Corrects two probed claims that measurement contradicted: a hard page break
does *not* swallow the block's space above (only an automatic one does), and
the header band is max(declared height, content + gap) rather than a line
count. Adds the chapter-numbering model, the mirrored master pair, and the
harness ceiling a CSS-drawn label is.
A break between spanned rows stretched the cell across the gap: the
merged table drew over the page gap and the next page's header. A
group too tall for a page still breaks per row. An in-flow text box
took the same treatment for floats — CSS lays a block box over one
instead of beside it, which had the vertical frame covering its
neighbour. Both match LibreOffice's render of the same document.
The always-on absolutely-positioned ¶ pseudo made Chromium give up
balancing the multicol: every block stacked into column 1, while Word
and LibreOffice balance the section end. Static inside .columns-node
(zero width, line breaking unchanged); past a trailing BR the mark is
dropped there so an empty paragraph stays one line.
View ▸ Split (Ctrl+Alt+S): a second live view of the editor's state, stacked
below the first and editable the same way. The panes share a width, so the
second renders the pagination the first measures — pageBreaks, columnsFlow
and tabStops skip the pass on a split pane rather than run it twice.
Every widget decoration now builds its DOM in a function: two views render
one decoration set, and a shared node is taken back and forth between the
two documents until the tab locks up (measured on a two-page document).
A section narrower than the reserved sheet now splits the difference
between its insets instead of hanging it all on the right, and every
per-page layer draws from the box's own left — sheet, header/footer, page
border and line numbers. The ribbon's Size menu gains the same per-section
block Orientation already had.
The page grid was one repeating cycle, so every page had to be the same
size. It is now a table of height runs the placement pass grows as each
section's first page becomes known, and the sheets are one box per page
instead of a repeating gradient — .paper reserves the widest section's
paper, a narrower one takes the difference as a right inset.
Layout ▸ Orientation offers "this section"; the size stays document-wide
in the ribbon but round-trips per section from a file.
ODF style:page-usage="mirrored" and Word's w:mirrorMargins make the
declared left/right the inner/outer pair, which an even page swaps.
Read, rendered and written back in both formats; the body rides the
per-block inset a section's own margins already use, the header and
footer band moves with it. Measured on a mirrored two-column file:
its page 2 went from 12.7-184.1mm to 31.8-203.2, which is LibreOffice's
to 0.1mm, and its running head with it.
Three measurement bugs in columnsFlow, each worth a page on its own:
rects became lines by their top, so a raised run counted as a line of
its own (192 for 111); the split probed `left + 1`, which is a line's
*end* right-to-left, so it kept a word too many and the flow then shaved
one line per pass; and one median advance under-measured a paragraph
whose lines differ in height (2834px real against 2453 estimated), by
enough to budget 74 lines into room for 64. The chain used to flip
between one fragment and two forever; it now settles after one split.
Collapsed borders paint a shared edge once, so the spacer <tr> bridging
the gap took the row separator with it and left one fragment open —
visible in a fixture as tables missing their frame on a page. The spacer now
draws the line LibreOffice draws there: the separator the break falls on,
or the table's own box where the rows carry none. Absolutely positioned,
since a collapsed border on the spacer shifts every row below by half.
.tiptap's padding draws one pair of side margins for the whole document, so a
title page declaring 3.18cm left sat 3.85mm off. The difference now rides a
--sec-inset-left/-right node decoration on the section's top-level blocks, and
everything that writes a block margin adds it in — an indent, a named style's
and a table's alignment all compose with it instead of losing to it.
ODF style:may-break-between-rows="false" — probed: LibreOffice leaves four
lines of a page empty rather than split such a table, and splits the identical
one when the attribute says "true". pageBreaks now emits a single atomic leaf
for the wrapper unless the table is taller than a page, where the rule cannot
be met. Imported, exported and round-tripped; Word has no equivalent.
A heading only moved down when its successor's *first* line no longer fit. But
widow-orphan control carries a lone first line over anyway, so the heading was
left stranded at the page foot — LibreOffice moves the pair. The fit test now
asks for MIN_KEPT_LINES of the successor.
Only the ordinary-block push site added the space the page-top rule had
dropped above it; the index, text box, columns and table-row sites subtracted
the spacers alone, so each of those measured too high by whatever the pages
above had swallowed and got an equally oversized spacer — a fixture's figure
frames landed 16mm down their page. One naturalTopOf() now serves all six.
Both probed against LibreOffice while chasing the render-parity corpus:
it compresses word spacing to fit one more word on a justified line
(CSS only stretches), and it drops the space above a block that a page
break put at a page top.
An as-character image is now a line box in getLineRects, so a paragraph of
stacked images breaks between them instead of overflowing the page. Only a
floated image keeps its paragraph atomic, and the leaf's height reaches down
to the lowest float — a float hangs out of the paragraph's box, so
offsetHeight alone let the image walk off the page foot. prevPlacementsKey
damps the ping-pong that overhang causes between two layouts.
The root file is read at every session start and had grown to 87 KB, so every
session paid for knowledge it did not use. Move the per-directory detail into
nested CLAUDE.md files, which load when a file in that directory is touched,
and the five deep-dive topics into docs/. The root keeps the commands, the hard
rules and an index of where the rest lives.
Content moved verbatim — the probed Word/LibreOffice behaviour, sentinel order
and shared constants are what the code cannot say. CI fails the root over 120
lines, since a prose budget alone is what let it grow.
Steffen Becker