The button leaves the header bar for the world, centered on the columns
the next purchase unlocks and panning with them until it leaves the view.
Those columns now take a lighter tint than the locked ground behind them,
so the boundary between the two says how far one purchase reaches.
Nothing in the code reported a scroll: the camera knew it had moved and
told only the hover. A ViewScrolledEvent says it now, which is what a
widget keeping a place in the world rather than on the screen needs. The
panels want none of it -- they are placed against the screen and stay put
as the world moves under them -- so the button is placed after their pass
and never joins the rectangles they step around.
Its face is two lines, "Expand" over the cost. That makes the cost a
display of its own, so it carries the building-block tooltip the way every
other item value does, on hover alone since the click buys the expansion.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ne3mejABZoLWKLh8fgpM3x
Hiding the layer takes the focus off whatever stood on it and leaves the
window with no focus widget at all, so the world view answered no keys
afterwards: Escape opened the menu once, closed it, and then did nothing.
Each modal now notes what held the keyboard when it opened and gives it
back on the way out -- the world view for the first of a stack, the modal
beneath for one above it, which was dead the same way.
The window's own focus guard now also catches the focus going nowhere,
rather than only landing on the wrong widget.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ne3mejABZoLWKLh8fgpM3x
Every display naming an item now shows that item's production tooltip, not
only the selection panel's chips: the header block stock, the total cost of
a building multi-selection, the remaining scrap of a debris selection, the
icons of every recipe summary, and a blueprint card's cost.
ItemTooltip moves out of the selection panel and takes an
ItemTooltipContext of its own -- an item is named all over the UI, and the
panel's context carries visuals and a debug flag no tooltip reads. A
recipe line wraps each icon with its amount so the pair can be pointed at
as one statement, and attaches tooltips only where asked: a module button
and the item tooltip itself draw the same line and stay silent.
Hover only wherever the display sits on something the player clicks, whose
click is not free to explain. Note that a recipe line on an option button
can no longer be transparent to the mouse -- Qt never looks inside a
transparent widget for the cursor -- so it relies on press propagation to
keep picking the option.
The header block stock loses world.building_blocks_tooltip, showing what
every other block icon shows instead. The Expand button keeps no tooltip:
its cost is painted into its face with nowhere to hang one, and the button
is due to be removed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ne3mejABZoLWKLh8fgpM3x
The last dialogs wearing system chrome were the three QMessageBoxes -- the
escape menu, the game-over and win screens -- and the two QInputDialogs that
name a blueprint. They are the ones the player meets at the sharpest moments
of a run, and they looked like alerts from the operating system.
MessageDialog and NameInputDialog replace them on the layer. The message
dialog names its buttons by the index addButton hands back and reports which
was clicked, with Escape standing for a button the caller nominates rather
than for a dismissal of its own -- Continue in the escape menu, as it was
explicitly set to before, and Quit on the two state screens, which is where
the reject role sent it. Neither dialog takes Q or a click outside: every
button is a decision, and a half-typed name is work in progress
(REQ-UI-DIALOG-DISMISS).
Two things the layer needed for the nested case: it now tracks the scroll
area each modal is shown in, so a modal opened with no anchor centers on the
one it was opened from rather than on the window -- which is where the
Create Blueprint prompt belongs (REQ-UI-PANEL-MODAL) -- and findFor() walks
a widget's parents to the layer, so the blueprint panel buried in the layout
dialog can open a modal without every widget in between carrying a pointer.
The three error boxes stay system dialogs on purpose: config load, config
reload, and blueprint file load all report a failure that may leave nothing
to draw on (REQ-UI-MODAL-CHROME).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ne3mejABZoLWKLh8fgpM3x
Every dialog was an OS window with a dim widget behind it in the game
window. That is why a click beside one could not reach it: Qt drops mouse
events for a window a modal blocks, so the dim -- a child of the blocked
window -- never saw the press.
ModalLayer is that dim grown up: still a child of the main window covering
its rect, but it now hosts the open modal, places it, and runs its event
loop, so it is the widget the clicks beside a modal land on. It keeps a
stack, paints one dim however many modals are open, and a hold keeps it up
while one modal hands over to the next. ModalDimOverlay and ModalDimScope
are gone; ModalLayerHold replaces the scope at the sites that still open a
system message box.
ModalDialog is what a dialog inherits in place of the window it lost: the
panel background, the drawn header with its optional close button, and the
dismissal gestures. isDismissible() governs Q and, next, the click outside
-- one predicate because both gestures reach the same dialogs. Its default
refuses, so a dialog opts in. requestDismiss() is what a gesture asks for,
and ShipLayoutDialog overrides it with the one-step ladder its Q handler
used to spell out, which is now reached by both. DialogDismiss.h folded into
the base.
The four dialogs the player meets keep their contents unchanged and lose
their title bars: the recipe/schematic selection, blueprint selection, ship
layout, and schematic choice dialogs. Placement moved with them --
placeOnSelectionPanel became getSelectionPanelAnchor, and the layer centers
on that rectangle in window coordinates, with no global mapping and no
frame height to guess at.
Two things followed from dropping OS modality: the focus guard that hands
focus back to the game world now also asks the layer whether a modal is
open, and closeEvent refuses to close the window while one is, since its
nested loop runs over widgets the window owns.
The escape menu, the game-over and win screens, and the two name dialogs are
still system dialogs; they are dimmed by a layer hold until they are
converted next.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ne3mejABZoLWKLh8fgpM3x
Clearing a shipyard with "(None)" opened the layout configuration dialog on a
grid of no cells. The auto-open guard asked only whether the chosen id differs
from the current one, and the "(None)" option carries the empty id, which
differs from every schematic; ShipLayoutDialog then found no ship def and
derived a 0x0 grid. Predates this branch -- 698dd4d, 2026-07-13.
The question all three sites were answering by hand is now one:
findLayoutShipDef() returns the ship to configure a layout against, or nullptr
when no schematic is set, the id names no ship, or the ship defines no grid.
The auto-open path and the LayoutDialogRequestedEvent handler now ask it
before opening, and ShipyardContent asks it instead of spelling the same test
out for the preview and the Configure button.
The event handler was reachable only through a button ShipyardContent already
disables, so guarding it changes nothing today; it is guarded because the
dialog's precondition belongs to the dialog's entry, not to the widget that
happens to be the only caller.
REQ-MOD-UI-AUTO-DIALOG said "differs" and left clearing implicit, which is the
reading the code took. It now says clearing opens nothing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ne3mejABZoLWKLh8fgpM3x
Three dialogs take Q as a second way out beside Escape: the recipe/schematic
selection dialog and the blueprint selection dialog close outright, and the
layout configuration dialog steps out one level per press -- the module being
placed, then remove mode, then the session. Both of those mode exits now go
through the handler the Remove button uses, extracted from a lambda into
onRemoveButtonClicked(), so the key and the button cannot leave different
state behind.
The schematic choice dialog goes the other way and declines reject(). It had
no close button but Escape still closed it, and the caller then applied
choiceIndex 0 -- awarding whichever option happened to be first. Refusing
reject() covers Escape, Alt+F4, and the window manager together, since all
three funnel through it. It is also the only dialog whose dismissal would
strand state: the poll that opened it does not reopen it while the choices
stay pending, so a drop dismissed is a drop lost.
The key itself is spelled once in DialogDismiss.h rather than in three key
handlers. It stays out of the ControlAction table on purpose: that table
answers what an input does in the player's current situation, and a dialog
has none -- it holds focus and takes the key whatever the world is doing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ne3mejABZoLWKLh8fgpM3x
Add deterministic record/playback for a run.
Recording captures `(seed, config hash, ordered tick-tagged commands)` and re-simulates on playback — no state snapshots. `DotaFactory.exe --replay <file>` re-plays a recorded run view-only with manual speed/pause.
Reviewed-on: #4
Co-authored-by: Malte Langkabel <malte.langkabel@gmail.com>
Co-committed-by: Malte Langkabel <malte.langkabel@gmail.com>
