Commit Graph

429 Commits

Author SHA1 Message Date
daniel 9347c861a7 Merge plugin-editor drop: a card dragged onto a third-party plugin's editor is handed to the OS as a file drop 2026-09-25 14:15:47 -04:00
daniel 7e523b0505 Clean up stale editor-hInstance comments, add isOwnModulePath edge tests
Drops a false claim about ReaSampler's own editor's hInstance, corrects the real reason for registering the wndproc class against our own module, and covers isOwnModulePath's separator edge cases.
2026-09-25 14:15:36 -04:00
daniel faa513cafb Fix SWELL drag guard, channel-fork the VST editor window class, dedupe own-module match
Panel drag no longer misfires on SWELL's always-zero wParam; stable/beta VST editors register distinct window classes against their own module instead of colliding on REAPER's; own-module basename matching moves to a tested pure helper.
2026-09-25 14:11:22 -04:00
daniel 4567744b8f Fix spline-drag rollback, bake contour leak, held-note live miss, amp clamp
Editor close mid-drag now rolls back published contours like capture-loss does; bake detaches liveContours so a printed capture can't observe a live drag; a held note now receives a live publish that lands in the same block as a note-on; amp spline read clamped to [0,1].
2026-09-25 13:59:37 -04:00
daniel fee941a83d Exempt ReaSampler 9000's own editor from the plugin OS drag hand-off
A dragged bank card over our own instrument's editor was OLE-dropping like a third-party plugin. Classify our channel's own .vst3 module as Instrument, not Plugin, so it keeps the normal instrument-drop outcome.
2026-09-25 13:57:11 -04:00
daniel f7505d0e53 Make spline EG drags audible live: held and new notes follow a dragged knot
Contours ride their own seqlock block into a double-buffered engine copy, and the swap step is smoothed. Each drag move publishes contours without a model write; release still commits and reloads.
2026-09-25 13:42:25 -04:00
daniel fe1aec5925 Hand a card dragged onto a plugin's own editor to the OS as a file drop; withdraw the OLE drag when the pointer returns to REAPER
A code-ownership proof (window class procedure's module, via the pure window_origin) plus REAPER's FX-surface token routes a plugin editor to OLE; FX chrome keeps the instrument drop. The OLE loop re-asks the law each turn, so passing a plugin editor stays reversible.
2026-09-25 13:39:39 -04:00
daniel 3fa4d10b91 Lift the sign-off DRAFT note, fix footer/cursor CLAUDE.md overclaim, honest-en the readout-width test, trim duplicated rationale
Removed DRAFT marker on approved readout strings; corrected "paint and cursor cannot disagree" to note drag-cue suppression; renamed the width test to state it only bounds character count, not pixels; trimmed restated rationale to its CLAUDE.md/header home.
2026-08-21 15:59:32 -04:00
daniel 052fbb0d52 Make the background FX drain visible: a busy cursor over the panel and a footer readout that names its direction, driven off fxReclaimActivity 2026-08-21 14:45:25 -04:00
daniel f6f09d983c Close the FX reclaim span when REAPER's own Undo/Redo is dispatched, so the switch's point mints first
Both command hooks intercept Edit: Undo (40029) and Edit: Redo (40030) and consume
nothing. Undo routes that never dispatch an action remain uncovered.
2026-08-09 13:40:28 -04:00
daniel 20fef2a3bf Keep the extension's tick out of REAPER's own project load, and re-earn the park holds across one load 2026-08-07 18:03:58 -04:00
daniel 3d95eb715c Undo-block review fixes: exception-safe depth, corrected docs, real test teeth
Fixes ctor exception safety, a ValidatePtr2 guard, false SDK-citation and
enforcement claims, the span-depth invariant, and vacuous fx_reclaim tests.
2026-08-07 17:07:56 -04:00
daniel b7b93a7ac3 One undo-block opener for the whole tree: fourteen sites route through it, and the reclaim span can no longer swallow a user's act
The drain refuses a tick inside a foreign block; applyMode wakes the reclaimer
before it opens the span.
2026-08-07 16:38:11 -04:00
daniel 39529a5ac4 Design View: drop the undo blocks around writes that mint nothing; the membership actions own theirs
Flag writes are silent from a timer, so a block around a reapply or the reconcile
repair could only manufacture a point. The real switch's span is unchanged.
2026-08-07 14:41:17 -04:00
daniel 67d9ec3da9 Put the FX reclaimer to sleep on a project-state load instead of waking it
abandon() clears the wakeup where invalidate() raises it, so File > Revert
cannot converge FX or spend park records against a model nothing re-read.
2026-08-07 14:23:56 -04:00
daniel 9b6f725747 Close the FX reclaimer's undo span at the state-load hook, and expose the drain's direction
A Ctrl-Z mid-span used to close the pre-undo block over post-undo state; it now
terminates at BeginLoadProjectState and stops the drain where it stands.
fxReclaimActivity reports Online/Offline/Idle; fxReclaimPending is unchanged.
2026-08-07 13:45:58 -04:00
daniel 86aa8dabba Collapse a Design View mode switch and its background FX parking to ONE undo point
applyMode now returns with its undo block open; the reclaimer closes it when the drain
settles. A transport deferral or a tab switch ends it early -- two points, never a leak.
2026-08-07 12:53:42 -04:00
daniel 4f6cf60740 Narrow the FX reclaimer's transport gate to defer offline work only; let online drain through, and always run the park clock's observe 2026-08-06 23:04:02 -04:00
daniel c95203718a fx_reclaim: use GetCurrentProjectInLoadSave for the save hook, stop disarming ReleaseFlush on one write, validate project before cached-track lookup 2026-08-06 22:52:57 -04:00
daniel 02790f1270 Gate the FX reclaimer on the transport, both directions; move the release's persist off the drain and onto the project save
The drain no longer dirties a project on its own initiative — a spent park record rides the user's next save via OnSaveExtensionConfig.
2026-08-06 22:31:47 -04:00
daniel 224dcf84df Fix the reclaimer stranding a tab-switched-into project: lift the pin and wakeup into FxReclaimQueue, retarget instead of going idle
Also validate the cached track in its project, and stop a drop-free tick resetting the report memo.
2026-08-06 22:16:41 -04:00
daniel 7687ea76fc Persist the reclaimer's FX release so a reload cannot re-drive a spent record, and report the FX-restore drops it was swallowing 2026-08-06 21:26:55 -04:00
daniel 5dcfd3a53f Reclaim drain owns its work: invalidation only bumps a generation, and a generation that moved aborts the tick's commit and its release. 2026-08-06 21:08:34 -04:00
daniel f04c704353 Make an FX reclaim step one whole track's chain: drop the resume cursor, yield only between tracks 2026-08-06 20:48:17 -04:00
daniel 5a35b71dc0 Add the FX reclaimer: a debounced, time-boxed convergence loop that offlines parked chains and restores returning ones off the switch path 2026-08-06 19:07:17 -04:00
daniel 0f6df5fd6d Design View: expose a park-shaped no-record track visible instead of adopting or refusing it
Adopting silently wipes hand repairs or hides untouched tracks forever; refusing was rejected too. EXPOSE hands the track back and self-heals on the next apply.
2026-08-06 18:35:24 -04:00
daniel 024c619884 Fix Wave-2 review findings: FX union on re-park, per-project reconcile memo, shared undo mask, decayed test oracle 2026-08-06 17:50:49 -04:00
daniel c4038a9e9c Strip FX writes from the Design View mode switch; split view.cpp into park/lanes/reconcile/tracks and add the record-scoped undo repair pass 2026-08-06 17:19:43 -04:00
daniel b064081a3c Fix five Wave-1 pure-layer FX/park review findings: resume-point stall, stale doc comments, implicit mode read, keying reset 2026-08-06 16:50:43 -04:00
daniel 2423225f0c Split FX out of the park plan: flag-only plans, derived online-first FX work list, two-half record release, applied-state fingerprint 2026-08-06 16:31:59 -04:00
daniel 8c06383fe7 Collapse a Design View mode switch to ONE undo point: FX writes run inline in applyMode's block, deferred park queue deleted 2026-08-06 04:32:26 -04:00
daniel 8a7056d19d Fix undo-label misnomer, tighten drain-nesting/abort-path docs, mark inferred undo behavior as unverified
Renamed the drain's undo point to "Design View FX state" (it restores as often as it parks). Extended the nesting audit to cover a block opening inside the drain's own block. Marked post-fix undo behavior as inference pending DAW checks.
2026-08-05 20:54:31 -04:00
daniel 13c824c63d Collapse the FX-park drain's writes into one undo point instead of one per TrackFX_SetOffline 2026-08-05 20:40:45 -04:00
daniel bd86ede5b3 Merge FX-park snapshot guard: refuse to park when the pre-park truth is gone
A snapshot is never taken from a chain a park has touched, on either
half -- flags or per-FX offline. Stranded tracks are refused, named
once in the console, and left recoverable by hand.
2026-08-05 20:17:00 -04:00
daniel 4b306dd436 Name refused tracks before Undo_EndBlock2, not after; tighten the route-list doc; pin the refusal memo's gate
Fixes D (handle lifetime), C (ASCII dashes), G (comment compression), I (shouldReport extracted + tested); CLAUDE.md route list corrected per A/B/H. E pushed back -- cross-module GUID plumbing is architectural, not a targeted fix.
2026-08-05 17:34:49 -04:00
daniel 5376ab085c Extend the park trust test to the FX chain, name the refused tracks, and stop reprinting an unchanged refusal 2026-08-05 17:10:19 -04:00
daniel 67524206fb 2nd Pass Render 2026-08-05 16:21:17 -04:00
daniel 203961f41c Refuse to park a track whose pre-park snapshot is gone, rather than re-snapshotting the already-parked chain as the user's state 2026-08-03 18:30:40 -04:00
daniel 2f96dd3da5 Revert Ω-W2-T4: FX-park state stranded across save/reopen
Backs out the deferred park drain split. Forced synchronous FX-park
drains return, and with them the mode-switch hitch, until the
re-snapshot-after-discard fix lands.
2026-08-03 18:05:04 -04:00
daniel 8487833342 Revert "Merge Ω-W2-T4: split the deferred FX-park drain by kind"
This reverts commit 70bd29dc89, reversing
changes made to 84afd1b545.
2026-08-03 17:46:17 -04:00
daniel d22446ef7d Move Re-Capture and Insert Conform from the panel top bar to the overflow menu; re-pin action_bar width test to the new 4-button top bar. 2026-08-03 17:44:00 -04:00
daniel 4b19556513 Merge Insert as FX: load ReaSampler 9000 onto the selected track from the panel
# Conflicts:
#	src/shell/actions/CLAUDE.md
2026-08-03 16:46:40 -04:00
daniel a5febbd6aa Merge drag-out hand-off fix: keep the drag internal while the pointer is over REAPER 2026-08-03 16:45:14 -04:00
daniel 51c7505dc0 Fix Insert-as-FX review findings: id rename, selection-count drift, refusal tooltip, success msg, master-only, reserve test
Renames the permanent action id out of the placement family, makes the button's
painted state and pressed outcome share one selection count, surfaces
panel-known refusals in the tooltip, reports the target track on success,
distinguishes a master-only selection, and pins the overflow-reserve test to the
real bar spec.
2026-08-03 16:43:10 -04:00
daniel b71a05fbef Close the bridged-process hole in the drag-out hand-off gate
OsHandoff now needs the window-ownership proof AND a hit-test that named
nothing, so a bridged plugin's UI can't read as off-REAPER.
2026-08-03 16:42:59 -04:00
daniel aa87d81f1f Merge Ω-W2-T5: bound the waveform zero-crossing snap to a pixel radius
Single-cycle loop marks stop teleporting to the one interior crossing. Ctrl on a
marker drag defeats the snap, and no longer loses the grab to a coincident node.
2026-08-03 16:09:27 -04:00
daniel 70bd29dc89 Merge Ω-W2-T4: split the deferred FX-park drain by kind
Restores stay forced before the view model serializes; parks go one FX per idle
tick behind a 1s coalescing debounce, so a rapid A-B-A flip costs no plugin work.
2026-08-03 16:09:20 -04:00
daniel f91054276b fix(instrument): stop Ctrl-before-click stealing a waveform mark grab from a coincident node
Ctrl held before mouse-down forced the node/toggle win over a smaller cap or
column regardless of area; now it defers to the ordinary smallest-area
arbitration like a plain click. Also amends product docs, VERIFICATION.md, and
adds sparse-material/narrow-overlay test fixtures.
2026-08-03 16:07:49 -04:00
daniel fcd1ed022c feat(panel): add Insert as FX — ReaSampler 9000 onto the selected track, preloaded with the focused capture, over the existing instrument-drop body 2026-08-03 16:06:22 -04:00
daniel 9599e96e0c Fix drag-out giving live drags to OLE: prove the pointer left REAPER
OffReaper leaves the surface vocabulary; OsHandoff now gates on a window-ownership
proof resolved in the shell, so no in-REAPER surface can reach DoDragDrop.
2026-08-03 15:50:25 -04:00