feat(controls): add per-control reset and confirm binding changes #2126

Merged
HumanBean merged 5 commits from Phishinabowl/Longshot:feat/improved-control-defaults into main 2026-11-21 22:23:24 +00:00
Member

Summary

Controls now lets you reset a single row. The square curved-arrow button is available when the row's gamepad or keyboard bindings, or its Toggle, differ from the current defaults. The table headings sit below Layout preset and Button prompts. Resetting a row or all controls asks for confirmation. When a new input is already in use, the dialog names the affected control before reassigning it.

The grid gives Toggle and Reset a little more space after Keyboard. With a controller, Clear uses the available top face button or another free face button if Bind or Back uses it. The footer shows the assigned button, and keyboard input shows Del.

On main, open Controls → Controls and move down in Toggle from ZR toward Plus and Minus. Navigation reaches a row without a Toggle and moves focus to Keyboard. With this change, Up and Down stay in Toggle or Reset and skip unavailable cells. Left and Right change columns.

Start with gpu/vk/vk_menu.cpp for focus and confirmation, and host/nn_hid.c for the effective defaults and binding updates.

Testing

Tested:

  • Windows and native laptop Linux at 42841c4d before rebase. Both built and passed 12 focused game-free checks each. In game, Up and Down skipped disabled cells without leaving Toggle or Reset. Cancel preserved the binding. Confirm restored the default and disabled the reset arrow.
  • Windows and Linux game-free comparison on 971f77f5 and 3cbdeef1 plus the layout and fixture changes. The same navigation check missed both expected Toggle targets on main and reached both on the branch.
  • Seven focused Linux checks passed, including the input-recording and output-preview fixtures. Six focused Windows checks passed. The Windows input-recording fixture also fails on clean main on LLVM-MinGW 23.1.5, where the compiler reports that tp_cancels is set but unused.
  • Strict laptop Linux comparison before the spacing and Clear edits: main 813 passed, 106 failed, 18 skipped; topic 814 passed, 105 failed, 18 skipped. The topic has no additional failing names and passes test_menu_keys.py, which fails on main.
  • Earlier Windows strict runs: main 777 passed, 107 failed, 50 skipped; topic 769 passed, 115 failed, 50 skipped. Both timed out in test_vbytes.py. Eight branch-only failures passed on retry: test_mesh_first_types.py, test_import_windows.py, test_cbuf_table_extent.py, test_dump_ssbo_read.py, test_mesh_dynamic_state.py, test_timeline_modes.py, test_capture_copy_nt.py, and test_capture_copy_runs.py. The ninth, test_hos_ab.py, now passes after the test was updated for the appended ABI callbacks. The final Windows focused rerun passed six checks; Ruff passed for the three changed Python files.
  • Final spacing and Clear update: four focused game-free checks passed on Windows and laptop Linux. Both isolated runtimes used their existing executables and rebuilt native renderers. Manual play-tests passed on both platforms for spacing, controller Clear and the matching footer glyph. The laptop play-test also confirmed the return to Del under keyboard input.
  • Final runtime hardware: Windows 10 build 19045 with an RTX 3080 and NVIDIA 621.24; native Ubuntu with an RTX 5090 Laptop GPU and NVIDIA Vulkan driver 0x94d8c080.
  • Reviewed the matched UI-only Controls captures from the pinned main and the final source, including the Classic theme and controller footer. A Graphics comparison reproduced identical pixels; the guide source review and offline check passed. The captures use synthetic bindings and contain no game data.
  • BeanBot follow-up at 3541a0d2, rebased onto 8aa74983. The Controls help and its existing translations now explain that confirming a conflict removes the previous binding. The new fixture executes the actual menu confirmation handlers for row reset, Reset All and conflicting rebind. It checks cancellation and acceptance against bindings and Toggles. A mutation that accepted Cancel failed at the expected assertion.
  • The final focused Windows and Linux runs each passed seven checks covering Controls, ABI and output preview. Windows also reproduced main's test_pads_after_menu.py and test_xinput_thread.py unused-variable failures. The translation-key check fails on the same unrelated Output resolution string on both branches. Ruff and the offline Graphics guide source check passed. The refreshed Graphics captures differ only by the version label.
  • Live output preview passed on Windows with the rebuilt renderer and the existing ABI 44 test host. At 1280×720, a 1024×576 trial opened the countdown, expired, restored 1280×720 and released its trial storage. The next trial was kept and saved 1024×576 to the test settings file.
  • Strict, full Linux comparison on current main 8aa74983 and published head 3541a0d2: main 813 passed, 107 failed, 18 skipped; PR 814 passed, 106 failed, 18 skipped. All 106 PR failures also occurred on main by test name, including the test_vbytes.py timeout. test_menu_keys.py failed on main and passed on the PR.
  • The website build fails locally with Incomplete graphics setting on both 8aa74983 and 3541a0d2. The upstream Graphics guide has empty default cells for two ReShade action rows. A local check with those cells filled reached a separate upstream DLSS/FSR visibility documentation mismatch. No website repair is included in this Controls PR. The PR's dedicated guide image and source review check passed.
Linux failures reproduced on main (106)

test_appimage_player_build.py, test_bug_report_defaults.py, test_capture_read_once.py, test_capture_readback_visibility.py, test_cmpflags.py, test_compile_retry_jobs.py,
test_compile_stops_on_exception.py, test_crash_dialog.py, test_crash_record.py, test_crt_options.py, test_damage_debug_menu.py, test_dev_packs.py,
test_device_lost_exit.py, test_device_policy.py, test_display_startup_failure.py, test_dlss_feature_size.py, test_draw_binds.py, test_draw_stream_visibility.py,
test_failed_compile_keeps_objects.py, test_finish_output.py, test_flush_nothing_held.py, test_frame_overlap.py, test_framegen_preset.py, test_fsr3fi_hud_layer.py,
test_fsr_image_retirement.py, test_fsr_uniform_ownership.py, test_gpu_wait_stats.py, test_guest_imported.py, test_hdr_encode.py, test_high_fps_fixes.py,
test_host_guest_writes.py, test_import_reach.py, test_input_background.py, test_intel_prepare_side_by_side.py, test_libc_hooks.py, test_lift_cache.py,
test_lift_cleanup_error.py, test_linux_preflight.py, test_linux_pressure.py, test_linux_tools.py, test_linux_update.py, test_log_open_failure.py,
test_low_vram_texture_mips.py, test_make_game_ngx.py, test_menu_env_overrides.py, test_menu_i18n.py, test_menu_ini_keys.py, test_menu_module_slots.py,
test_menu_modules.py, test_menu_sections.py, test_menu_text_field.py, test_mesh_conv.py, test_mise_build.py, test_missing_nup.py,
test_mods_list.py, test_mods_size_table.py, test_ngx_sdk.py, test_nonuniform_image_arrays.py, test_oot_shadow_fit.py, test_pass_barrier.py,
test_pending_guest_import.py, test_pipeline_play_cache.py, test_pipeline_tour.py, test_pipelines_outlive_cache_format.py, test_prepare_cache_save.py, test_prepare_stall.py,
test_present_ahead.py, test_present_device_loss.py, test_present_queue_default.py, test_present_source_crop.py, test_profiling_mode.py, test_program_cache_cleanup.py,
test_program_cache_one_variant.py, test_prompts_size_table.py, test_psychov35_fsr3_fg.py, test_readback_timeout_source.py, test_recorder_own_idle.py, test_recorder_sync.py,
test_reflex_swapchain_lifetime.py, test_renderer_new_handler.py, test_renhdrum.py, test_replay_write_order.py, test_rumble_menu.py, test_run_all_mingw.py,
test_run_all_output.py, test_run_all_reap.py, test_sampler_admission.py, test_settings_table.py, test_slow_frames.py, test_song_note_cue.py,
test_sparse_mirror.py, test_spirv_pack.py, test_staging_burst.py, test_submit_oom.py, test_target_reuse_memory.py, test_texel_buffer_ownership.py,
test_ub_update_batch.py, test_ui_record.py, test_upscaler_round_trip.py, test_upscaler_rows.py, test_vbytes.py, test_vcore_escape.py,
test_vcore_mask0.py, test_vibration_menu.py, test_vk_occlusion_queries.py, test_xinput_thread.py.

Still to test:

  • Windows, by hand: complete the strict full-suite comparison against current main. The paired Windows runs did not reach their summaries. Focused Windows results are recorded above.
  • AMD RDNA and Steam Deck, by hand with a controller: open Controls, confirm the displayed Clear button clears only the selected binding, and verify row reset and focus skipping over disabled cells.

Measurements

Verdict: Vertical input reaches the next available Toggle cell in either direction. The tested path runs in the open Controls grid. No frame-time claim is made for this menu interaction.

Setup: native LLVM-MinGW and Linux Clang game-free fixtures, same synthetic binding rows and press directions, compiled from each revision's navigation code.

Check Before: main @ 8aa74983 After: this PR @ 3541a0d2
Correct destination after skipping Plus and Minus 0/2 directions 2/2 directions

Risks

Binding conflicts, slot swaps, layout presets, Sprint's effective defaults, Toggle persistence, and the keyboard or controller focus path can regress if both sides of the host/renderer ABI diverge. A reset can be canceled before changing bindings. Settings continue to use the existing options file. The change adds no switch and needs no save or shader-cache migration. Revert the change to restore the previous Controls behavior.

Checklist

Ready for review

  • One topic, on current main; the title is a Conventional Commit.
  • Tested on my own PC, in the game for a change the game runs, as Testing says.
  • mise run test passes (any failure that also fails on main is named). The equivalent strict Linux suite reported no PR-only failing tests; see Testing for the baseline exceptions.
  • Measured against main, or Measurements says why there is nothing to measure.
  • No game files, build output or hand-edited generated files; third_party/ untouched; nothing GPL-incompatible.

When it applies

  • A fix: a game-free test that fails on main and passes here.
  • A check under tests/ that reads changed source is updated, not weakened.
  • MKWVK_API_VERSION bumped (MkwVkHost or MkwVkRenderer in gpu/vk/vk_api.h changed). Version 44 preserves main's output-preview and input-recording callbacks.
  • Host code builds on Windows (llvm-mingw) and Linux, with UTF-8 paths. Full host builds and UTF-8 paths were tested on the previous base; the follow-up was checked with a Windows renderer-only rebuild and both hosts' ABI fixtures.
  • Screenshots of the same spot on main and here, for a change that shows on screen.
  • Docs updated.
  • AI assistance recorded with Assisted-by: / Co-Authored-By: trailers.

Ready to merge

  • Everything in Still to test has run, its results in Tested.
  • Tier 1 hardware it can affect is tested (docs/dev/standards.md, "Hardware tiers").
  • Approved in review, every comment answered.
  • CI passes on the head; on current main, measured again or a comment says why the numbers still hold.
## Summary Controls now lets you reset a single row. The square curved-arrow button is available when the row's gamepad or keyboard bindings, or its Toggle, differ from the current defaults. The table headings sit below Layout preset and Button prompts. Resetting a row or all controls asks for confirmation. When a new input is already in use, the dialog names the affected control before reassigning it. The grid gives Toggle and Reset a little more space after Keyboard. With a controller, Clear uses the available top face button or another free face button if Bind or Back uses it. The footer shows the assigned button, and keyboard input shows Del. On main, open **Controls → Controls** and move down in Toggle from ZR toward Plus and Minus. Navigation reaches a row without a Toggle and moves focus to Keyboard. With this change, Up and Down stay in Toggle or Reset and skip unavailable cells. Left and Right change columns. Start with `gpu/vk/vk_menu.cpp` for focus and confirmation, and `host/nn_hid.c` for the effective defaults and binding updates. ## Testing **Tested:** - Windows and native laptop Linux at `42841c4d` before rebase. Both built and passed 12 focused game-free checks each. In game, Up and Down skipped disabled cells without leaving Toggle or Reset. Cancel preserved the binding. Confirm restored the default and disabled the reset arrow. - Windows and Linux game-free comparison on `971f77f5` and `3cbdeef1` plus the layout and fixture changes. The same navigation check missed both expected Toggle targets on main and reached both on the branch. - Seven focused Linux checks passed, including the input-recording and output-preview fixtures. Six focused Windows checks passed. The Windows input-recording fixture also fails on clean main on LLVM-MinGW 23.1.5, where the compiler reports that `tp_cancels` is set but unused. - Strict laptop Linux comparison before the spacing and Clear edits: main 813 passed, 106 failed, 18 skipped; topic 814 passed, 105 failed, 18 skipped. The topic has no additional failing names and passes `test_menu_keys.py`, which fails on main. - Earlier Windows strict runs: main 777 passed, 107 failed, 50 skipped; topic 769 passed, 115 failed, 50 skipped. Both timed out in `test_vbytes.py`. Eight branch-only failures passed on retry: `test_mesh_first_types.py`, `test_import_windows.py`, `test_cbuf_table_extent.py`, `test_dump_ssbo_read.py`, `test_mesh_dynamic_state.py`, `test_timeline_modes.py`, `test_capture_copy_nt.py`, and `test_capture_copy_runs.py`. The ninth, `test_hos_ab.py`, now passes after the test was updated for the appended ABI callbacks. The final Windows focused rerun passed six checks; Ruff passed for the three changed Python files. - Final spacing and Clear update: four focused game-free checks passed on Windows and laptop Linux. Both isolated runtimes used their existing executables and rebuilt native renderers. Manual play-tests passed on both platforms for spacing, controller Clear and the matching footer glyph. The laptop play-test also confirmed the return to Del under keyboard input. - Final runtime hardware: Windows 10 build 19045 with an RTX 3080 and NVIDIA 621.24; native Ubuntu with an RTX 5090 Laptop GPU and NVIDIA Vulkan driver `0x94d8c080`. - Reviewed the matched UI-only Controls captures from the pinned main and the final source, including the Classic theme and controller footer. A Graphics comparison reproduced identical pixels; the guide source review and offline check passed. The captures use synthetic bindings and contain no game data. - BeanBot follow-up at `3541a0d2`, rebased onto `8aa74983`. The Controls help and its existing translations now explain that confirming a conflict removes the previous binding. The new fixture executes the actual menu confirmation handlers for row reset, Reset All and conflicting rebind. It checks cancellation and acceptance against bindings and Toggles. A mutation that accepted Cancel failed at the expected assertion. - The final focused Windows and Linux runs each passed seven checks covering Controls, ABI and output preview. Windows also reproduced main's `test_pads_after_menu.py` and `test_xinput_thread.py` unused-variable failures. The translation-key check fails on the same unrelated Output resolution string on both branches. Ruff and the offline Graphics guide source check passed. The refreshed Graphics captures differ only by the version label. - Live output preview passed on Windows with the rebuilt renderer and the existing ABI 44 test host. At 1280×720, a 1024×576 trial opened the countdown, expired, restored 1280×720 and released its trial storage. The next trial was kept and saved 1024×576 to the test settings file. - Strict, full Linux comparison on current main `8aa74983` and published head `3541a0d2`: main 813 passed, 107 failed, 18 skipped; PR 814 passed, 106 failed, 18 skipped. All 106 PR failures also occurred on main by test name, including the `test_vbytes.py` timeout. `test_menu_keys.py` failed on main and passed on the PR. - The website build fails locally with `Incomplete graphics setting` on both `8aa74983` and `3541a0d2`. The upstream Graphics guide has empty default cells for two ReShade action rows. A local check with those cells filled reached a separate upstream DLSS/FSR visibility documentation mismatch. No website repair is included in this Controls PR. The PR's dedicated guide image and source review check passed. <details> <summary>Linux failures reproduced on main (106)</summary> `test_appimage_player_build.py`, `test_bug_report_defaults.py`, `test_capture_read_once.py`, `test_capture_readback_visibility.py`, `test_cmpflags.py`, `test_compile_retry_jobs.py`, `test_compile_stops_on_exception.py`, `test_crash_dialog.py`, `test_crash_record.py`, `test_crt_options.py`, `test_damage_debug_menu.py`, `test_dev_packs.py`, `test_device_lost_exit.py`, `test_device_policy.py`, `test_display_startup_failure.py`, `test_dlss_feature_size.py`, `test_draw_binds.py`, `test_draw_stream_visibility.py`, `test_failed_compile_keeps_objects.py`, `test_finish_output.py`, `test_flush_nothing_held.py`, `test_frame_overlap.py`, `test_framegen_preset.py`, `test_fsr3fi_hud_layer.py`, `test_fsr_image_retirement.py`, `test_fsr_uniform_ownership.py`, `test_gpu_wait_stats.py`, `test_guest_imported.py`, `test_hdr_encode.py`, `test_high_fps_fixes.py`, `test_host_guest_writes.py`, `test_import_reach.py`, `test_input_background.py`, `test_intel_prepare_side_by_side.py`, `test_libc_hooks.py`, `test_lift_cache.py`, `test_lift_cleanup_error.py`, `test_linux_preflight.py`, `test_linux_pressure.py`, `test_linux_tools.py`, `test_linux_update.py`, `test_log_open_failure.py`, `test_low_vram_texture_mips.py`, `test_make_game_ngx.py`, `test_menu_env_overrides.py`, `test_menu_i18n.py`, `test_menu_ini_keys.py`, `test_menu_module_slots.py`, `test_menu_modules.py`, `test_menu_sections.py`, `test_menu_text_field.py`, `test_mesh_conv.py`, `test_mise_build.py`, `test_missing_nup.py`, `test_mods_list.py`, `test_mods_size_table.py`, `test_ngx_sdk.py`, `test_nonuniform_image_arrays.py`, `test_oot_shadow_fit.py`, `test_pass_barrier.py`, `test_pending_guest_import.py`, `test_pipeline_play_cache.py`, `test_pipeline_tour.py`, `test_pipelines_outlive_cache_format.py`, `test_prepare_cache_save.py`, `test_prepare_stall.py`, `test_present_ahead.py`, `test_present_device_loss.py`, `test_present_queue_default.py`, `test_present_source_crop.py`, `test_profiling_mode.py`, `test_program_cache_cleanup.py`, `test_program_cache_one_variant.py`, `test_prompts_size_table.py`, `test_psychov35_fsr3_fg.py`, `test_readback_timeout_source.py`, `test_recorder_own_idle.py`, `test_recorder_sync.py`, `test_reflex_swapchain_lifetime.py`, `test_renderer_new_handler.py`, `test_renhdrum.py`, `test_replay_write_order.py`, `test_rumble_menu.py`, `test_run_all_mingw.py`, `test_run_all_output.py`, `test_run_all_reap.py`, `test_sampler_admission.py`, `test_settings_table.py`, `test_slow_frames.py`, `test_song_note_cue.py`, `test_sparse_mirror.py`, `test_spirv_pack.py`, `test_staging_burst.py`, `test_submit_oom.py`, `test_target_reuse_memory.py`, `test_texel_buffer_ownership.py`, `test_ub_update_batch.py`, `test_ui_record.py`, `test_upscaler_round_trip.py`, `test_upscaler_rows.py`, `test_vbytes.py`, `test_vcore_escape.py`, `test_vcore_mask0.py`, `test_vibration_menu.py`, `test_vk_occlusion_queries.py`, `test_xinput_thread.py`. </details> **Still to test:** - Windows, by hand: complete the strict full-suite comparison against current main. The paired Windows runs did not reach their summaries. Focused Windows results are recorded above. - AMD RDNA and Steam Deck, by hand with a controller: open Controls, confirm the displayed Clear button clears only the selected binding, and verify row reset and focus skipping over disabled cells. ## Measurements **Verdict:** Vertical input reaches the next available Toggle cell in either direction. The tested path runs in the open Controls grid. No frame-time claim is made for this menu interaction. Setup: native LLVM-MinGW and Linux Clang game-free fixtures, same synthetic binding rows and press directions, compiled from each revision's navigation code. | Check | Before: main @ `8aa74983` | After: this PR @ `3541a0d2` | |---|---|---| | Correct destination after skipping Plus and Minus | 0/2 directions | 2/2 directions | ## Risks Binding conflicts, slot swaps, layout presets, Sprint's effective defaults, Toggle persistence, and the keyboard or controller focus path can regress if both sides of the host/renderer ABI diverge. A reset can be canceled before changing bindings. Settings continue to use the existing options file. The change adds no switch and needs no save or shader-cache migration. Revert the change to restore the previous Controls behavior. ## Checklist ### Ready for review - [x] One topic, on current `main`; the title is a Conventional Commit. - [x] Tested on my own PC, in the game for a change the game runs, as Testing says. - [x] `mise run test` passes (any failure that also fails on `main` is named). The equivalent strict Linux suite reported no PR-only failing tests; see Testing for the baseline exceptions. - [x] Measured against `main`, or Measurements says why there is nothing to measure. - [x] No game files, build output or hand-edited generated files; `third_party/` untouched; nothing GPL-incompatible. ### When it applies - [x] A fix: a game-free test that fails on `main` and passes here. - [x] A check under `tests/` that reads changed source is updated, not weakened. - [x] `MKWVK_API_VERSION` bumped (`MkwVkHost` or `MkwVkRenderer` in `gpu/vk/vk_api.h` changed). Version 44 preserves main's output-preview and input-recording callbacks. - [x] Host code builds on Windows (llvm-mingw) and Linux, with UTF-8 paths. Full host builds and UTF-8 paths were tested on the previous base; the follow-up was checked with a Windows renderer-only rebuild and both hosts' ABI fixtures. - [x] Screenshots of the same spot on `main` and here, for a change that shows on screen. - [x] Docs updated. - [x] AI assistance recorded with `Assisted-by:` / `Co-Authored-By:` trailers. ### Ready to merge - [ ] Everything in Still to test has run, its results in Tested. - [ ] Tier 1 hardware it can affect is tested (docs/dev/standards.md, "Hardware tiers"). - [ ] Approved in review, every comment answered. - [ ] CI passes on the head; on current `main`, measured again or a comment says why the numbers still hold.
Restore individual control defaults, confirm resets and conflicting rebinds, and use the standard action styling for Reset to defaults.

Assisted-by: OpenAI Codex (GPT-6.2)
Show a reset arrow only as actionable when the current bindings or toggle differ from their effective defaults. Keep vertical navigation in the selected Toggle or Reset column.

Assisted-by: OpenAI Codex (GPT-6.2)
fix(controls): add controller Clear and refine the reset layout
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
Website / Website on Cloudflare (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 10s
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Successful in 14s
CI / Workflow lint (actionlint) (pull_request) Successful in 16s
Guide screenshots / Guide images and local UI source review (pull_request) Successful in 17s
Third-party parts / As their notes say (pull_request) Successful in 19s
CI / Website build and checks (pull_request) Failing after 23s
ddc420ce0d
Assisted-by: OpenAI Codex (GPT-6.2)
Author
Member

Tested in game: Tested in Controls on Windows 10 (RTX 3080, NVIDIA 621.24) and native Ubuntu (RTX 5090 Laptop GPU, Vulkan driver 0x94d8c080). Row reset, cancellation and navigation past disabled cells passed. On the final renderer update, both systems passed the spacing, controller Clear and matching footer icon checks. Linux also confirmed the keyboard Del hint.

Tested in game: Tested in Controls on Windows 10 (RTX 3080, NVIDIA 621.24) and native Ubuntu (RTX 5090 Laptop GPU, Vulkan driver 0x94d8c080). Row reset, cancellation and navigation past disabled cells passed. On the final renderer update, both systems passed the spacing, controller Clear and matching footer icon checks. Linux also confirmed the keyboard Del hint.
Author
Member

UI-only captures of the same Controls grid at 1280×720. The bindings are synthetic.

Theme Main at 971f77f5 PR at ddc420ce
Modern dark Main, Modern dark PR, Modern dark
Classic Main, Classic PR, Classic

The controller footer shows the face button assigned to Clear.

PR, controller Clear

UI-only captures of the same Controls grid at 1280×720. The bindings are synthetic. | Theme | Main at `971f77f5` | PR at `ddc420ce` | |---|---|---| | Modern dark | ![Main, Modern dark](https://git.magicbean.dev/attachments/502acc21-8aa1-4353-9999-b54e0aa63166) | ![PR, Modern dark](https://git.magicbean.dev/attachments/9eecdc2b-f707-4423-a089-ad8939745c98) | | Classic | ![Main, Classic](https://git.magicbean.dev/attachments/7280736e-d209-4f33-a17b-084fe78d2961) | ![PR, Classic](https://git.magicbean.dev/attachments/ff197903-eec7-47d8-8e6d-cd84322c2e80) | The controller footer shows the face button assigned to Clear. ![PR, controller Clear](https://git.magicbean.dev/attachments/8b68fc95-19d3-44db-84c1-e9efab9be5f1)
Member
@review
BeanBot requested changes 2026-11-21 21:27:59 +00:00
Dismissed
BeanBot left a comment

Recommendation: Request changes. The Controls help still tells players that a binding cannot be moved in a case this change now permits. The new confirmation paths also need behavioral regression coverage.

Adds per-control resets, confirmation for resets and conflicting bindings, controller Clear, and navigation that skips unavailable cells. It improves the tested navigation path but adds host and menu state that needs coverage and current player guidance.

@fix will do (2)

  1. L1 gpu/vk/vk_menu.cpp:2854: The Controls help still says a button's only gamepad input cannot move to another row's empty Input 2, but the new confirmed rebind permits exactly that and can leave the original row unbound. Update this help and its translations to explain the confirmation and removal behavior.
  2. L2 (minor) tests/test_menu_keys.py:477: The confirmation checks match source text rather than exercising the dialog. (more on the line)

Before merging, for the author (2) (a run, a measurement or a decision; it doesn't hold the recommendation)

  • L3 (minor): The required final-head checks and controller runs on AMD RDNA and Steam Deck are still outstanding. Run the checks and verify Clear, row reset and disabled-cell navigation on those systems before merge.
  • L4: It conflicts with its base branch: bring it up to date (@​fix or @​rebase try first).

Checks: Failing on this head, which is behind its base, where it passes: CI / Website build and checks (pull_request). @rebase (or @fix) brings it up to date and runs it again; not held against it.

More: does it help, other systems, switches, what was read

Does it help: Yes for the tested navigation case: the PR reports 0/2 correct destinations on the base and 2/2 on the branch using the same synthetic rows. The author also reports successful in-game checks on Windows and Linux NVIDIA systems. This is not a performance change.

Other systems: The host binding changes affect keyboard, controller, Sprint, shared note and HUD bindings on Windows and Linux. The reported in-game runs cover NVIDIA systems; AMD RDNA and Steam Deck controller checks, final-head full checks, and the output-preview manual check remain outstanding.

Switches: None added, changed or removed.

Reviewed: ddc420ce0d against 971f77f525: Read the repository rules, PR template, standards, commit log, host binding changes and surrounding Sprint code, menu diff and relevant input handling, and changed test diffs. Skimmed translations, the screenshot manifest and guide screenshot rules. The full changed code files and all surrounding callers were not read in full; the Tooling AGENTS.md was unavailable in the adjacent checkout.; nothing run (a code review).

Checked and fine:

  • The host interface version is bumped and the callbacks are appended after the version-42 callbacks.
  • Each branch commit records AI assistance.
  • The changed file list contains no game-derived output or third-party changes.
  • The host fixture covers several reset, conflict and Sprint-default cases.

Testing: No tests or project scripts were run during this read-only review. The PR reports focused Windows and Linux runs and identifies the remaining validation.

Overlaps and merge order: Unknown; network access was unavailable.


Automated code review by BeanBot (gpt-6.1-terra), asked for by llun. Nothing was run; a maintainer decides.

**Recommendation: Request changes.** The Controls help still tells players that a binding cannot be moved in a case this change now permits. The new confirmation paths also need behavioral regression coverage. Adds per-control resets, confirmation for resets and conflicting bindings, controller Clear, and navigation that skips unavailable cells. It improves the tested navigation path but adds host and menu state that needs coverage and current player guidance. **`@fix` will do (2)** 1. **L1** `gpu/vk/vk_menu.cpp:2854`: The Controls help still says a button's only gamepad input cannot move to another row's empty Input 2, but the new confirmed rebind permits exactly that and can leave the original row unbound. Update this help and its translations to explain the confirmation and removal behavior. 2. **L2** (minor) `tests/test_menu_keys.py:477`: The confirmation checks match source text rather than exercising the dialog. (more on the line) **Before merging, for the author (2)** (a run, a measurement or a decision; it doesn't hold the recommendation) - **L3** (minor): The required final-head checks and controller runs on AMD RDNA and Steam Deck are still outstanding. Run the checks and verify Clear, row reset and disabled-cell navigation on those systems before merge. - **L4**: It conflicts with its base branch: bring it up to date (`@​fix` or `@​rebase` try first). **Checks:** Failing on this head, which is behind its base, where it passes: `CI / Website build and checks (pull_request)`. `@rebase` (or `@fix`) brings it up to date and runs it again; not held against it. <details> <summary>More: does it help, other systems, switches, what was read</summary> **Does it help:** Yes for the tested navigation case: the PR reports 0/2 correct destinations on the base and 2/2 on the branch using the same synthetic rows. The author also reports successful in-game checks on Windows and Linux NVIDIA systems. This is not a performance change. **Other systems:** The host binding changes affect keyboard, controller, Sprint, shared note and HUD bindings on Windows and Linux. The reported in-game runs cover NVIDIA systems; AMD RDNA and Steam Deck controller checks, final-head full checks, and the output-preview manual check remain outstanding. **Switches:** None added, changed or removed. **Reviewed:** ddc420ce0d18 against 971f77f5251a: Read the repository rules, PR template, standards, commit log, host binding changes and surrounding Sprint code, menu diff and relevant input handling, and changed test diffs. Skimmed translations, the screenshot manifest and guide screenshot rules. The full changed code files and all surrounding callers were not read in full; the Tooling AGENTS.md was unavailable in the adjacent checkout.; nothing run (a code review). **Checked and fine:** - The host interface version is bumped and the callbacks are appended after the version-42 callbacks. - Each branch commit records AI assistance. - The changed file list contains no game-derived output or third-party changes. - The host fixture covers several reset, conflict and Sprint-default cases. **Testing:** No tests or project scripts were run during this read-only review. The PR reports focused Windows and Linux runs and identifies the remaining validation. **Overlaps and merge order:** Unknown; network access was unavailable. </details> --- <sub>Automated code review by BeanBot (gpt-6.1-terra), asked for by llun. Nothing was run; a maintainer decides.</sub>
@ -433,1 +474,4 @@
"vertical navigation must use the column-aware row selection")
need(tick, r"while \(!control_column_enabled\(i, ccol\)\);",
"keyboard navigation must skip Plus and Minus Toggle cells")
need(tick, r"if \(!press\[5\] && ask.choice == 1\) \{\s*if \(ask.action == ControlAction::Rebind\)",
Member

L2 (minor): The confirmation checks match source text rather than exercising the dialog. For example, they would still pass if the reset callback ran before the Cancel choice was checked. Add a game-free behavioral test that opens each confirmation, cancels and confirms it, and checks the resulting bindings and toggles.

**L2** (minor): The confirmation checks match source text rather than exercising the dialog. For example, they would still pass if the reset callback ran before the Cancel choice was checked. Add a game-free behavioral test that opens each confirmation, cancels and confirms it, and checks the resulting bindings and toggles.
Owner
@fix
Phishinabowl force-pushed feat/improved-control-defaults from ddc420ce0d
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
Website / Website on Cloudflare (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 10s
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Successful in 14s
CI / Workflow lint (actionlint) (pull_request) Successful in 16s
Guide screenshots / Guide images and local UI source review (pull_request) Successful in 17s
Third-party parts / As their notes say (pull_request) Successful in 19s
CI / Website build and checks (pull_request) Failing after 23s
to 3541a0d20c
Some checks failed
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been skipped
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been skipped
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been skipped
Third-party parts / The same as upstream (pull_request) Has been skipped
Website / Website on Cloudflare (pull_request) Has been skipped
CI / Conventional Commit title (pull_request) Successful in 20s
CI / Workflow lint (actionlint) (pull_request) Successful in 30s
Guide screenshots / Guide images and local UI source review (pull_request) Successful in 31s
Third-party parts / As their notes say (pull_request) Successful in 32s
CI / Website build and checks (pull_request) Failing after 39s
PR description / Description follows the template (pull_request_target) Successful in 6s
2026-11-21 21:38:19 +00:00
Compare
Author
Member

BeanBot follow-up in 3541a0d2: L1 now describes the confirmation and removal behavior in Controls help and its translations. L2 adds game-free coverage of the real reset and rebind dialog handlers, including cancellation and acceptance. L4 was addressed by rebasing onto 8aa74983. Focused checks passed on Windows and Linux; remaining live preview, final CI and hardware checks are listed above.

BeanBot follow-up in 3541a0d2: L1 now describes the confirmation and removal behavior in Controls help and its translations. L2 adds game-free coverage of the real reset and rebind dialog handlers, including cancellation and acceptance. L4 was addressed by rebasing onto 8aa74983. Focused checks passed on Windows and Linux; remaining live preview, final CI and hardware checks are listed above.
Author
Member

Refreshed UI-only captures following the rebase and help update. All use the same 1280×720 viewport and synthetic bindings.

Theme Main at 8aa74983 PR at 3541a0d2
Modern dark Main, Modern dark PR, Modern dark
Classic Main, Classic PR, Classic

Updated help and controller Clear

Refreshed UI-only captures following the rebase and help update. All use the same 1280×720 viewport and synthetic bindings. | Theme | Main at `8aa74983` | PR at `3541a0d2` | |---|---|---| | Modern dark | ![Main, Modern dark](https://git.magicbean.dev/attachments/84b70922-2082-4ea3-b607-f37d4b90a024) | ![PR, Modern dark](https://git.magicbean.dev/attachments/bedfc6d6-6b81-4d05-8898-6c775331aae8) | | Classic | ![Main, Classic](https://git.magicbean.dev/attachments/24469dd2-c513-48e8-9dea-192ee5008784) | ![PR, Classic](https://git.magicbean.dev/attachments/11b0b5d0-4cff-4806-9a7b-d38325f05820) | ![Updated help and controller Clear](https://git.magicbean.dev/attachments/e0b62ef3-75c0-4d6c-bc5d-8056b4a8b903)
BeanBot left a comment

Recommendation: Request changes. The help and confirmation tests now address the earlier findings, but the pull request is still reported as unmergeable. Final-head checks and controller validation on AMD RDNA and Steam Deck also remain outstanding.

The change adds per-control reset and confirmation for resets and conflicting bindings, and makes navigation skip unavailable cells. It improves control over bindings at the cost of additional confirmation and host–renderer coordination.

Before merging, for the author (1) (a run, a measurement or a decision; it doesn't hold the recommendation)

  • L3 (minor): The required final-head checks and controller runs on AMD RDNA and Steam Deck are still outstanding. Run the checks and verify Clear, row reset and disabled-cell navigation on those systems before merge.

Since the last review (on ddc420ce0d)
Resolved: L1 (The Controls help still says a button's only gamepad input cannot move to another row's em…); L2 (The confirmation checks match source text rather than exercising the dialog. For example,…); L4 (It conflicts with its base branch: bring it up to date (@​fix or @​rebase try first).)
Still open: L3 (The required final-head checks and controller runs on AMD RDNA and Steam Deck are still ou…)

Checks: Still running: CI / Website build and checks (pull_request), PR description / Description follows the template (pull_request_target), CI / Conventional Commit title (pull_request), CI / Workflow lint (actionlint) (pull_request), Guide screenshots / Guide images and local UI source review (pull_request), Third-party parts / As their notes say (pull_request).

Also noticed (1; not asked for, not blocking)
  • The live output-preview confirmation and expiry check remains listed as outstanding.
More: does it help, other systems, switches, what was read

Does it help: Yes for the tested navigation case: the supplied game-free comparison reached 2 of 2 expected destinations on the branch versus 0 of 2 on main. The author also reports successful in-game reset, cancel, navigation and Clear checks on Windows and Linux; no performance gain is claimed.

Other systems: The binding changes affect keyboard and controller input, Sprint defaults, saved options and the host–renderer interface. Windows and Linux NVIDIA play tests are reported; AMD RDNA, Steam Deck and the final-head checks remain outstanding.

Switches: None added, changed or removed.

Reviewed: 3541a0d20c against 8aa7498345: Read AGENTS.md, CONTRIBUTING.md, the pull request template, the two follow-up commits and the relevant confirmation, help and binding changes. The full changed code files and all dependent callers were not read in full; translations, the manifest and other changed tests were skimmed. docs/review-checklist.md is absent.; nothing run (a code review).

Checked and fine:

  • The updated help removes the obsolete restriction and describes confirmed removal.
  • The confirmation fixture checks that cancellation preserves bindings and that confirmation applies the action.

Testing: No commands that build or run project code were executed. The supplied CI checks are still running; the author reports earlier Windows and Linux checks and in-game tests.

Overlaps and merge order: Unknown; the network is closed.


Automated code review by BeanBot (gpt-6.1-terra), asked for by HumanBean. Nothing was run; a maintainer decides.

**Recommendation: Request changes.** The help and confirmation tests now address the earlier findings, but the pull request is still reported as unmergeable. Final-head checks and controller validation on AMD RDNA and Steam Deck also remain outstanding. The change adds per-control reset and confirmation for resets and conflicting bindings, and makes navigation skip unavailable cells. It improves control over bindings at the cost of additional confirmation and host–renderer coordination. **Before merging, for the author (1)** (a run, a measurement or a decision; it doesn't hold the recommendation) - **L3** (minor): The required final-head checks and controller runs on AMD RDNA and Steam Deck are still outstanding. Run the checks and verify Clear, row reset and disabled-cell navigation on those systems before merge. **Since the last review** (on ddc420ce0d18) Resolved: L1 (The Controls help still says a button's only gamepad input cannot move to another row's em…); L2 (The confirmation checks match source text rather than exercising the dialog. For example,…); L4 (It conflicts with its base branch: bring it up to date (`@​fix` or `@​rebase` try first).) Still open: L3 (The required final-head checks and controller runs on AMD RDNA and Steam Deck are still ou…) **Checks:** Still running: `CI / Website build and checks (pull_request)`, `PR description / Description follows the template (pull_request_target)`, `CI / Conventional Commit title (pull_request)`, `CI / Workflow lint (actionlint) (pull_request)`, `Guide screenshots / Guide images and local UI source review (pull_request)`, `Third-party parts / As their notes say (pull_request)`. <details> <summary>Also noticed (1; not asked for, not blocking)</summary> - The live output-preview confirmation and expiry check remains listed as outstanding. </details> <details> <summary>More: does it help, other systems, switches, what was read</summary> **Does it help:** Yes for the tested navigation case: the supplied game-free comparison reached 2 of 2 expected destinations on the branch versus 0 of 2 on main. The author also reports successful in-game reset, cancel, navigation and Clear checks on Windows and Linux; no performance gain is claimed. **Other systems:** The binding changes affect keyboard and controller input, Sprint defaults, saved options and the host–renderer interface. Windows and Linux NVIDIA play tests are reported; AMD RDNA, Steam Deck and the final-head checks remain outstanding. **Switches:** None added, changed or removed. **Reviewed:** 3541a0d20c11 against 8aa7498345df: Read AGENTS.md, CONTRIBUTING.md, the pull request template, the two follow-up commits and the relevant confirmation, help and binding changes. The full changed code files and all dependent callers were not read in full; translations, the manifest and other changed tests were skimmed. docs/review-checklist.md is absent.; nothing run (a code review). **Checked and fine:** - The updated help removes the obsolete restriction and describes confirmed removal. - The confirmation fixture checks that cancellation preserves bindings and that confirmation applies the action. **Testing:** No commands that build or run project code were executed. The supplied CI checks are still running; the author reports earlier Windows and Linux checks and in-game tests. **Overlaps and merge order:** Unknown; the network is closed. </details> --- <sub>Automated code review by BeanBot (gpt-6.1-terra), asked for by HumanBean. Nothing was run; a maintainer decides.</sub>
Member

The branch moved while I worked (someone pushed 3541a0d20c), so I did it again on top of that. There is nothing for an edit to fix on this pull request. It was already up to date with main.

For the author (no edit does these)

  • L3 (minor): The required final-head checks and controller runs on AMD RDNA and Steam Deck are still outstanding. Run the checks and verify Clear, row reset and disabled-cell navigation on those systems before merge.

Automated fix by BeanBot (gpt-6.1-terra), asked for by HumanBean. Nothing was run; a maintainer decides.

The branch moved while I worked (someone pushed 3541a0d20c11), so I did it again on top of that. There is nothing for an edit to fix on this pull request. It was already up to date with `main`. **For the author** (no edit does these) - **L3** (minor): The required final-head checks and controller runs on AMD RDNA and Steam Deck are still outstanding. Run the checks and verify Clear, row reset and disabled-cell navigation on those systems before merge. --- <sub>Automated fix by BeanBot (gpt-6.1-terra), asked for by HumanBean. Nothing was run; a maintainer decides.</sub>
Author
Member

Live Windows output preview passed: the first 1024×576 trial expired, restored 1280×720 and released its storage; choosing Yes on the second saved 1024×576. The full strict Linux comparison found no PR-only failing tests against 8aa74983. The remaining website failure was reproduced locally on clean main as well as this branch; the dedicated guide screenshot CI passed.

Live Windows output preview passed: the first 1024×576 trial expired, restored 1280×720 and released its storage; choosing Yes on the second saved 1024×576. The full strict Linux comparison found no PR-only failing tests against 8aa74983. The remaining website failure was reproduced locally on clean main as well as this branch; the dedicated guide screenshot CI passed.
Phishinabowl changed title from WIP: feat(controls): add per-control reset and confirm binding changes to feat(controls): add per-control reset and confirm binding changes 2026-11-21 22:10:13 +00:00
HumanBean deleted branch feat/improved-control-defaults 2026-11-21 22:23:24 +00:00
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
4 participants
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
MagicBeanDev/Longshot!2126
No description provided.