The escalation remains necessary for DoRenderLineCanvas and ClearNeedsDraw,
but it no longer drives parent self-content redraws — those are now gated on
the needsDrawSelf snapshot captured before this escalation. Updated the inline
comment to reflect the narrowed purpose.

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

…ing draw

Remove the redundant SetNeedsDraw() call in the base AdornmentView's
OnClearingViewport. The thickness has just been drawn; calling SetNeedsDraw
mid-pass cascades to the parent's SubViewNeedsDraw and adds churn that
competes with the needsDrawSelf gate introduced for gui-cs#5358.

MarginView already overrides without this call. BorderView and PaddingView,
which inherit the base implementation, also benefit. Also refresh the stale
class doc comment — Border and Padding now extend AdornmentView too.

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Five tests covering the needsDrawSelf gate:
- ChildOnlyDirty: parent does not clear/redraw self
- ParentDirectlyDirty: full redraw path still works (regression guard)
- TransparentParent: transparent early-return composes with the gate
- ParentWithBorder: parent skips clear/text even with an adornment present
- IOutput layer: child's render still reaches IOutput after the fix

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

…-cs#5358

DrawComplete fires unconditionally for clip-exclusion bookkeeping even when
NeedsDraw is false, so it is the wrong signal for measuring whether a view
actually did draw work. Add a DrawingText counter to both the diagnostic and
integration tests — DrawingText is gated on NeedsDraw inside DoDrawText and is
not intercepted by Code (which does intercept ClearedViewport / DrawingContent
via OnClearingViewport / OnDrawingContent returning true). Use DrawingText to
verify per-Code inactive-tab behavior, and use tabsCounts.ClearedViewport
to verify the parent's own behavior.

TabsFanOutDiagnosticTests (synthetic Layout/Draw path) now asserts:
- inactive tabs have DrawingText == 0
- Tabs container has ClearedViewport == 0
- text-draw fan-out ratio == 1.0 (parity with single-Code baseline)

TabsFanOutIntegrationTests (end-to-end via main loop) keeps the existing
"documents remaining fan-out" assertion: a separate cascade source remains
in ApplicationImpl.LayoutAndDraw (force=true when neededLayout, calling
SetNeedsDraw on the top runnable which cascades to overlapping subviews).
Removing that force=true uncovers stale-content bugs in the shrink/move
path (already covered by ShadowTests / BorderViewTests) and is out of
scope for gui-cs#5358. Layout fan-out is fully eliminated by PR gui-cs#5373.

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

The previous union math at View.NeedsDraw.cs lines 87-92 was buggy:

  int x = Math.Min (Viewport.X, viewPortRelativeRegion.X);
  int y = Math.Min (Viewport.Y, viewPortRelativeRegion.Y);
  int w = Math.Max (Viewport.Width, viewPortRelativeRegion.Width);
  int h = Math.Max (Viewport.Height, viewPortRelativeRegion.Height);

It unioned against Viewport (not against the existing NeedsDrawRect), and used
max-of-Width / max-of-Height rather than Rectangle.Union. The result was that
NeedsDrawRect was effectively clamped to the current Viewport size on every
call — accumulated partial regions were silently widened to ~viewport-wide,
making NeedsDrawRect useless for narrowing draw work.

Replace with Rectangle.Union (NeedsDrawRect, viewPortRelativeRegion) so
NeedsDrawRect accurately tracks the accumulated dirty area. This is required
for issue gui-cs#5358's region-aware ClearViewport.

Update NeedsDrawTests to reflect the new accumulating semantics:
- NeedsDrawRect_Is_Viewport_Relative: shrink scenarios now keep the larger
  accumulated area (Union of prior larger frame and current smaller one)
- SetNeedsDraw_MultipleRectangles_Expands: union of two non-overlapping
  rects is the bounding rect, not the full viewport

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

When NeedsDrawRect is set and genuinely smaller than the full viewport (i.e.
a real partial-region invalidation, not SetNeedsDraw()-no-arg which sets
the full Viewport), narrow the clear and the trailing SetNeedsDraw cascade
to just that region. Otherwise preserve the existing full-clear contract
used by Code.OnClearingViewport, MarkdownCodeBlock, and direct test callers
(e.g. ClearViewport_FillsViewportArea, ClearViewport_SetsNeedsDraw).

This is the second half of the infrastructure for issue gui-cs#5358's narrowed
draw work: combined with the SuperView invalidation on Frame change added
in the next commit, geometry changes will only clear and cascade to the
affected region instead of forcing the whole viewport and all overlapping
peers to redraw.

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

When a view's Frame shrinks or moves, the SuperView's old-frame area is now
uncovered and must be cleared on the next draw — otherwise stale content
remains (e.g. an old shadow after a view shrinks). Add a SuperView.SetNeedsDraw
(Rectangle.Union(originalFrame, Frame)) call in View.Layout(Size) when
frameChanged. Combined with the region-aware ClearViewport from the previous
commit, this clears only the union area and cascades NeedsDraw only to
overlapping subviews intersecting that union — not the entire viewport.

Note: this only catches Frame changes. Adornment thickness changes (e.g. a
Margin shrinking when a shadow is removed) don't change Frame, so this fix
doesn't help them. A future change can extend this by tracking adornment
thickness deltas during Layout — at which point ApplicationImpl.LayoutAndDraw
can drop force=true on neededLayout and the integration-level draw fan-out
in TabsFanOutIntegrationTests can be flipped to assert == 0.

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

This is the cascade source that keeps TabsFanOutIntegrationTests bandaged.
Removing force=true when only neededLayout (not forceRedraw) requires
tracking adornment thickness changes so the SuperView can be invalidated
precisely. Until that lands, ShadowTests and BorderViewTests rely on
force=true to clear stale content after adornment thickness rebalances
that don't change Frame.

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

…tests

SetFrame is the direct-assignment path for view.Frame = ... (used by code that
sets Frame outright). It bypasses Layout()'s frameChanged detection because
SetFrame updates _frame before Layout runs. Add the same SuperView.SetNeedsDraw
(Union(oldFrame, newFrame)) call in SetFrame so direct Frame assignments also
trigger the region-aware uncovered-area clearing.

Add RegionAwareClearViewportTests with 5 focused tests:
- DirectCaller_EmptyNeedsDrawRect_ClearsFullViewport — backward-compat contract
- PartialNeedsDrawRect_ClearsOnlyDirtyRegion — narrowing fires when partial
- FullViewportNeedsDrawRect_ClearsFullViewport — SetNeedsDraw() no-arg path
- FrameShrink_InvalidatesSuperViewWithUnionOfOldAndNewFrames — Step 3 catches it
- Scroll_DoesNotInvalidateSuperView — pure scroll does not fan out

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

…serve transparent cache

CoPilot code review surfaced three correctness issues with the region-aware
ClearViewport from the prior commit:

1. Coordinate-space inconsistency. SetNeedsDraw(Rectangle) cascades to subviews
   using frame-local coordinates (subtracts subview.Frame.X/Y), while the no-arg
   variant passes content-coord Viewport. For an unscrolled view those coincide;
   for a scrolled view they disagree, and my narrowing math (subtracting
   Viewport.X/Y) would shift the clear off-screen on a scrolled subview.

2. Code/MarkdownCodeBlock full-clear contract. Both call public ClearViewport()
   from their OnClearingViewport overrides expecting a full fill of the
   code-block background. With NeedsDrawRect set partial by the cascade, the
   narrowed clear left stale background in the un-cleared part of the viewport
   (visible when content has fewer lines than the viewport, etc.).

Fix: revert public ClearViewport() to always full-clear. Move narrowing logic
into a private CanNarrowClearToNeedsDrawRect helper invoked only from
DoClearViewport (the framework's draw-pipeline entry point), guarded by
Viewport.Location == Point.Empty so the scrolled-cascade ambiguity is
sidestepped entirely. The Item 1 normalization (frame-local vs content-coord
NeedsDrawRect) is a deeper refactor deferred as a follow-up.

3. TransparentMouse hit-region wipe. _localDrawContext was recreated
   unconditionally before the needsDrawSelf gate, but on child-only passes
   DoDrawText/DoDrawContent are skipped, so _localDrawContext stayed empty.
   DoDrawComplete then wrote that empty region into CachedDrawnRegion, breaking
   mouse hit-testing for transparent views until the next full self-redraw.

Fix: only recreate _localDrawContext when needsDrawSelf is true. On child-only
passes the prior context is preserved, so DoDrawComplete keeps the previously-
cached drawn region. SetNeedsDraw nulls CachedDrawnRegion when the view is
genuinely invalidated, so stale cache values are bounded by view lifecycle.

Updated tests:
- RegionAwareClearViewportTests: switched from testing the public ClearViewport
  API to testing the framework's DoClearViewport via Draw(). Added a
  FrameworkDraw_ScrolledView_FallsBackToFullClear test that locks in the
  scrolled-view guard from fix 1.
- SubViewOnlyRedrawTests: added TransparentMouseParent_ChildOnlyDirty_
  PreservesCachedDrawnRegion to lock in fix 3.

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

CoPilot review item 4: SetFrame() already calls SuperView.SetNeedsDraw(union)
for every Frame mutation (both direct view.Frame=... and layout-driven paths
go through SetFrame). The matching block in Layout(Size) repeated the union
and the cascade work on the hot path this PR is trying to trim. The duplicate
was idempotent (Rectangle.Union of identical rects produces the same rect),
but the cascade-to-overlapping-subviews work ran twice.

Refs gui-cs#5358

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>