'Delete' keyboard shortcut does not work in diagrams #15
No Assignees
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: sorokin.oleg.edu/ARCADIA-Studio#15
Reference in New Issue
Block a user
You can't delete an element (either from model or from diagram) by pressing 'Delete' button
This is a part of #1
Codex-956atf /root claims this bounded diagram keyboard outcome on codex/behavior-authoring-20260910, alongside existing #3. Initial source shows Delete handles selected route bends only; model and placement removal already have separate commands. I will define predictable diagram-focus Delete behavior using existing removal/confirmation flows, preserve text editing and route-bend priority, and verify actual App-tree dispatch plus undo/source invariants. No overlapping implementation claim appears in current issue comments; other sessions retain their existing work. Initial investigation is read-only while the allocation gate is frozen.
Codex-956atf /root resumed on the user's explicit current instruction to continue. This supersedes my pause notice. Existing ownership remains: codex/behavior-authoring-20260910, allocation source3840 verified796 tests, separate Delete candidate7dcfb3c. Next: correct live-drag outside wording, independently review and integrate Delete/focus changes preserving the allocation toolbar, then execute combined Windows verification. Issue15 acceptance remains actual focused removal with correct placement/model semantics, editor protection and undo/source consistency across diagram types; current architecture/state candidate alone does not satisfy the entire issue. Linux/WSL remains deferred under the separate networking instruction. No new overlapping owner appears in current comments.
Codex-956atf /root: integrated architecture/state candidate60ea483; focused Windows diagram_removal_tests passed10/10 at source0d6f677 with all529 inputs unchanged. Independent review found placement-properties identity mismatch;
dc1506ffixes Delete/ShiftDelete from placement-ID selection and adds OA/SA mounted coverage through the generated menu action and blank-canvas focus. Full807 desktop suite is running ondc1506f(not yet terminal). Separate worker implements process/scenario occurrence removal with confirmed semantic scope, reusable-target preservation, qualified-process-selection protection and captured owner identity. State-action/internal-transition row gaps are being mapped. Old stop notes are superseded/corrected locally; work continues. No issue closure or native acceptance claimed.Codex-956atf /root: full
dc1506fgate closed805pass/2fail; both failures were legacy raw Canvas F2 tests asserting obsolete private focus ownership. Preserved rename/history checks and moved focus refusal coverage to actual Mounted App; source28e21a0 passes19 diagram tests, including real F2/Inspector/Tab and state Entry/Do/Exit/Internal-row relation deletion with exact Undo/source preservation. Derived process/scenario candidate c2e651b integrated49a4cb1. Root12b2237 wires captured removal explanations and actual rendered confirmation-button tests;2d9e9c1aligns guide/reST8.1. Independent derived review found no blocking source defect, and is strengthening real Inspector-click/keyboard-reentry evaluators before next executable gate. These changes remain unaccepted until combined full/strict/GUI/native checks; no issue closure. Tests serialized; Linux untouched.Codex-956atf /root checkpoint: published
42511f0f8aon codex/release1-956atf and verified the remote SHA. Full source4c387cc closed817 passed/1 failed with531 inputs unchanged; all new architecture/state/process/scenario removal cases passed. Sole failure was evaluator sequencing: it deliberately retained a focus Apply then expected another F6 while that operation was pending.42511f0explicitly verifies pending F6 refusal, completes native synchronous Apply, and retains the same operation for all stale chooser/review ownership replay assertions. Production focus protection unchanged. Complete suite running on42511f0; full/strict/GUI/native acceptance remains open. Exact earlier failed and passing receipts plus reviews archived in harness/experiments/diagram-removal-windows-20260912 and linked release-diagram-removal-windows-20260912.md. Linux remains deferred.Codex-956atf native checkpoint: published source
f7d84fapasses strict all-target Windows desktop Clippy and GUI build (531 unchanged identified inputs); full 818-test pass remains precisely scoped to42511f0with subsequent verified test-only/dead-accessor corrections. Native f7 binary SHA256 adea83fd13b68a517bb43067169e492c67ed3c9364ffbb3aa8896d98596a3612: actual plain Delete removes only the selected placement; Save and one-step Undo/Redo match the complete independent baseline-derived Project and exact two-file SourceBundle. Model deletion Cancel and confirmed delete/Undo also pass exact checks. Actual Save/Undo then Shift+Delete revealed lost canvas focus until a click. This is a confirmed remaining interaction defect, currently being corrected in isolated codex/delete-undo-focus-956atf-20260912 with mounted-tree tests and a fresh critic; native acceptance and issue closure are withheld. Current root owns native input; Linux remains deferred.Codex-956atf checkpoint: normal push and remote readback confirm
e2c027e54con codex/release1-956atf. The focused Windows gate at24bb248closed 22 passed / 1 failed with 531 unchanged inputs: architecture/allocation Save/Undo and suspended-focus/gesture cases passed; persisted state placement Save/Undo still lost its actual keyboard owner. Source investigation identified positional Iced child replacement when the unplaced-element picker appeared/disappeared.e2c027eretains a stable canvas child slot and strengthens the same failing workflow's intermediate focus checks; it has not passed runtime acceptance yet.The combined source also includes #21 activation rendering and #25 redundant control removal. Strict all-target Clippy at
e2c027eclosed with two production-unused Focus event variants left by #25 (533 unchanged inputs). I am removing the stale panel-mode paths and migrating their tests, then running fresh combined tests/lint/GUI and native regression. No warning suppression, passing gate, issue closure, or completed release is claimed. Failure receipts and prepared bound OA/SA scenario fixtures are being preserved in versioned evidence. Linux/WSL remains deferred.Codex-956atf checkpoint: combined source
83f53c7c6epasses all 834 Windows desktop tests, strict all-target Clippy with warnings denied, and matching GUI build. All 533 input files agree across the closed gates and remained unchanged. This includes the previously failing persisted state-placement Save/Undo focus regression. Portable receipts and reviews are published at419c78con codex/release1-956atf in harness/release-canvas-followup-20260912.md; 156 archive payloads verified against Git-index and workspace bytes. Fresh native acceptance is underway using the identified staged executable and disposable source fixtures, with exact Project/SourceBundle checkpoints. This is not final native/release acceptance. Linux remains deferred.