The edge band and window.panel.dock to an outer edge anchor on the canvas only where the edge area narrows to the canvas column, else on a panel that is not protected, so tree order no longer decides the frame. The renderer tries the drop and offers the band only when it wraps the panels the band shows. A top (bottom) edge also narrows a split of rows to its first (last) row when only that row holds the canvas. to_actions takes the permissions, and FloatSpace records side_edge_zones.
With SidePolicy::OneSlot a drop can reveal a pane the frame hides, so the landing slot is
worked out from every open pane. A side pane is no wider than the room the panes outside it
leave, as the preview assumes; egui placed a wider one partly beyond the host's edge.
* docking: drop targets follow the pointer over the whole group
Outside side_edge_zones, a group body's outer third on each side docks
beside it and the middle joins as a tab; at a corner the relatively
nearer side wins, and a group that takes no tabs maps its whole body to
its edges. Dock-column groups get the same zones: top and bottom stack,
left and right make a column. Border bands (seam, column edge, new
dock, root edge) are 28 pt and reach only 4 pt into a tab strip that
takes tabs.
Previews are a drop_line_width line along the edge a drop docks to, or
the whole group's outline with a light fill for a tab join, for tab and
group drags alike. side_edge_zones keeps its 40 pt picture-frame zones,
overlay and previews (SIDE_EDGE_ZONE_DEPTH).
DockStyle: side_zone_fraction replaces zone_depth; new column_edge_band
and drop_line_width; edge_band, seam_band and new_dock_band default to
28.
* docking: document the drop zones, bands and previews
* docking: dock groups and the outer edge band keep the split-tree sizes with side edge zones
* docking: drop zones for hosts that resolve drops, lines kept in view, stripless dock groups
resolve_drop_by and DockStyle::zone_rule give hosts the dock area's zones and previews.
A stack line on an edge scrolled out of a dock column shows at the view's edge, and a dock
group without a tab strip takes its zones from the top of its body. A style needs a drop
line wider than zero and a side zone fraction of at most one half.
It came in with the first proofing commit (bcbe09df8) and has never had a caller; `#[allow(dead_code)]` kept the warning quiet. Because it is `pub(crate)`, nothing outside wordcraft-proof can reach it.
SonarCloud's analysis of main at b4da59f0c lists 23 closures such as `|v| v.as_slice()` that hand their argument straight to one method. This passes the method itself: `Vec::as_slice`, `str::to_ascii_lowercase`, `f64::total_cmp` and the like, all forms the repo already uses elsewhere. Behaviour is unchanged.
Five stay. The four in crates/proof/src/es/mod.rs sit in code that storytold/wordcraft#4191 replaces, and its replacement already uses `PoisonError::into_inner`. In crates/engine/src/io/safe_save.rs the path would be `<dyn std::error::Error + Send + Sync>::is::<Guided>`, which reads worse than `|inner| inner.is::<Guided>()`.
Review follow-up. When two vendor namespaces trade prefixes, both prefixes stay
declared, so a name written with one of them as an element or attribute name
(`<ns0000:Feature/>`, `ns0000:flag="1"`) still parsed and resolved after the
root declarations changed, but to the other namespace. respell only rewrites
`name`/`constrained` values and QName values, so such a name would not follow.
respell now returns None when any element or attribute is named in a namespace
whose prefix changes. Names in a namespace that keeps its prefix do not block
the respelling, and the real Brother tickets (vendor names only in `name`
attributes and QName values) are respelled as before.
Tests: regression cases for element and attribute names in a swap and in a
prefix chain, and a control for names in an untouched namespace. Each ticket
is asserted well-formed with every prefix resolving, so the case is refused
for the name, not for a parse failure. The refusal tests fail with the guard
disabled.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FQ5PzHVUcntEHU97zniM5U
* docking: blocking input hands the keyboard from a native window to the main window
* docking: the keyboard goes to the main window only if a native window still has it after a short wait
Review of #307:
- respell rewrites the root's namespace declarations, so an element below the
root that declares either prefix of a rename again (ns0000 or brpsk) would make
a name resolve to another namespace once the prefixes change. It now returns
None for such a ticket, and a test covers both prefixes.
- The vendor-element test replaced only an opening tag, left the closing tag
behind and so never parsed: it passed because respell refused the broken XML,
not because of what it was meant to check. Every refusal fixture is now
asserted well-formed, and the element is replaced whole. Doing so showed the
element guard was redundant: an element or attribute name in a renamed
namespace leaves an undeclared prefix, which the final parse of the result
already refuses. The guard is removed and the test now exercises that check.
- Doc comments on the new constants and tests.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FQ5PzHVUcntEHU97znim5u
Windows' PTMergeAndValidatePrintTicket names a vendor namespace with a generated
prefix (ns0000) as soon as the merged ticket uses a name from it. driver_ticket
merges the driver's own paper option from its capabilities, which carries vendor
properties, so a Brother HL-L2390DW's ticket came back with ns0000 where the
driver's own default ticket says brpsk. XML-wise the two are the same; the
driver's print pipeline reads the prefix as text. The job was accepted and then
ID2D1PrintControl::AddPage (or Close, in older builds) failed with 0x80040003,
reported by winprint as "Render error" / "Failed to close print control: No such
interface supported".
The same ticket sent raw with brpsk prints; with only the prefix changed to
ns0000 it fails the same way, so the prefix is the whole difference.
driver_ticket now re-spells the validated ticket with the prefixes of the driver's
own default ticket (an empty ticket, validated), through ticket::respell: pure
XML, tested on every platform, and a no-op unless a vendor namespace is spelled
differently. It rewrites only the root declarations and the vendor names in name
/ constrained attributes and QName values, handles two namespaces trading
prefixes, and leaves the ticket as it is when a rename could be ambiguous or leave
a prefix undeclared.
Verified on a Brother HL-L2390DW: the job spools, and pdfcraft head's doc_print
that failed at AddPage prints.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FQ5PzHVUcntEHU97znim5u
* docking: route app drags in panel bodies across native windows
While an app's drag that began in a panel body lasts, the owner window's
input hook sends pointer moves over another of the app's windows to that
window in its coordinates (PointerMoved, PointerGone, and the primary
release), and the owner sees the pointer gone there. Drop sites in the
target window see hover and release with no app changes.
* docking: wrap the body-drag docs
* docking: a stopped body-drag route takes back the events it queued
* docking: route only body drags that carry a payload, and hold the owner's release
A body drag reaches other windows only with an egui drag-and-drop
payload or when the app calls docking::route_app_drag that frame or the
last; sliders, scrollbars and other plain drags keep the owner's
pointer. When the release goes to another window, the owner's own
release waits for its next input, so a target drawn before the owner
reads the release while egui's payload still exists. A stop never
withdraws that held release.
* docking: routed and held events come before a window's own input
* docking: a held release waits for the main window's next pass, with the pointer events after it
* docking: the owner sees the pointer go in the pass its release is held
The driver's PrintCapabilities tell a driver that dropped a requested value
from one that never had the feature. Grayscale on a driver with no colour
option and uncollated copies on one that can't collate are accepted, like
one-sided on a queue without duplex. Collated copies on a driver that can't
collate are refused with their own message. The capabilities are read only
when the strict check fails.
* docking: native windows ignore input while the app shows a modal
DockArea::input_blocked disables each native floating window's widgets, drops its pointer, key, text and scroll input, covers it with DockStyle::blocked_dim over a layer that takes the pointer, offers no drop targets, forwards no events and cancels its OS close request. Turning it on cancels a drag in progress.
* docking: a blocked native window also drops hovered and dropped files
* docking: blocked palettes drop to the normal window level
* docking: a blocked native window still forwards key releases
* docking: while input is blocked, activation raises the main window last
* docking: blocked palettes stay shown, and the main window comes in front on the app's return
* docking: normal floating windows come forward when the app is activated
* docking: drop windows that became palettes from the raise queue
* docking: the main window comes forward with the normal windows, in focus order
* docking: replacing the workspace cancels a drag in progress
A dock area keeps the layout it drew while a drag is in progress, with
the actions it returned. When the next layout is neither that one nor
that one with the actions applied, the app replaced or changed it, and
the area cancels its tab, group, column, window bar, native bar, divider
and cross-window drags itself. cancel_drag also ends a native window's
bar drag.
* docking: replacing the layout also ends a divider drag in a native window
libjpeg's encoder drops trailing zero bytes, so a flat tail of a few thousand blocks decodes from zeros supplied at EOI. Damage is data that runs to the end of the file or a lost restart marker. whitetail-ari.jpg (cjpeg -arithmetic) and its Huffman transcode cover it.
A dictionary is read outside the lock. If the dictionaries folder changed
while it was being read (File › Options › Proofing), the reload emptied
the slot first and the late load then filled it with the old folder's
dictionary. It stayed until the next reload. Every language's copy of
the loader had this window; the shared slot now closes it in one place.
The slot counts reloads. A load that started before a reload still
answers the check that asked for it, but it isn't kept: the next check
reads the new folder.
Found by CodeRabbit on storytold/wordcraft#4191. Tests: a_load_from_before_a_reload_is_not_kept;
not counting reloads fails it. Proof: 288 tests.
The suggestion rounds don't run in rising cost: a MAP diacritic drop (11)
comes before a REP confusion (10, 9 for Russian). The engine checked each
candidate once, at the first cost it was offered, so the same word found
again more cheaply kept the dearer cost and could rank one place too low.
Czech's engine always did this, and the merged engine took it over.
The candidate set is now a map to the cheapest cost offered so far, and
an offer that is no cheaper is skipped as before.
Found by CodeRabbit on storytold/wordcraft#4191. Tests: a_cheaper_later_offer_wins; skipping
every repeat offer again fails it. Proof: 288 tests.
* docking: an input hook and dropped files for every native window
eframe runs raw_input_hook only for the root viewport, and each native window's
dropped files arrive only in that window's input. DockArea::native_input_hook runs
an app hook on each native window's RawInput through an egui plugin, before egui
reads it, and DockOutput::window_files reports files hovered or dropped on native
windows with the window and the panel under the pointer. The test host takes
hovered and dropped files per window.
* docking: a discarded pass reports no hovered files; share hook state through a mutex
* docking: one file report per native window when its pass is discarded
* docking: panels closed one by one reopen into one group
Showing a hidden panel whose anchor is hidden too follows the chain of
hidden anchors to its far end, so a floating location no longer opens a
second window. The hidden panels the restore passed then anchor back on
the panel before them in the chain, on the recorded side, so reopening
them joins the restored panel instead of rebuilding their group or
window. A restored tab whose next tab is not in the anchor's group goes
last.
* docking: a panel beside an edge frame keeps its own location on reopen
* docking: a panel reopened through a hidden neighbour keeps its own group state; the chain stops reversing at an edge frame
* docking: viewer panels float as normal windows
DockContent::window_kind marks panels as palettes (the default) or normal
windows. A native floating window holding any Normal panel gets the normal
window level and stays shown while the app is in the background; drags
treat it as stacked with the main window by focus.
* docs: a normal window behind the main window takes drags outside it
* docking: tabs shrink before they overflow
Tabs in a narrow strip shrink, widest first, down to DockStyle::min_tab_width (54 pt at the
default font), cutting their labels with "…" and naming them in a tooltip. Only tabs that do
not fit at that width go to the overflow menu.
* docking: the drag bar narrows once the strip is narrower than the header plus the bar
* docking: floating windows stay in the app window on Wayland
The compositor reports no window positions, so native windows could not
dock again by drag or reopen where they were. The dock area detects
Wayland (WAYLAND_DISPLAY or WAYLAND_SOCKET set and a main window that never
reported its outer rect) and draws floats in-window, keeping their native
rects. DockArea::native_windows(bool) overrides the choice.
* docking: a missing DISPLAY also counts as a possible Wayland backend
* Revert "docking: a missing DISPLAY also counts as a possible Wayland backend"
This reverts commit a804afb1d3.
* docking: the Wayland session is a pure function of the environment; without a Wayland variable only a real window counts
aiongg
skylartaylor
aiongg
VoiceLessQ
Claude