Owner's second live look: "On items tab, we show some options there
right of the items we add. That should not be there. Remove that."
VTank's real Items tab is exactly 6 controls (docs/research/vtank-kb/
08-ui-views.md §1 "Tab: Items" — 2 labels, the 2-column
clWeaponName/clHandedness list, Add, Add (no buffs)); mosstank.xml's
Items group also carried a right-of-list block with no VTank
counterpart: the Weapon/Offhand readout (MonsterEquipmentText), the
Refill Worn Mana toggle + slider + its two status labels
(RefillWornManaText/ItemManaRechargeStatus), and the ProfileNotice
hint label. All five are deleted from the markup.
The pre-existing "Remove" button stays — it traces back to the
plugin's very first automation PoC commit (`4e6e9bc9d`, long before
Campaign VT slice 7 existed), the accepted "slice-1 Delete" carried
forward per the round's own scope.
RefillWornMana/RefillWornMana-Item-ManaPercent are real
VtankOptionCatalog entries (Bool/Int, not tString) — nothing was
deleted from the plugin, they remain real settings, still editable in
the Advanced Options popup or via `/vt opt set`; only this second,
redundant Items-tab surface for them is gone. ItemManaRechargeStatus
was a pure runtime status readout with no setting behind it and no
VTank equivalent, so it has no replacement — matching VTank's own tab,
which shows no such status either.
Mutation shown to fail first: the new
ItemsTabIsVtankOnlyPlusTheAcceptedSliceOneRemoveButton test asserted
the Items group has no toggle/slider and no label bound to any of the
four removed properties, against the UNCHANGED markup — failed on the
Refill Worn Mana toggle still being present — then passed once the
block was deleted from mosstank.xml.
MossTank suite 723 -> 724 (one new pin); App markup/plugin filter
258/258 (unaffected — no App-side markup change).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner's second live look: "Buff_Profile banes and prots got duplicated.
Only the one with ALL works as intended." BuffProfile-Prots/-Banes are
tString rows (VtankOptionCatalog.DeclaredType == String); VTank's own
Advanced Options list has NO case for tString at all (KB
01-settings-and-profiles.md, db.cs:132-166 for the list switch and
:201-203 for the click handler — both "Not exposed"), so the hyphen
rows never belonged in the list next to their real tEnum twins
(BuffProfile_Prots/_Banes) — that mismatched pair is the "duplicate"
the owner saw. FilteredAdvancedOptionNames now excludes every tString
row (also catches BlacklistedSpellComps, the catalog's third tString
entry); the enum rows stay listed and clickable exactly as before.
tString settings are unaffected otherwise — still real, still settable
via `/vt opt set`, only no longer rendered in this list.
Mutation shown to fail first: the new
AdvancedOptionListHidesTStringSettingsButKeepsTheirEnumCounterparts
test asserted BuffProfile-Prots/-Banes/BlacklistedSpellComps absent and
the two enum names present against the UNCHANGED filter — failed with
"BuffProfile-Prots" found in the collection — then passed once
FilteredAdvancedOptionNames gained the DeclaredType != String guard.
MossTank suite 722 -> 723 (one new pin); App markup/plugin filter
258/258 (unaffected — no App-side markup change).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner's second live look (2026-09-07): "The whole bottom window of
advanced options should not be there." VTank's own AdvancedOptionsView
is a bare list + category list + edit field + description (KB
08-ui-views.md §1) — no notice line, no extra controls. Deletes the
"Editing X." notice label and the entire "MossTank Extras" section fix
round B item 8/9 added below it (Checkpoint/Jump/Remove/Set Follow
Target + status label/Follow Corners/Open Doors/Nav Priority/
Follow-Nav min distance stepper), shrinking the popup back to VTank's
392x300 (resizable, min = that; was 392x450).
Every control removed had a real path already, so nothing is actually
unreachable:
- Checkpoint/Jump: `/mt addnavcheckpoint` / `/mt addnavjump` (already
documented in the /vt help text).
- Remove: click any row of the Route tab's own waypoint grid
(DeleteRouteWaypointAt) — RemoveRouteWaypointCore and
DeleteRouteWaypointAtCore are the identical removal logic, so the
button duplicated a control that already exists.
- Set Follow Target: re-select "Follow" in the Route tab's own
nav-mode menu — SelectRouteMode already calls CaptureFollowTarget
every time RouteMode.Target is (re-)selected.
- Follow Corners/Open Doors/Nav Priority/Follow-Nav min distance:
real VtankOptionCatalog entries (FollowAroundCorners, OpenDoors,
NavPriorityBoost, NavCloseStopRange) already listed and editable in
THIS SAME popup's own option list/field above, or via
`/vt opt set <Name> <value>`.
No C# state or command was deleted — MossTankPanel's properties/actions
(RemoveRouteWaypoint, SetFollowTarget, AddRouteCheckpoint, AddRouteJump,
NavigationSettings.FollowAroundCorners/OpenDoors/Priority/
MinimumDistanceMeters) are untouched; only the redundant second UI
surface for them is gone.
Mutation shown to fail first: MossTankMarkupContractTests'
ExpectedPopupBounds and MossTankMarkupBuildOverRealFilesTests'
WideningTheRealAdvancedOptionsPopupGrowsTheOptionListWithoutOverlappingItsSibling
were changed to expect 392x300 before the markup shrink — both failed
with "Expected: 300 Actual: 450" against the still-450-tall file, then
passed once mosstank-advanced.xml's own h/minh dropped to 300.
MossTank suite 722/722 (no count change, no control removed here had a
dedicated test); App markup/plugin filter 258/258.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Records round E (the architecture re-check of round D): all six items
fixed, commit SHAs, test-count deltas per item, and the two real test-
methodology bugs plus one test-harness false positive found while
building D-6's resolved-geometry pin.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Every real-file re-layout pin so far inspects ONE named element after a
resize (the Monsters list, the Advanced Options option/category
lists). Adds a whole-tree pin at both ends of the main panel's declared
resize range — the minw/minh floor (856x236) and one enlarged size
(1100x320) past the 984x271 default — for each of the nine tabs in
turn: builds the real mosstank.xml through MarkupDocument.Build, draws
once at the authored default to capture anchor baselines, resizes, and
asserts no sibling overlap and no child crossing its parent's edge.
Adapts MossTankMarkupContractTests' AssertNoSiblingOverlap/
AssertWithinParent algorithm (rectangle intersection, ignore
touching/zero-size boxes, two bare <group> tab pages never overlap-
check each other) from the authored XElement tree to the resolved
UiElement tree (real Left/Top/Width/Height post-anchor), since a resize
only exists at that level.
Two real bugs surfaced and fixed in the test itself while building this:
(1) reusing one built tree across all nine tabs let an earlier tab's
resize leave the shared root already at the target size by the time a
later tab's own descendants took their "first" (baseline) anchor
capture — fixed by building a fresh tree per tab; (2) walking into an
INVISIBLE tab group's descendants compared their stale, never-
anchored-for-this-size geometry against the group's own (correctly
resized) bounds — fixed by skipping recursion past an invisible node,
since it never got a real anchor pass this round. A genuine false
positive from the test harness itself was also found and excluded:
UiLabel.OnDraw (the CT-GF1 fix) always overwrites Width/Height to the
current measured text extent, and this stub renderer has no real
DatFont, so it falls through to a crude "text.Length * 7f" placeholder
far wider than the authored box or the real DAT font would ever
measure — labels are excluded from the size checks (position still
checked via recursion) for the same reason D-3's own comment cites this
constraint.
Mutation shown to fail first: removing the Monsters list's
anchor="left right top bottom" attribute (temporary edit, reverted —
net diff on mosstank.xml is zero) reproduced "UiMarkupList @
(0,16,976,151) crosses the right edge of UiPanel (w=848)" at the 856
floor, exactly the failure this pin exists to catch.
App markup/plugin filter 244 -> 246; MossTank suite holds 722/722
(App-side test only).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
UiMarkupList.OnDraw/DrawColumns re-ran their "keep the selected row
visible" clamp on EVERY draw call, not only when SelectedIndexSource
actually changed. A real bound list (MossTank's Advanced Options,
selected="{SelectedAdvancedOptionIndex}") holds a stable selected index
while the user operates the scrollbar, so the very next frame's OnDraw
saw that (unchanged) row now outside the just-scrolled view and yanked
_topRow straight back to it — undoing the arrow click or thumb drag
before it was ever visible. This is exactly the "sometimes works"
symptom: a scroll that happens to keep the selected row in view
survives; one that moves it out of view gets reverted on the next
frame. Gate the clamp on an observed change in the selected index
(int.MinValue sentinel so the first frame still reveals it), so a
stable selection no longer fights user-driven scrolling while a
genuinely new selection is still auto-revealed once.
New tests build a real nested Panel > Group > List tree at non-zero
offsets and drive it exclusively through UiRoot's public mouse entry
points, ruling out a coordinate-frame bug in the hit-test/dispatch
chain as well as proving the fix for both single-column and
<column>-mode lists (arrow click and thumb drag).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
RefreshAdvancedOptions' own doc comment lists every real mutation site
that must call it (category toggle, edit/apply, selection change,
profile load) but ShowAdvancedOptions itself was missing — opening the
popup only called LoadAdvancedOptionDraft. A setting changed while the
popup was closed (/vt opt set, one of the Options tab's own direct
checkboxes, or a profile load) left the cached AdvancedOptionNames/
AdvancedOptionValueColumn on whatever they held from construction or
the popup's last open, until some in-popup interaction happened to
refresh them.
Mutation shown to fail first: reverting the source fix (git checkout,
patch saved and reapplied) reproduced the stale column — toggling
CombatEnabled then opening the popup still showed the pre-toggle value.
MossTank suite 721 -> 722; App markup/plugin filter holds 244/244.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
mosstank.xml's own top-of-file comment already claimed the Advanced
Options popup "gets its own resizable=\"true\" + anchored lists" — the
popup file itself still carried the dead resize="none" attribute
(MarkupDocument.cs: "resize" only ever means anything once
resizable="true" already arms the master switch), so the claim was
false and the popup was permanently fixed-size.
resizable="true" now floors minw/minh at the pre-fix authored size
(392x450 — never shrink below the tested layout, matching the main
panel's own rule). lOptionList grows WIDTH ONLY: the value-edit field,
description, and notice all sit directly below it with zero vertical
slack and none of them anchor to shift down in lockstep, so any height
growth here would walk the list straight into the field below — the
same "grow only in the direction with nothing in the way" principle the
main panel already applies to Items/Consumables/Buffs/Route's lists.
lFilterList (the category checklist beside it) tracks the growing right
edge via its own anchor="right" so the widening option list never walks
into it, matching the main panel's own right-pinned-sibling pattern.
AdvancedOptionDescription is declared anchor="left top right" for
markup consistency, but documented as a real no-op: UiLabel.OnDraw
(UiPanel.cs) always overwrites Width/Height to the current measured
text extent on every draw, right after the anchor pass runs in the same
frame, so nothing a label's anchor computes is ever visible — a
genuinely working vertical anchor was deliberately avoided since the
notice label directly below isn't anchored either and a large resize
would walk description into it; fully anchoring the rest of the
vertical stack is a larger redesign left for a future round.
Mutation shown to fail first: reverting the markup (git checkout, patch
saved and reapplied) failed at Assert.True(built.Resizable) — the popup
was still fixed-size.
App markup/plugin filter 243 -> 244; MossTank suite holds 721/721
(this item touches App-side markup only).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
RouteRecallNames/SelectedRouteRecall used Enum.GetNames/ToString(),
showing bare C# identifiers ("PrimaryPortalRecall",
"ParadoxTouchedOlthoiInfestedAreaRecall", ...) in the Route tab's
recall combo instead of VTank's real recall names. Both now route
through RouteWaypoint.RecallDisplayName (the same table
AddRouteRecallCore/RecallLabel already use), matching how
RouteInsertModeNames/RouteModeNames already show dedicated display
captions rather than raw enum text. SelectRouteRecall now parses that
same display string back (Enum.TryParse against the enum identifier no
longer matches what the combo sends).
mosstank.xml's cmbRecallType widened 120 -> 300: VTank's own combo is
also 120 wide, but only because its captions are hand-abbreviated
("LS", "Mt. Lethe", "PtOIA"); RecallDisplayName deliberately keeps
metaf's full names (round D item 3), so the box needs real room for
the longest one, "Paradox-touched Olthoi Infested Area Recall".
Nothing else shares that row past x=474 and the group floors at ~848
wide even at the panel's minw, so 300 has ample clearance.
Mutation shown to fail first: reverting MossTankPanel.cs's fix (git
checkout, patch saved and reapplied) reproduced the bare-identifier
list and the "PrimaryPortalRecall" default selection.
MossTank suite 720 -> 721; App markup/plugin filter holds 243/243
(no overlap from the widened menu).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The pre-cutover RouteRecallKind was {Lifestone=0, Marketplace=1,
PrimaryPortal=2, SecondaryPortal=3} — completely different kinds at the
SAME ordinals as today's much larger round-D enum (PrimaryPortalRecall=0,
SecondaryPortalRecall=1, LifestoneRecall=2, LifestoneSending=3, ...).
LegacyWaypointDocument.Recall was still typed RouteRecallKind, so
System.Text.Json deserialized a legacy JSON route's bare "Recall"
integer straight into the new enum — silently remapping every one of
the four old ordinals to the wrong new recall. Enum.IsDefined never
caught it (0..3 are all valid members of the new enum too, just for
different spells).
The field is now a bare int, translated through an explicit
MapLegacyRecall(legacyOrdinal) table in ToWaypoint() that also derives
RecallSpellId/RecallSpellName from the mapped kind (the pre-cutover
schema predates both fields entirely).
Mutation shown to fail first: reverting the source fix (git checkout,
patch saved and reapplied) reproduced the bug for all four legacy
ordinals — ordinal 2 (old PrimaryPortal) even landed on LifestoneRecall
instead of PrimaryPortalRecall, confirming the "still IsDefined, still
wrong" failure mode. The new theory asserts against the migrated
route's own written .af "rcl" line (RecallDisplayName) rather than the
value re-parsed back through LoadCurrent, since the .af format's
own name-only round-trip is a separate, pre-existing limitation
(no spell id, and "Marketplace Recall" cannot resolve through any
catalog because it isn't a real spell) unrelated to this fix.
MossTank suite 716 -> 720; App markup/plugin filter holds 243/243.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
#490 part 2 (plugin layout persistence has no revision bump) is fixed at
05f22d46f (RetailWindowManager.ComputeAuthoredGeometryRevision +
MigrateAuthoredGeometry's inequality comparison + MountPlugins wiring).
Part 1 (StartVisible=false + ShowInSidePanel=false is permanently
unshowable) remains open -- not touched by this work.
Also adds one sentence to docs/plugin-ui-markup.md's "Resizable panels and
anchors" section: changing a panel's authored w/h/minw/minh/resizable in a
later plugin update resets every user's stored size to the new default
exactly once, automatically -- no plugin-author call site to remember.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
MountPlugins registered every plugin window with authoredGeometryRevision
hard-coded to 0 (RetailUiRuntime.cs), so RetailWindowLayoutPersistence's
MigrateAuthoredGeometry -- gated on "saved revision >= authored revision"
-- never migrated a plugin window's saved size: 0 >= 0 forever. MossTank's
panel went 856x236 -> 984x271 and every user with a stored layout stayed
stuck at 856x236 with no way to see the new default.
Two changes:
1. MountPlugins now passes RetailWindowManager.ComputeAuthoredGeometryRevision
(added previous commit) as the plugin window's authoredGeometryRevision,
derived from the panel's own authored width/height/minw/minh/resizable.
2. MigrateAuthoredGeometry now compares revisions for INEQUALITY
(saved.Revision == authored.Revision) instead of ordering
(saved.Revision >= authored.Revision). A hash is not an incrementing
counter -- two different authored sizes can hash in either order -- so
"the authored size changed" has to mean "the value differs", not "the
value went up". Built-in windows' hand-picked incrementing literals
(chat: authoredGeometryRevision = 1) are unaffected: no existing saved
revision is ever equal to a later, different literal either way.
Mutation shown to fail first: the two new PluginMarkupPanel_AuthoredSizeChanged_*
tests in RetailWindowLayoutPersistenceTests.cs reproduce the exact bug with
concrete literals (856x236/400/150/true -> 984x271/... and
200x100/100/80/true -> 220x110/...) chosen so ComputeAuthoredGeometryRevision's
OLD hash is >= the NEW hash for each pair -- confirmed via a throwaway probe
before writing the assertions, so the pre-fix run fails deterministically
rather than by chance of hash ordering. Both failed before this commit
(size stayed at the old authored extent) and pass after
(PluginMarkupPanel_AuthoredSizeUnchanged_KeepsUserResizedSize, unaffected
either way, is a regression-safety companion). Full RetailWindow/Markup/
PluginSidePanel filter: 249 passed (was 246), 0 failed, 0 skipped.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Slice-7 resemblance re-check found UiMarkupList's column-less items= mode
and its <column> mode disagreed on whether a selected row gets a
persistent SelectedColor fill, so a plugin's Buffs lists highlighted a
row while the Monsters/Meta grids looked different for the same widget.
Real VVS lists (VTank's HudList) draw no such fill at all.
UiMarkupList.SelectionBandEnabled (default false) now gates the fill in
both OnDraw's legacy branch and DrawColumns; MarkupDocument parses
<list selectionband="true"> (same literal-bool convention as
openupward/clearonsubmit) to opt a single list back in. selected/onchange
semantics and scroll-into-view of the selected row are unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Part of #490 part 2: plugin panels register with authoredGeometryRevision
hard-coded to 0, so a plugin author who ships a new authored panel size
(MossTank: 856x236 -> 984x271) has no way to signal the change short of
adding a manual revision-bump call site, and every user's stored layout
keeps the old size forever. Built-in retail-imported windows solve this
with an explicit incrementing int literal at each Register call; a plugin
author does not maintain that call site by hand.
RetailWindowManager.ComputeAuthoredGeometryRevision derives the revision
from the authored geometry tuple itself (width, height, minw, minh,
resizable) via a fixed FNV-1a-style combine over the values' raw IEEE-754
bit patterns -- deliberately not System.HashCode, whose per-process reseed
would make the "same" authored geometry hash differently on every launch.
The sign bit is masked off so the result is never negative (Register's own
Math.Max(0, revision) would otherwise silently fold distinct negative
hashes onto the same "unversioned" 0 bucket used by legacy saves).
Mutation shown to fail first: without this method,
RetailWindowManagerTests.ComputeAuthoredGeometryRevision_* (7 new tests)
fails to compile (CS0117, method does not exist). No wiring yet -- this
commit only adds the pure, inert helper; MountPlugins still passes no
revision. That lands next along with the comparison-semantics fix it
depends on.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
An .af "rcl" node whose spell name failed to resolve against the
character's known spells (MetafSerializer.ReadNavNode's "rcl" case)
leaves RecallSpellId at 0 and Recall at its default ordinal
(RouteRecallKind.PrimaryPortalRecall — the enum field is never touched
on that path). SubmitRecall's old Recall-based fallback then silently
cast Primary Portal Recall (spell 48) for a waypoint that was never
authored as one. TickRecall now refuses the cast, names the unresolved
spell in the status notice, and skips the waypoint (CompleteAction) —
the same "refuse and continue" shape TickUse already uses for an
unavailable object. SubmitRecall's dead fallback branch (every genuine
path — AddRouteRecallCore, the binary .nav loader, the D-1 legacy-JSON
migration — writes RecallSpellId directly) is removed.
Mutation shown to fail first:
RecallWaypointWithUnresolvedSpellNameRefusesAndSkipsWithoutCasting
against the pre-fix code cast spell 48 (magic.CastSpellIds == [48]).
MossTank suite 715 -> 716; App markup/plugin filter holds 243/243.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Records the merge of claude/latest-main-sync-497549 plus round D's
four fix items (Advanced Options category names, VTank's click model,
the full 26-recall table, resizable/enlarged main panel) on the panel
worktree, with commit SHAs, test-count deltas, and the carried
deviations (#491, the RouteRecallKind-ordinal migration risk).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner live report 2026-09-07: "The size of the entire window needs to
be enlarged for default and should also be resizeable." mosstank.xml's
main panel is now resizable="true" with minw/minh floored at the
PRE-round-D authored size (856x236) and a new default authored size
~15% larger in both directions (984x271) — every one of the nine tab
groups grows by the identical 128px-wide/35px-tall delta (848x194 ->
976x229), preserving the original 0px group-to-panel margin design
exactly.
Every tab <group> gets anchor="left top right bottom" so a future
manual drag-resize keeps growing it. The two single-list tabs with
nothing beside their list (Monsters, Meta) give their list the SAME
full stretch, widened AND heightened by the same 128x35 delta at
author time — anchors only react to CHANGE from their first-draw
baseline, so bumping only the outer group while leaving the list's own
geometry untouched would leave it visually unchanged at the new
default (the growth invisibly reserved as margin until a further
manual resize). Every other tab's list(s) grow in HEIGHT only (+35),
pinned to whichever horizontal edge doesn't have a sibling list/button
column in the way (Items/Consumables/Buffs/Route all have one) — a
right-pinned sibling (Consumables' Excluded Scarab list, Buffs'
Blacklisted Buff Families list) is repositioned +128 at author time so
it already sits flush against the enlarged default's right edge.
Trailing "Add" rows/status labels below a taller list shift down by
the same +35 and get anchor="bottom" to keep tracking. Options/
Profiles/Vitals need no internal changes — only their own group grows.
Deviations, both documented in mosstank.xml's own Round D item 4
comment: Monsters' extra 128px of list width lands in its LAST column
(the MoveDown icon, which always absorbs the remainder per the markup
grammar) — a wider icon cell, not a redesigned grid; seven of nine
tabs' groups grow wider than their content uses, leaving harmless
empty space on the right at the enlarged default rather than
redesigning nine tabs' pixel geometry in one pass (a deliberate,
minimal-risk choice — nothing moved INTO another control's space, only
into previously-empty margin, so every existing AssertNoSiblingOverlap/
AssertWithinParent guarantee still holds).
Tests added:
- MossTankMarkupContractTests.AuthoredShellFitsTheMinimumCanvasAnd...
updated to the new 984x271 default.
- PanelIsResizableFlooredAtThePreRoundDAuthoredSize (resizable="true",
minw/minh = 856/236).
- EveryStretchingListDeclaresARealAnchor (contract pin: every list in
Monsters/Items/Consumables/Buffs/Route/Meta declares a real anchor
attribute, not the silent left-top default).
- MossTankMarkupBuildOverRealFilesTests.
WideningTheRealMainPanelWidensTheRealMonstersList (App.Tests): builds
the REAL mosstank.xml against a real MossTankPanel, widens the built
root by 100px, and confirms the Monsters list's own Width grows in
turn — the same mechanism MarkupResizableAnchorTests already proves
against synthetic markup, now proven against the shipped file.
Mutation shown to fail: temporarily removing the Monsters list's own
anchor attribute failed the new re-layout test (976 -> 976, no growth);
restored and confirmed green. A build-copy gotcha surfaced while
writing these tests: dotnet build's default incremental copy did NOT
refresh the test projects' deployed mosstank.xml after editing the
source file (PreserveNewest apparently didn't see it as newer under
normal incremental evaluation) — dotnet build --no-incremental was
needed to get a fresh copy into bin/ before the size/anchor tests
would actually exercise the new markup instead of a stale cached copy.
Verified: dotnet build AcDream.slnx -c Release (clean, --no-incremental)
green; MossTank suite 715/715 (713 -> 715, two new markup contract
pins); App markup/plugin filter 243/243 (242 -> 243, one new re-layout
test).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner live report 2026-09-07: "Under Route, Add recall is missing a
lot of recalls that VTank has and they do not work in routes yet."
RouteRecallKind's old 4-entry model (Lifestone/Marketplace/
PrimaryPortal/SecondaryPortal) is replaced with VTank's real
cmbRecallType table: 26 castable recall spells from metaf's own
NRecall table (metaf_monolithic.py:10981-11008) plus Marketplace
Recall, VTank's one entry with no spell (issued as /marketplace).
Deviation (no refs/vtank/ checkout in this worktree to confirm VTank's
real combo layout directly): the old Lifestone member, which issued
"/lifestone" as a slash command, is DROPPED rather than kept alongside
the new spell-based LifestoneRecall (1635) — VTank's real combo has
ONE "Lifestone Recall" entry and it is a real castable spell, so
keeping both would show two menu rows reading "Lifestone Recall" with
different behavior. Marketplace is the one member kept as a slash
command per the explicit "keep ours" instruction. PrimaryPortal/
SecondaryPortal are renamed to PrimaryPortalRecall/SecondaryPortalRecall
and now use their real spell ids (48, 2647) instead of the old runtime
KnownSelfBuffs name lookup. This reshuffles the enum's underlying
ordinals; the one place that mattered (MossTankRouteProfileStore's
legacy pre-cutover JSON migration DTO, which stores Recall as a raw
int) already guards with Enum.IsDefined and is documented as a
migration-only path for not-yet-migrated files — its fallback default
moved from the deleted Lifestone to PrimaryPortalRecall.
Changes:
- RouteRecallKind: 26 named members in VTank's own combo order (values
are plain sequential indices, not spell ids, so Enum.GetNames/
GetValues — which sort by underlying VALUE — reproduce that order)
plus Marketplace appended last.
- RouteWaypoint.RecallDisplayName: VTank's exact label text per kind.
- RouteWaypoint.SpellIdForRecall (new): the real spell id per kind
(Marketplace = 0, the existing "no spell" sentinel).
- NavigationController.SubmitRecall: unchanged fast path (RecallSpellId
!= 0 -> cast) now covers every recall added through the UI; the
fallback for a waypoint with no recorded id resolves through
SpellIdForRecall instead of the old runtime spell-name lookup, and
Marketplace still falls through to "/marketplace".
- MossTankPanel.AddRouteRecallCore: populates RecallSpellId AND
RecallSpellName on the new waypoint (matching what a metaf import or
the binary .nav loader already produces), so a waypoint added from
the Route tab's own combo executes identically to one round-tripped
through a real route file.
- mosstank.xml: the recall <menu> comment updated; rows 4->7 now that
scrolling through 27 entries is real, not grammar-only.
Tests added (NavigationTests.cs, MetafSerializerTests.cs):
- RouteRecallKindListsVTanksTwentySixRecallsInOrderPlusMarketplaceLast
(the 26-entry order via Enum.GetNames).
- RecallNameAndSpellIdTablesAgree (27-case Theory: name<->id both ways;
RouteRecallKind is internal so the Theory parameter is the public int
ordinal, cast back inside the method — a public method cannot expose
an internal-typed parameter, CS0051).
- RecallWaypointWithNonZeroSpellIdCastsThatSpell /
RecallWaypointForMarketplaceSubmitsTheSlashCommandNotACast (execution,
via a new FakeMagic tracking fake — FakeAutomation.Magic is now
settable instead of always NoOpAutomationSurface).
- RecallNodeRoundTripsByNameAndResolvesTheRealSpellIdFromTheCatalog
(three recalls through SaveNav/TryLoadNav with a new FakeSpellCatalog
that actually knows the spells, proving both the name AND the
resolved id survive the .af round trip).
Mutations shown to fail, then reverted: swapping the first two enum
members failed the order test; reducing SubmitRecall to `return false`
failed both execution tests (no cast recorded, "/marketplace" not
submitted).
Verified: dotnet build AcDream.slnx -c Release green; MossTank suite
713/713 (682 -> 713, 31 new tests); App markup/plugin filter 242/242
(unchanged).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner live report 2026-09-07, screenshots 2/3 (relayed VTank click
model — this worktree has no refs/vtank/ checkout to read
db.cs:118-165 directly): clicking a VALUE cell used to always just
select the row, no matter the setting's type. VTank's own model
dispatches by declared type: a tBool value flips in place, a tEnum
value cycles to the next label, and only int/double/single/string
select the row and load the current value into the edit field below
for typing. Apply/Back buttons are gone entirely (retail's own
AdvancedOptionsView has neither; Enter on the field already applies
via its existing onsubmit) — the popup closes from its own title bar
or the Options tab's toggle, both already independent of this content.
Changes:
- VtankDefaultSettingsDatabase.SettingEnumValues parses VTank's own
SettingsEnumInfo table (3 columns, 33 rows: Setting/Value/EnumValue)
from the embedded .usd — the real per-setting enum code->label table
(UseArcs: 1=No, 2=At Range, 3=Yes), not a hand-typed guess.
- MossTankPanel.DisplayAdvancedOptionValue shows the enum LABEL for
Enum-typed settings in the value column instead of the raw stored
integer.
- MossTankPanel.ClickAdvancedOptionValue dispatches to
FlipAdvancedOptionBool / CycleAdvancedOptionEnum (wraps past the
last entry) / SelectAdvancedOption by VtankOptionCatalog.DeclaredType.
- mosstank-advanced.xml: the value column's onclick is now
{ClickAdvancedOptionValue}; Apply/Back buttons removed; the notice
label and the whole "MossTank Extras" section below it moved up 26px
to reclaim the space; panel height 476->450.
- Removed the now-unreferenced ApplyAdvancedOption wrapper property
(ApplyAdvancedOptionCore is still used by the field's own onsubmit).
Mutations shown to fail: (1) reducing ClickAdvancedOptionValue to a
bare SelectAdvancedOption(index) call failed both the bool-flip test
(expected False, got True — the click never flipped it) and the
enum-cycle test (expected "At Range", got "No" — the click never
advanced); the numeric-select test correctly stayed green since
select-only is still its own expected behavior. Restored and
confirmed green.
Verified: dotnet build AcDream.slnx -c Release green; MossTank suite
682/682 (679 -> 682, three new interaction tests); App markup/plugin
filter 242/242 (unchanged, mosstank-advanced.xml's new footprint
392x450 updated in ExpectedPopupBounds).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner live report 2026-09-07: "Some Hex numbers with green button to
the right. What is that?" — the Advanced Options popup's category
filter checklist showed raw bitmask hex ("0x1", "0x2", ...) instead of
VTank's real category names (Misc, Recharge, MeleeCombat, SpellCombat,
Ranges, Navigation, Buffing, Crafting, Looting).
This worktree has no refs/vtank/ checkout, so VTank's own name/enum
string literal for each bit isn't directly readable. Derived the
mapping instead from real data already in the embedded .usd: for each
category name, docs/research/vtank-kb/01-settings-and-profiles.md §2
was searched for a setting whose Category column names EXACTLY that
one category (no `|` combination) — e.g. row 45 RandomHelperBuffs is
pure "Misc", row 18 AttackDistance is pure "Ranges" — then that
setting's own recorded bitmask was read back from
VtankDefaultSettingsDatabase.SettingCategoryBitmasks (never a typed-in
hex literal), so the mapping tracks the shipped database instead of
silently drifting from it. All 9 of VtankOptionCatalog.CategoryBits
resolve to a real name this way (VtankOptionCatalog.
CategoryNamesByBit), in VTank's own ascending-bit order.
Mutation shown to fail: reverting AdvancedOptionCategoryNames to the
old `$"0x{bit:X}"` projection failed the new
AdvancedOptionCategoryNamesShowRealNamesNotHexBitmasks test with
["0x1", "0x2", ...] instead of ["Misc", "Recharge", ...]; restored and
confirmed green.
Verified: dotnet build AcDream.slnx -c Release green; MossTank suite
679/679 (678 -> 679); App markup/plugin filter 242/242 (unchanged).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Merges claude/latest-main-sync-497549 into the slice-7 panel worktree
so round D can build on both: b71a8ea37 (retail scrollbar chrome on
plain <menu> popups and overflowing <list>s — chat/inventory sprite
ids, always-scrollable single-column menu) and 2e63391cc (<panel
resizable minw minh> + anchor="left top right bottom" markup).
Conflicts resolved keeping both intents:
- Ledger (docs/plans/2026-09-07-campaign-vt-slice7-tabs.md): unioned
both branches' entries into one chronological timeline instead of
picking a side.
- docs/plugin-ui-markup.md: kept both attribute additions per element
(slider min/max/style, menu scroll/style) AND anchor on every row.
- src/AcDream.App/UI/UiMarkupList.cs: kept this branch's fix round B
item 10 (VVS HudList grids have no row-selection highlight) over the
sync branch's older SelectedColor band draw in the <column> grid
path — the legacy single-column list path is unaffected either way.
- src/AcDream.App/UI/MarkupDocument.cs: the auto-merge left two
`Scrollable =` initializers on the same <menu> object (CS1912).
Kept the sync branch's `Scrollable = true` (VTank's HudCombo is
always a single scrolling column, never a wrapping grid) and
dropped this branch's `Scrollable = B(el, "scroll", false)` opt-in,
since the owner-driven always-scrollable design supersedes the
S7.2 opt-in one. Updated MarkupDocumentTests.cs to match: removed
Build_MenuWithNoScrollAttribute_KeepsScrollableFalse (asserted the
now-false opt-in default) and
Menu_Scroll_DrawsAPlainFlatThumbFillWhenTheMarkupItemCountOverflowsTheVisibleRows
(asserted a flat DrawFill thumb; the scrollbar is sprite-chrome for
every menu style now) — both fully superseded by
Menu_Markup_IsAlwaysScrollable_WithRetailScrollbarChromeWired and
UiMenuPlainStyleTests.Plain_OpenPopup_ScrollableOverflow_
DrawsRetailScrollbarChrome_RowsStayPlain.
Verified: dotnet build AcDream.slnx -c Release green; MossTank suite
678/678; App markup/plugin filter 242/242 (241 before this commit's
test-file trim, +1 net from the merge's own new tests, 0 red).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
EveryInteractiveControlDeclaresARealHandlerBinding asserted an exact
Assert.Equal(156, controls.Length), backed by ~60 lines of running
commentary recording every markup edit that ever bumped the number. That
count was a pure change detector: it carried no signal the test's own
per-control handler/enabled loop couldn't already catch on its own, and
it forced an edit to THIS test file every time an unrelated tab gained
or lost a single control.
Replaced the exact count with Assert.NotEmpty(controls) — it still
guards the selector itself (a broken interactive-element-name filter
that matched nothing would otherwise pass the loop vacuously) — and kept
the real assertion (the per-control handler/enabled loop) unchanged.
Moved the removed count's full history into this same commit's ledger
entry in docs/plans/2026-09-07-campaign-vt-slice7-tabs.md, per item
F13's own instruction.
Mutation named: temporarily broke the controls selector (appended
`&& false` to the interactive-element filter) and confirmed
Assert.NotEmpty fails ("Collection was empty") before restoring it.
MossTank suite holds at 678 (no test added or removed, one assertion
replaced). Full solution build green; App markup/plugin filter 203/203.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Adds the "Resizable panels and anchors" section to plugin-ui-markup.md:
the resizable/minw/minh attribute table for <panel>, the anchor grammar
for every element that now supports it, the AnchorEdges semantics
(left top default, left right / top bottom stretch, right/bottom pin-
and-move), group-relative child anchoring, and one worked example
(a resizable panel with a group and list that both stretch on drag).
Also updates the Elements table, the "every registered window" intro
paragraph, the ApplyCommon common-attributes paragraph, and the Testing
conventions section to reference the new grammar and its test coverage.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner live-client report 2026-09-07: "For scrollable dropdown or the meta
window we use the same assets as we do in for example chat or inventory
window."
<menu> markup wiring (MarkupDocument.cs): a plugin <menu> is now always
Scrollable (single-column, VTank HudCombo shape) instead of wrapping
overflow into more grid columns, with PopupScrollbarHideWhenDisabled=true
so the bar is entirely absent while the item count fits the "rows"
window. RetailScrollbarChrome.ApplyToMenuPopup wires the same chrome ids
the previous commit taught DrawScrollablePopupPlain to draw, for both
style="plain" and style="retail" markup menus.
<list> markup (UiMarkupList.cs / MarkupDocument.cs): a plugin <list>
(single-column or <column> multi-column) that overflows its own row
viewport now draws the retail scrollbar chrome at its right edge (VVS's
own placement, 16px wide) instead of being wheel-scroll-only with no
visible bar. The reserved 16px column only exists while rows actually
overflow, in both column-layout modes (ComputeColumnLayout receives the
already-shrunk width so the last/auto column absorbs the remainder
correctly); the bar is fully interactive (up/down arrows, track paging,
thumb drag) via a small UiScrollable projection kept in sync with the
list's own _topRow, which stays the single source of truth. Wheel
scrolling and a no-resolver hand-built list (draws nothing, no crash) are
unchanged.
Mutation shown to fail first: new
UiMarkupListScrollbarTests/MarkupDocumentTests cases were written against
pre-change UiMarkupList/MarkupDocument and failed (no scrollbar sprites
ever emitted since UiMarkupList had no SpriteResolve property at all, and
<menu> markup never set Scrollable) before the implementation landed;
after: SingleColumn_Overflowing_DrawsRetailScrollbarChromeAtRightEdge and
Columns_Overflowing_ReservesSixteenPixels_LastColumnShrinksAccordingly
pin sprite ids + exact reserved-width geometry,
*_ContentFits_DrawsNo(Scrollbar|ReservationLastColumnKeepsFullRemainder)
pin the no-overflow/no-bar case, *_UpArrowClick_ScrollsUpByOneRow and
ThumbDrag_MovesTopRowAndIsReadableByASubsequentClick pin interactivity via
a following row click resolving to the moved position (mirroring
MarkupListColumnsTests' own wheel-scroll pin), and the four new
MarkupDocumentTests menu cases pin Scrollable/PopupScrollbarHideWhenDisabled/
the six chrome-id properties plus an end-to-end open-popup draw for both
the overflowing (draws chrome) and non-overflowing (draws none) cases.
Every pre-existing MarkupListColumnsTests/MarkupDocumentTests case stays
green unchanged (none of their fixtures overflow their own viewport).
docs/plugin-ui-markup.md updated: the <menu> style paragraph and a new
<list> "Scrollbar" section describe the new chrome + auto-reservation, and
the PITCH-transcription guidance is corrected to say the 16px scrollbar
column is now automatic (no more manual fold-in/double-reservation advice).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner live-client report 2026-09-07: "For scrollable dropdown or the meta
window we use the same assets as we do in for example chat or inventory
window." The plain-style <menu> popup's scrollable-overflow scrollbar
(DrawScrollablePopupPlain / DrawPopupScrollbarPlain in UiMenu.cs) drew a
home-made flat 1px track + flat thumb instead of the gold track + up/down
arrow buttons + thumb the chat SpewBox and inventory UiItemList already
use through RetailScrollbarChrome. The owner only ever objected to the
retail ROW art (checkmark glyph, gradient panel) — the bar itself was
never in scope for the plain-row fix, so this change touches only the
scrollbar draw call and leaves the plain row rendering untouched.
DrawScrollablePopupPlain now calls the existing DrawPopupScrollbar helper
(the same procedural sprite-chrome draw VendorUiController/
ConfigOptionsPageController already use) whenever a SpriteResolve is
wired, falling back to the old flat DrawPopupScrollbarPlain only for a
hand-built UiMenu with no resolver at all. New
RetailScrollbarChrome.ApplyToMenuPopup(UiMenu) wires the same vertical
skin ids (Track/Up/Down/ThumbTop/Mid/Bot Normal) the chat/inventory
scrollbar uses onto a menu's own ScrollTrackSprite/etc properties.
Mutation shown to fail first: UiMenuPlainStyleTests's
Plain_OpenPopup_ScrollableOverflow_DrawsPlainTrackAndFlatThumb_NoDatArt
and Plain_ScrollablePopup_ContentFits_DrawsTrackWithNoThumb asserted
resolveCalls==0 and an all-fill scrollbar — both failed (6 resolve calls,
6 sprite quads instead of 0) against the new DrawPopupScrollbar call
before being rewritten to
Plain_OpenPopup_ScrollableOverflow_DrawsRetailScrollbarChrome_RowsStayPlain
and Plain_ScrollablePopup_ContentFits_DrawsNoScrollbarAtAll, which pin the
new sprite-chrome behavior (6 resolved ids on overflow: track, up, down,
thumb top/mid/bottom; 3 on content-fits: track+up+down, no thumb; 0 on a
menu built with no resolver) while re-asserting the rows are still plain
fills with zero retail row-sprite quads. Retail's own
RetailButtonArt=true popup path (DrawGridPopup/DrawScrollablePopup) is
untouched — its regression golden
(Retail_OpenPopup_DrawIsByteForByteUnchanged_RegressionGolden) still
passes byte-for-byte.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
BuffSettings.ExtraBuffSpellNames/BlacklistedBuffFamilyNames and their two
Buffs-tab tooltips already documented that BuffPlan.Build does not
consume these sets yet, but didn't name the tracked issue. Both field
doc comments and both tooltip strings now cite #491 (filed on the
campaign branch: "shown and persisted, not consumed by BuffPlan.Build
until slice 4"), so anyone hitting the gap has a concrete issue to read
instead of just a prose warning.
Text/comment-only change; no test behavior to pin. Full solution build
green, MossTank suite 678/678, App markup/plugin filter 203/203.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Extends the resizable="true"/minw/minh markup grammar's coverage past
MarkupDocument's own parse tests to the two host seams a resizable
plugin panel actually flows through, proving no host-side wiring beyond
what MarkupDocument.Build already sets on the panel was needed:
- RetailWindowManagerTests: a resizable="true" markup panel accepts
RetailWindowManager.ResizeTo within its minw/minh floor (and clamps
to it below the floor); a plain (non-resizable) markup panel refuses
— Width/Height unchanged and no Resized event, exactly today's
fixed-size behavior.
- RetailWindowLayoutPersistenceTests: a resizable panel's dragged size
round-trips through save/restore into a fresh session, and a saved
size below the panel's CURRENT minw/minh floor (a legacy save, or a
plugin update that raised its floor) clamps UP to the floor on
restore rather than restoring the too-small legacy value.
Mutation proof: reverted MarkupDocument.cs to its pre-feature state and
reran the new tests — the two RetailWindowManagerTests cases failed
(the old UiNineSlicePanel ctor default of Resizable=true/MinWidth=40
let the "fixed" window resize and let the "resizable" window shrink
below the new floor), and the persistence floor-clamp case failed
(80x60 came back instead of clamping to 200x150). The plain
save/restore round-trip case passed either way — the old ctor default
was already resizable, so it exercises a real but coincidentally
already-covered path; kept for its own documentation value. Restoring
the implementation returns 256/257 (1 pre-existing unrelated skip) on
the full Markup/PluginSidePanel/RetailWindow/Anchor-filtered App suite
and 9/9 on the MossTank markup-filtered suite, both green.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
SecondaryPopupPanelsFitTheirOwnBoundsAndEveryBindingResolves was a
[Theory] over a hand-maintained [InlineData] file list — every popup
added or renamed since (buffpicker, metaeditor) already needed a manual
row here, and a forgotten one would leave a new popup completely
uncovered with no test failure to flag it.
Converted to a single [Fact] that discovers popup files with the exact
same Directory.GetFiles(AppContext.BaseDirectory, "mosstank*.xml") glob
AuthoredControlsInTheSameContainerNeverOverlapASibling and
NoButtonAnywhereUsesTheUnrenderableArrowGlyphs already use (excluding
mosstank.xml itself, the main panel covered by its own tests). Expected
width/height per file now lives in an ExpectedPopupBounds dictionary; a
discovered file with no entry fails loudly, naming the file, instead of
silently going unchecked.
Mutation named: temporarily commented out the mosstank-buffpicker.xml
entry and confirmed the test now fails with the new "has no expected
width/height entry" message (rather than the old behavior, where an
unlisted file was simply never discovered) before restoring the entry.
MossTank suite 681 -> 678 (the four [InlineData] cases collapse into one
[Fact] with the same coverage; net -3 test count, not a coverage loss).
Full solution build green; App markup/plugin filter 203/203.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
ItemHandsColumn ran a fresh .Select(...).ToArray() over _itemRows, PLUS a
per-row BaseItemName suffix-strip to recover the undecorated name, on
every single retained-UI draw. Folded per item F12: RefreshItemEditors
now captures the raw SortedCombatItemNames() result once as
_itemBaseNames — the SAME array used both to build the decorated
_itemRows (append " [no buffs]" where it applies) and to compute
_itemHandsColumn directly, so the " [no buffs]" suffix has exactly one
definition (added going forward) instead of two (added in RefreshItemEditors,
parsed back off in the old ItemHandsColumn getter).
CycleItemHandsAtCore (the grid's own "click cycles handedness" cell
action) mutates _itemHandedness but did not call RefreshItemEditors —
unlike the Monsters grid's mutators, which all already refresh after
mutating. Added the call so the cached column stays correct; also
switched it to read the cached _itemBaseNames instead of a second
SortedCombatItemNames() call.
Mutation named: temporarily restored the old live-recomputing
ItemHandsColumn getter (with a local copy of the suffix-stripping helper)
and confirmed the new
ItemHandsColumnDoesNotReallocateOnEveryReadAndNoBuffSuffixNeverLeaks pin
fails (Assert.Same throws — different array instances per read) before
restoring the fix. The same test also proves the no-buffs case still
resolves handedness correctly through the folded base-name array.
MossTank suite 680 -> 681 (one new pin). Full solution build green; App
markup/plugin filter 203/203.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Owner direction (2026-09-07): plugin panels need a bigger default size
and real resizability. Plugin markup panels were fixed-size with no way
to opt in to drag-resize, and only <meter> exposed an anchor attribute
(comma-separated, silently dropping unknown tokens) — no other element
could stretch or reposition when its window resized.
<panel resizable="true" minw= minh=> is now the explicit opt-in (default
false — a panel with none of these attributes gets Resizable=false,
ResizeX=false, ResizeY=false, matching every plugin panel shipped today,
e.g. mosstank.xml's resize="none"). resizable="true" arms both axes and
defaults the min size to the authored w/h so a resizable panel never
shrinks below the layout its author tested; the pre-existing resize=
attribute still narrows to one axis on top of that.
anchor="left top right bottom" (space-separated, case-insensitive) now
applies uniformly via ApplyCommon to every element (<group>, <list>,
<menu>, <field>, <label>, <button>, <icon>, plus <meter>/<tab>/<toggle>/
<slider> for free) instead of just <meter>'s own comma-separated,
non-throwing parse. An unknown token now throws FormatException naming
the element, matching this file's "malformed markup throws at Build"
convention everywhere else. No new plumbing is needed for live re-layout
or group-relative child anchoring — UiElement.ApplyAnchor/AnchorEdges
already measure a child's margins against its own direct Parent's
Width/Height every draw, and RetailWindowManager.ResizeTo/UiRoot's
existing edge-drag resize already respect Resizable/ResizeX/ResizeY/
MinWidth/MinHeight generically for any registered window.
Mutation proof: reverted MarkupDocument.cs to its pre-change state and
reran the 25 new MarkupResizableAnchorTests — 17 failed (the anchor
grammar, resizable/minw/minh parsing, live re-layout, and golden-draw
tests), 8 passed trivially (cases asserting the unchanged no-attribute
default). Restoring the implementation turned all 25 green with no
regression in the existing 227 Markup/PluginSidePanel/RetailWindow/
Anchor-filtered tests (252/253, 1 pre-existing unrelated skip).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
AssertWithinParent only flagged a child crossing a group's bottom edge
when that child declared its own h attribute — Number(child, "h") reads
0 for an absent attribute, so an unsized <label>/<field>/<toggle>/
<button> positioned right against a group's bottom edge silently passed
even though the retained-UI runtime still gives it a real default row
height at draw time. Added EffectiveHeight: falls back to each widget
kind's own implicit default when h is absent (label/field 16, toggle 20,
button 16 as a defensive floor since real buttons in this markup range
16-25px and always declare h explicitly); <list>/<menu> keep the old
"0 when absent" behavior since they have no implicit default at all.
Mutation named: temporarily reverted the check back to Number(child, "h")
and confirmed the new AssertWithinParent_CatchesAnUnsizedLabelNearTheBottomEdge
pin fails (no exception thrown for a synthetic <label y="190"> with no h
inside a 194-tall group, which should clip 12px past the bottom once the
16px default is counted) before restoring the fix. Audited every real
mosstank*.xml file for label/toggle/field/button elements missing h
(grep for each without ` h="`) — none exist, so this stricter check
introduces no new failures against the shipped markup.
MossTank suite 679 -> 680 (one new pin). Full solution build green; App
markup/plugin filter 203/203.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
EnsureDefaultMonsterRule() is a _combatSettings.Rules mutator (it appends
a DEFAULT row when the list is empty), but the /mt refresh command
handler never called RefreshMonsterEditor() afterward, unlike every other
Rules mutator in the plugin (AddMonsterRuleCore, DeleteMonsterRuleAtCore,
MoveMonsterRuleAtCore, UpdateMonsterActionsAt, the constructor, and
ResetProfileConsumers's own EnsureDefaultMonsterRule call).
Audited every _combatSettings.Rules mutation site in the plugin (grep for
"_combatSettings.Rules" across all of src/AcDream.Plugins.MossTank):
AddMonsterRuleCore, DeleteMonsterRuleAtCore, MoveMonsterRuleAtCore, and
UpdateMonsterActionsAt already call RefreshMonsterEditor right after
mutating; the constructor and ResetProfileConsumers already pair their
own EnsureDefaultMonsterRule call with one. The /mt refresh handler was
the only gap.
Mutation named: since Rules can never actually be observed empty through
the panel's own public surface (delete refuses removing DEFAULT, and
every profile-load path already re-adds it via ResetProfileConsumers
before this handler could see it), the new pin reaches the private
CombatSettings instance via reflection to clear Rules directly, and also
pokes the cached _monsterNameColumn field to a "STALE" sentinel first —
otherwise EnsureDefaultMonsterRule() re-adding "DEFAULT" would coincide
with the grid's already-cached construction-time value and the test
would pass even with the fix missing. Confirmed it fails (shows "STALE"
instead of "DEFAULT") with the RefreshMonsterEditor() call removed,
before restoring the fix.
MossTank suite 678 -> 679 (one new pin). Full solution build green; App
markup/plugin filter 203/203.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
AdvancedOptionCategoryEnabled returned _advancedOptionCategoryEnabled (the
panel's own mutable bool[]) directly. Any caller holding the returned
IReadOnlyList<bool> could cast it back to bool[] and flip category flags
without going through ToggleAdvancedOptionCategoryAt — bypassing the
clamp/cache-refresh that action performs. The property now returns a
ReadOnlyCollection<bool> view built once over the same backing array
(built once since the array's own reference never changes, only its
elements, so no re-allocation is needed on every read): a cast back to
bool[] now throws InvalidCastException instead of handing out a mutable
reference.
Mutation named: reverted the property to return the raw array and
confirmed the new
AdvancedOptionCategoryEnabledIsNotTheMutableBackingArray pin fails (no
exception thrown) before restoring the fix.
MossTank suite 677 -> 678 (one new pin). Full solution build green; App
markup/plugin filter 203/203.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
AdvancedOptionNames, AdvancedOptionValueColumn, AdvancedOptionName, and
AdvancedOptionDescription each re-ran FilteredAdvancedOptionNames() (a
Where over the full 163-entry VtankOptionCatalog) on EVERY retained-UI
draw of the popup, and the value column additionally ran
GetMetaOption(name).ToDisplayString() per row (Trim()/ToLowerInvariant()
allocations) every frame. All four are now materialized once into fields
by a new RefreshAdvancedOptions(), called from every real mutation point:
category toggle, edit/apply, selection change, and profile
load/create/clear/delete (via ResetProfileConsumers), plus construction.
Same pattern fix round B item 12 used for the Monsters/Meta/Route grids.
AdvancedOptionValueColumnMirrorsTheLiveSettingValue exercised a settings
change made OUTSIDE the popup's own mutators (ToggleCombatEnabled, a
main Options-tab checkbox) and expected the value column to reflect it
on the very next read — a real behavior this caching model intentionally
narrows (mirrored settings now catch up at the next real popup mutator,
not on every read). Updated the test to re-select the row afterward,
exercising the "selection change" refresh point, and documented why.
Mutation named: reverted the two property getters to their old live-
recomputing form and confirmed the new
AdvancedOptionsPopupBindingsDoNotReallocateOnEveryRead pin fails
(Assert.Same throws — different array instances per read) before
restoring the fix.
MossTank suite 676 -> 677 (one new pin). Full solution build green.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
