder.ts.
docPosBeforeElement is memoised per pass. At 124 pages the pass after a
split drops from 410 to 150 ms and the split keystroke from 219 to 50 ms.
The dirty ranges ride the plugin state and are mapped through every
transaction; the check after a pause swaps only those textblocks'
squiggles, and the whole document is checked on a language or
dictionary change. The pause after a keystroke in a 124-page document
drops from 363 ms to 11 ms (Chromium) and from 1.4 s to 17 ms (Firefox).
A unit test drives the recheck; the dom run asserts the pause is free.
An edit's pass waits for 300 ms of quiet instead of running between keys, so
while the typing goes on only the keystroke itself costs. When the quiet comes,
edits that only put text into one textblock and left the content's bottom where
the last pass had it skip the pass: the block is as tall as before, so nothing
below it moved. A break inside the block still runs it. Measured at the top of
a 124-page document: 41 ms a keystroke in Chromium, no pass for a letter typed
and taken back. The dom run holds both.
The page-break decorations are mapped through every transaction and each
spacer widget carries a key, so a pass that lands the same spacers keeps
their DOM instead of rebuilding all of them. An edit within 150 ms of the
last pass waits for a pause in the typing. Measured at the top of a
124-page document: 1.5 s a keystroke before, 40 ms after. The dom run
holds the budget.
The unload warning followed the changes a file had not got yet; a document
that never was a file got none, on the grounds that the autosave keeps it.
It now warns as soon as such a document has content — the same emptiness
test the confirm before a new document uses.
The format has to be settled before the bytes are built, so a save that goes
the download route cannot become .docx by renaming it in the browser's dialog.
Save on a document with no file behind it now opens SaveFormatDialog and hands
the choice to Save As; from then on it saves in that format without asking.
Every document-scoped key now goes through docKey(), which hangs it off the
tab's document id; the first document keeps the empty id, so existing browser
copies read back unchanged. A fresh tab takes up the most recently used
document no live tab holds, else starts a new one — one signed timestamp per
document (edentext-live@id) says which, written only by the tab holding it.
Snapshots and embedded fonts get a database per document. The picture store
stays shared and content-addressed, so its sweep now drops only keys no stored
document names any more, instead of the open document's own.
The shot follows the chapter heading rather than a fixed page, so it keeps both
page designs in view if the break ever moves. The gallery notes why the folio
and the document page differ there.
The book carries the chapter and the title towards the outer edge and its page
number centred at the foot; the newsletter opens with the masthead and runs a
head with a rule from the second page on. The thesis keeps chapter and number
in the head, so the three read as three page designs.
The collector walks node_modules, so a file copied into public/ had no entry.
The copy has no copyright line of its own; the text is upstream's LICENSE
section for count.js, with the project's copyright holder.
openApp() passes page options through, so the showcase run sets
deviceScaleFactor 2 — same CSS viewport and layout as the tests, 2800x2000 px
images that stay sharp where the README scales them down.
The MathML operator dictionary makes ( [ { | stretchy, so a plain bracket grew
to the tallest thing in its row — next to a sum with a fraction, C(b) came out
in full row height. Only a fence node emits stretchy brackets now.
GoatCounter: no cookies, no IP or user-agent stored, EU-hosted. count.js is
vendored into public/ so no file loads from a foreign origin — only the count
request itself leaves. The /count endpoint is documented as staying
compatible, so the copy does not go stale.
Everything since the 0.1.0 launch on 2026-08-17: named list styles, chapter
numbering, the letter templates and DIN 5008 marks, password protection in both
formats, comment threads with margin balloons, the book layout, the inline text
box, and the run of Word / LibreOffice interoperability, round-trip and page-layout
fixes behind the parity corpus, the fuzzed export options and the browser runs.
0.1.0 keeps its own date and scope; its open-items list drops password protection,
which now ships.
libreoffice-writer alone has no Math component, so Writer dropped every formula
it imported — Word's own OMML included — and the fuzz leg saw the loss.
node 20 cannot import tests/schemaValidate.ts, which the monkey run does; 24
strips the types. node 20 is end of life anyway.
A body sectPr without w:type means nextPage, so every section break started a
page the editor and the ODT export never make — the DOCX import then read that
break back as fo:break-before. It is continuous now unless the section has
zones or a page setup of its own, which is exactly when odt.ts mints it a
master page. Measured on LibreOffice 24.2: the kitchen sink paginates to five
pages in both formats again.
A shot names a ribbon tab and, optionally, where the caret goes: the tables
one lands in a cell, so the contextual Table Layout tab is up. The floating
table toolbar is hidden while shooting — it covered the table the ribbon is
already showing. `data-tab` on the tab buttons is what the script clicks.
`.paper-scaler` reserves page plus strip and centres the pair, so a document
with margin balloons sits left of centre — the ruler, sized from the page
alone, centred over a narrower box and hung to the right of the page it
measures. It now reserves the same strip as `padding-right`.
Inside `.columns-node` the space above a block is the margin and the space
below the padding: an engine truncates a margin at a column break but keeps
padding, so the columns of a section now begin at the same height and the
add model's two spacings still sum. `spaceAfter` carries `--space-after`
beside its margin, from the attr and from a named style, for the swap.
The 1x1 fixture carried an IDAT from a different image: its CRC fails and the
zlib stream is truncated. Chromium and WebKit render it anyway, Firefox reports
"Image corrupt or truncated", and the same bytes sat in eleven fixtures.
Header and footer go first, each with its own ops, and a variant flag is
flipped while a zone is open. The round trip compares a list's drawn kind and
marker rather than its node type and style name — a list that loses its name to
a formatted marker carries the style's own char — and skips the marks of an
inline atom, which no file keeps. A caret is snapped to a text position, a
swallowed File-menu click is retried, and WebKit's converging ResizeObserver
round is no page error. A new corpus pair brings all six zones through the
import path; the baseline holds it and a re-recorded showcase entry.
The two variant flags (a first page of its own, separate even pages) and the
header/footer distances were reachable only from the expanded toolbar. The
ribbon mode now has them in an Options menu beside the zone buttons.
A page break on a list paragraph is kept now, and a heading after a list item's
own paragraph reads back as a heading — the numbered chapter heading that *is*
the item stays a paragraph, which is how the editor numbers it. A cell
paragraph's break is dropped in both formats, since neither word processor
honours one there. The .docx importer strips exactly the one tab a note's text
carries, and a heading in a cell is measured against its own level's size
rather than the default style's, which made it carry that size as formatting.
LibreOffice keeps the two vertical margins in one item, so a heading declaring
only one carries its style's partner — at the top level and in a cell. A list
item's page break, the own formatting of its blocks past the first, and a
heading nested in it now survive the .odt, and a list in a cell no longer shifts
every following body item's style. In the .docx a note's own text keeps its
formula and its ruby — both passes now cover the note parts — and a field's
cached run carries the marks around it instead of going out bare.
A page break reaches the file only from a body block or a list item — a cell,
a frame and a note body all drop it — and an indent on a list item's paragraph
is the same; a plugin sweeps both. A note anchor keeps no marks and stays out
of a text box. Sub- and superscript exclude each other, a font stack the
contenteditable leaves behind keeps its first family, and a weight Bold already
claims is no second attribute. A table's first row is all header cells or none
(the promotion carrying each cell's own attrs), a replace into a table reads the
lost column weight back from the column, and a list item holds no table.
Three regression tests for what a 5000-seed sweep turned up, each checked to
fail without its fix. A run repeating its style's own plain underline is
lossy by design, so the generator no longer puts one under a named style.
An empty first master page was swallowed by the next one's match, so the
watermark and the fold marks missed the first section entirely.
Every master of a section now writes the running zone, blank where its own
page has none — the layout takes the band off the margin for the whole
section, and an opening page lost 0.8cm of its own.
Both importers compared whether a run underlines, not how, so a red or wavy
line under a plainly underlining style came back plain.
tests/layout/baseline.json records where each page starts and what the load cost, per
engine and platform; a page starting elsewhere is a layout change the run prints page by
page, a load over three times its recorded time a regression, and LAYOUT_UPDATE=1 records
it again. The monkey draws frame sizes, wraps, alignment and column counts too, holds the
history against the edits rather than the caches the editor recomputes, and leaves it
unchecked where the columns flow rewrote the document outside it. The differences both
importers make by design are named in the comparison; unhoist moved to normalize.ts.
An .odt kept only the first paragraph of a list item; the rest ride it SEG-separated and
are split back out inside the same item. A header row the table did not ask to repeat
came back as ordinary cells, and only DOCX wrote it. An endnote's picture pointed at a
relationship the package never minted, and neither note's pictures were read through
their own part's relationships. A frame band-wrapped flush left is at no side of its own.
Blocks wrapped into columns drop a page break nothing there honours, and a cell's content
expression now says what both exporters can write into one.
Steffen Becker