fix(ui): the text box's choices stay on a 16:10 or narrower screen #1233

Merged
HumanBean merged 3 commits from cuecuemoar/Longshot:fix/held-choices-narrow into main 2026-11-16 21:58:32 +00:00
Contributor

Type: fix · Risk: low · Tested on: Windows 10, RTX 4090 · Needs: none

Why

On a screen narrower than 16:9 (16:10 such as 1920x1200, 3:2, 4:3, 5:4), the answers you pick from in a conversation run off the right edge of the screen: their boxes are cut, and at 4:3 their text too ("Onde fica o caste..."). It happens in every conversation with choices, in every language, since #95 laid the HUD out for each screen shape. Fixes #1050.

How to reproduce

Seen on: Windows 10 22H2, RTX 4090, NVIDIA 617.42 (the report: Windows 11, RTX 5060 Laptop, a 1920x1200 display)
Settings and switches: Output resolution 1920x1200 (or HOS_OUTPUT_RES=1920x1200), Aspect ratio the monitor's shape; any language (the runs were in pt-BR, the report's)
Steps: load any save, go to the Market (the Cheats page's teleport, or the control bridge's teleport Market_Child), talk to the guard at the gate: "Precisa de alguma coisa?" / "Need anything?" and three choices
Expected: the three choices whole on the screen, as at 16:9
Actual: on main their boxes are cut at the right edge, and the column's scroll hint (the ↑ key) is off the screen; at 4:3 the texts are cut too

What changed

The text box (MessageDialog_00) is held where the game puts it, at the HUD's size (#95), but its choices sit in a column at the right edge of the game's 16:9 canvas. Down to 3:2 the HUD keeps its 16:9 size, so that canvas is wider than the screen and the column was left past the screen's edge. The column now moves in to the screen's right side exactly as far as the HUD's own side panes do, so it keeps the distance from the edge it has at 16:9; the box under it stays where it is, and nothing moves on a wider screen.

  • The plan (host/hos_wide_ui.c): a new rule, RULE_HELD_RIGHT, for one pane by name, MessageDialog_00's N_BtnNum_00 (the choices' column; in the game's file its boxes reach x 1930 of the 3840-wide canvas). plan() gives it the HUD's side move ui_ex only when that is a move in (ui_ex < 0: 16:10 and narrower). For it to be placed, the held layout's panes are now noted as a tree (hos_wide_ui_placed: n->tree for LK_HELD); the planner still walks none of them (visit skips anything not LK_PLAIN), so nothing else in the text box moves.
  • The check (tests/test_hud_layout.py, its own commit before the fix): the fixture's text box is now shaped like the game's: the box across the middle, the column at 1460 and a 940-wide choice made from it. It checks the column at 21:9 (held), 16:10 and 4:3 (moved in by the HUD's side move, its choice following without moving twice, as far from the screen's edge as at 16:9), the box held at every shape, and at every shape from 5:4 to 48:9 the choices no further past the edge than at 16:9. The old 4:3 check asserted the choice unmoved, the position that is cut off.
  • Runs (docs/tested/2026-11-16-held-choices-cuecuemoar-run01.md to run06): main and this change at 16:10, 4:3 and 21:9.

Measurements

Verdict: fixed: on main the choices are cut at the right edge at 16:10 and 4:3; with this change they are whole, with their scroll hint, and at 21:9 they are where main puts them. The game-free check fails on main and passes here. Nothing to measure for speed: the plan's time a present is the same (0.04 to 0.06 ms with a text box open, both sides).

Hardware: Windows 10 Pro 22H2 (19045), RTX 4090 24 GB, NVIDIA 617.42, Core i9-13900K, 32 GB
Scene: the Market gate guard's question and its three choices; HOS_OUTPUT_RES 1920x1200, 1440x1080 and 2560x1080, headless (offscreen, scripted input through the control bridge), pt-BR, otherwise defaults
Method: the kokiri-start repository save, teleport Market_Child, the guard talked to, a screenshot of the choices (RX_VK_SHOT_W/_H at the output size) and the plan's log (HOS_WIDE_UI_LOG=400); one run each way at each shape, plus a second main run at 16:10
How to reproduce: as above; scripts/dev_packs.py base of main e090cc48 and this change as a pack over it (see Testing for the commits)
Baseline: main e090cc48, main's head when the runs were made; the PR is now on 623a0ca3. The runs still hold there: none of the 188 commits between them changes host/hos_wide_ui.c, and of the other files the text box's drawing goes through, nx2/oot_wide.py gained only the NPC speech bubbles' site (#1191, its own layout, BalloonMessage_00) and host/rx_backend.c only audio and options changes (#1102, #1001, #729) and the Vulkan-only Windows build (#577). The game-free check ran on 623a0ca3, both ways.

Before: main @ e090cc48 After: this PR @ 6f9b566c (the same diff as cc8cc945)
1920x1200 (16:10): the choices cut at the right edge, the scroll hint off the screen (2 runs) whole, the scroll hint beside them
1440x1080 (4:3): the choices boxes and texts cut whole, the scroll hint beside them
2560x1080 (21:9): the choices whole, beside the box the same place as main
N_BtnNum_00 in the plan's log not noted 16:10 left -192.0, 4:3 left -300.0, 21:9 stays 0.0
The plan a present, a text box open 0.04-0.06 ms 0.04-0.06 ms
tests/test_hud_layout.py (main @ 623a0ca3 with the test commit 5fe2f5a6; this PR @ cc8cc945) FAIL PASS
wide ui: picture 1.6000: the HUD at 100% of its 16:9 size, its sides moved -192 and its top and bottom 0 canvas units; the menus at 90%
wide ui: MessageDialog_00:N_BtnNum_00 (pane 107d8c0c70 parent 107d8c0bc0, menu) type 0 x 1460.0 [1415.0 1505.0] y 0.0 [-60.0 60.0] rule 8: left -192.0, up 0.0, width x1.000
wide ui: picture 1.3333: the HUD at 89% of its 16:9 size, its sides moved -300 and its top and bottom 135 canvas units; the menus at 75%
wide ui: MessageDialog_00:N_BtnNum_00 (pane 107cfa0c70 parent 107cfa0bc0, menu) type 0 x 1460.0 [1415.0 1505.0] y 0.0 [-60.0 60.0] rule 8: left -300.0, up 0.0, width x1.000

main @ 623a0ca3 + 5fe2f5a6 (the test alone):
FAIL: plan failed:
FAIL:866 ds == 1.0f && dbx == 0 && dby == -704 && near(dcx, 1460 + e) && near(dbtn, dcx) && dcy == -500

this PR @ cc8cc945:
PASS: the HUD keeps its 16:9 size at each shape, at the screen's edges; the menus keep 16:9
PASS: the UI projection's site scales every screen not kept whole across and up, and 16:9 not at all

Screenshots, cropped to the choices: #1233 (comment) (main on the left, this PR on the right).

Flags and settings

None. HOS_WIDE_UI=0 (the HUD in the middle 16:9, or fitted to a narrower screen's width) is unchanged and leaves the column where the game puts it, as before.

Testing

Tested:

  • In the game (Windows 10 22H2, RTX 4090, NVIDIA 617.42; headless, the control bridge): the Market gate guard's choices at 1920x1200, 1440x1080 and 2560x1080, main against this change, as in Measurements. The game was built with llvm-mingw from main e090cc48 (dev_packs.py base) and this change as a pack over it (6f9b566c, before the branch was rebased and split into a test commit and the fix; the change is the same, git diff 6f9b566c^ 6f9b566c and git diff 623a0ca3 cc8cc945 have one patch ID, d0d83118). None of the 188 commits from e090cc48 to 623a0ca3 touch host/hos_wide_ui.c or its checks.
  • tests/test_hud_layout.py against broken fixes, each failing: the held layout's panes not noted, the column moved the wrong way, moved on wider screens too, and no rule for it.
  • tests/test_hud_layout.py and tests/test_wide_ui.py (the two checks that read hos_wide_ui.c) pass.
  • host/hos_wide_ui.c compiled: whole, for Windows, in the pack's build (llvm-mingw, no warning: in its log); on Linux with gcc -std=c11 -fsyntax-only -Wall -Wextra and the build's generated headers, no warning or error on this branch or on main, and the checks above compile the whole file's body with gcc on Linux.
  • python -I -B tests/run_all.py on Linux (WSL, Ubuntu 22.04): 565 passed, 26 failed (25 checks) on main @ ac778751; on this branch the same 25 fail and nothing else, the WSL machine lacking clang, the Vulkan headers and some Python modules for them (102 skipped): nx2/test_nup_zstd_heads, test_bug_report_defaults, test_check_third_party, test_cpp_preflight, test_crash_record, test_crt_options, test_crt_shaders, test_dlss_feature_size, test_failed_compile_keeps_objects, test_guest_imported, test_lift_lookups, test_link_skip, test_linux_update, test_longshot_saves, test_make_game_cache, test_missing_nup, test_mods_size_table, test_pending_guest_import, test_prof_clock, test_prompt_skip, test_run_all_changed, test_scene_tonemap_pass, test_vk_occlusion_queries, test_walk, test_walk_key. test_bug_report.py and test_hardware_report.py failed in the branch's first run only because ../Tooling was not cloned yet; with it they pass on both. That run was on the branch over ac778751; the last rebase, onto 623a0ca3, brought one commit (the menu's Frame generation and CRT rows, #1106), which touches none of this change's files; the two layout checks and the tested-runs check pass on it.

Not tested:

  • A real 16:10 or 4:3 monitor in fullscreen or a window: the runs set the output size (HOS_OUTPUT_RES) and drew offscreen; the plan reads only the picture's shape, which the log shows.
  • A full Linux build and run: the change is plain C in one portable file, with no platform code.
  • Conversations other than the gate guard's, and the text box in menus: every choice prompt uses the same layout, so the same column.

Must be tested:

  • None.

Risks

What could break: the text box's panes are now kept in the plan's table (about 80, plus their parts'), out of 65,536 slots; the planner walks none of them. A layout of another name with choices at the canvas's edge would still be cut: none was seen.

Turning it off: revert the commit (HOS_WIDE_UI=0 turns off the HUD's layout as a whole).

Compatibility: Nothing changes: no save, option, cache or interface (MKWVK_API_VERSION) is touched.

Issues

Fixes #1050
See #95

Checklist

Everyone

  • One topic.
  • The title is a Conventional Commit (type(scope): description) of the status line's type.
  • Why gives the reason plainly and links the issues that fit (the tracker searched).
  • Measurements has a verdict, or one line saying why there is nothing to measure.
  • Testing names what was tested, what wasn't, and what must still be tested, where and how.
  • Rebased on current main.
  • python -I -B tests/run_all.py passes (any failure that also fails on main is named).
  • No game files, dumps, build output or generated files.
  • third_party/ untouched (a new folder with a PROVENANCE.md instead).
  • The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled.

When it applies

  • AI assistance recorded with Assisted-by: / Co-Authored-By: trailers.
  • Screenshots of the same spot, captioned like the Measurements columns, for a change that shows on screen.
  • A file in docs/tested/ for each new measured run (format in docs/tested-configurations.md).
  • A game-free test added for the fix, failing on main and passing here.
  • A check under tests/ that reads changed source updated in this PR, not weakened.
  • Host code builds on both Windows (llvm-mingw) and Linux, with UTF-8 paths.
  • Docs updated.

🤖 Generated with Claude Code

https://claude.ai/code/session_01VxeaZib8ryJL5zYJuGgxcr

**Type:** fix · **Risk:** low · **Tested on:** Windows 10, RTX 4090 · **Needs:** none ## Why On a screen narrower than 16:9 (16:10 such as 1920x1200, 3:2, 4:3, 5:4), the answers you pick from in a conversation run off the right edge of the screen: their boxes are cut, and at 4:3 their text too ("Onde fica o caste..."). It happens in every conversation with choices, in every language, since #95 laid the HUD out for each screen shape. Fixes #1050. ## How to reproduce **Seen on:** Windows 10 22H2, RTX 4090, NVIDIA 617.42 (the report: Windows 11, RTX 5060 Laptop, a 1920x1200 display) **Settings and switches:** Output resolution 1920x1200 (or `HOS_OUTPUT_RES=1920x1200`), Aspect ratio the monitor's shape; any language (the runs were in pt-BR, the report's) **Steps:** load any save, go to the Market (the Cheats page's teleport, or the control bridge's `teleport Market_Child`), talk to the guard at the gate: "Precisa de alguma coisa?" / "Need anything?" and three choices **Expected:** the three choices whole on the screen, as at 16:9 **Actual:** on main their boxes are cut at the right edge, and the column's scroll hint (the ↑ key) is off the screen; at 4:3 the texts are cut too ## What changed The text box (`MessageDialog_00`) is held where the game puts it, at the HUD's size (#95), but its choices sit in a column at the right edge of the game's 16:9 canvas. Down to 3:2 the HUD keeps its 16:9 size, so that canvas is wider than the screen and the column was left past the screen's edge. The column now moves in to the screen's right side exactly as far as the HUD's own side panes do, so it keeps the distance from the edge it has at 16:9; the box under it stays where it is, and nothing moves on a wider screen. - **The plan** (`host/hos_wide_ui.c`): a new rule, `RULE_HELD_RIGHT`, for one pane by name, `MessageDialog_00`'s `N_BtnNum_00` (the choices' column; in the game's file its boxes reach x 1930 of the 3840-wide canvas). `plan()` gives it the HUD's side move `ui_ex` only when that is a move in (`ui_ex < 0`: 16:10 and narrower). For it to be placed, the held layout's panes are now noted as a tree (`hos_wide_ui_placed`: `n->tree` for `LK_HELD`); the planner still walks none of them (`visit` skips anything not `LK_PLAIN`), so nothing else in the text box moves. - **The check** (`tests/test_hud_layout.py`, its own commit before the fix): the fixture's text box is now shaped like the game's: the box across the middle, the column at 1460 and a 940-wide choice made from it. It checks the column at 21:9 (held), 16:10 and 4:3 (moved in by the HUD's side move, its choice following without moving twice, as far from the screen's edge as at 16:9), the box held at every shape, and at every shape from 5:4 to 48:9 the choices no further past the edge than at 16:9. The old 4:3 check asserted the choice unmoved, the position that is cut off. - **Runs** (`docs/tested/2026-11-16-held-choices-cuecuemoar-run01.md` to `run06`): main and this change at 16:10, 4:3 and 21:9. ## Measurements **Verdict:** fixed: on main the choices are cut at the right edge at 16:10 and 4:3; with this change they are whole, with their scroll hint, and at 21:9 they are where main puts them. The game-free check fails on main and passes here. Nothing to measure for speed: the plan's time a present is the same (0.04 to 0.06 ms with a text box open, both sides). Hardware: Windows 10 Pro 22H2 (19045), RTX 4090 24 GB, NVIDIA 617.42, Core i9-13900K, 32 GB Scene: the Market gate guard's question and its three choices; `HOS_OUTPUT_RES` 1920x1200, 1440x1080 and 2560x1080, headless (offscreen, scripted input through the control bridge), pt-BR, otherwise defaults Method: the `kokiri-start` repository save, `teleport Market_Child`, the guard talked to, a screenshot of the choices (`RX_VK_SHOT_W`/`_H` at the output size) and the plan's log (`HOS_WIDE_UI_LOG=400`); one run each way at each shape, plus a second main run at 16:10 How to reproduce: as above; `scripts/dev_packs.py` base of main e090cc48 and this change as a pack over it (see Testing for the commits) Baseline: main e090cc48, main's head when the runs were made; the PR is now on 623a0ca3. The runs still hold there: none of the 188 commits between them changes `host/hos_wide_ui.c`, and of the other files the text box's drawing goes through, `nx2/oot_wide.py` gained only the NPC speech bubbles' site (#1191, its own layout, `BalloonMessage_00`) and `host/rx_backend.c` only audio and options changes (#1102, #1001, #729) and the Vulkan-only Windows build (#577). The game-free check ran on 623a0ca3, both ways. | | Before: main @ e090cc48 | After: this PR @ 6f9b566c (the same diff as cc8cc945) | |---|---|---| | 1920x1200 (16:10): the choices | cut at the right edge, the scroll hint off the screen (2 runs) | whole, the scroll hint beside them | | 1440x1080 (4:3): the choices | boxes and texts cut | whole, the scroll hint beside them | | 2560x1080 (21:9): the choices | whole, beside the box | the same place as main | | `N_BtnNum_00` in the plan's log | not noted | 16:10 `left -192.0`, 4:3 `left -300.0`, 21:9 `stays 0.0` | | The plan a present, a text box open | 0.04-0.06 ms | 0.04-0.06 ms | | `tests/test_hud_layout.py` (main @ 623a0ca3 with the test commit 5fe2f5a6; this PR @ cc8cc945) | FAIL | PASS | ``` wide ui: picture 1.6000: the HUD at 100% of its 16:9 size, its sides moved -192 and its top and bottom 0 canvas units; the menus at 90% wide ui: MessageDialog_00:N_BtnNum_00 (pane 107d8c0c70 parent 107d8c0bc0, menu) type 0 x 1460.0 [1415.0 1505.0] y 0.0 [-60.0 60.0] rule 8: left -192.0, up 0.0, width x1.000 wide ui: picture 1.3333: the HUD at 89% of its 16:9 size, its sides moved -300 and its top and bottom 135 canvas units; the menus at 75% wide ui: MessageDialog_00:N_BtnNum_00 (pane 107cfa0c70 parent 107cfa0bc0, menu) type 0 x 1460.0 [1415.0 1505.0] y 0.0 [-60.0 60.0] rule 8: left -300.0, up 0.0, width x1.000 main @ 623a0ca3 + 5fe2f5a6 (the test alone): FAIL: plan failed: FAIL:866 ds == 1.0f && dbx == 0 && dby == -704 && near(dcx, 1460 + e) && near(dbtn, dcx) && dcy == -500 this PR @ cc8cc945: PASS: the HUD keeps its 16:9 size at each shape, at the screen's edges; the menus keep 16:9 PASS: the UI projection's site scales every screen not kept whole across and up, and 16:9 not at all ``` Screenshots, cropped to the choices: https://git.magicbean.dev/MagicBeanDev/Longshot/pulls/1233#issuecomment-37957 (main on the left, this PR on the right). ### Flags and settings None. `HOS_WIDE_UI=0` (the HUD in the middle 16:9, or fitted to a narrower screen's width) is unchanged and leaves the column where the game puts it, as before. ## Testing **Tested:** - In the game (Windows 10 22H2, RTX 4090, NVIDIA 617.42; headless, the control bridge): the Market gate guard's choices at 1920x1200, 1440x1080 and 2560x1080, main against this change, as in Measurements. The game was built with llvm-mingw from main e090cc48 (`dev_packs.py base`) and this change as a pack over it (6f9b566c, before the branch was rebased and split into a test commit and the fix; the change is the same, `git diff 6f9b566c^ 6f9b566c` and `git diff 623a0ca3 cc8cc945` have one patch ID, d0d83118). None of the 188 commits from e090cc48 to 623a0ca3 touch `host/hos_wide_ui.c` or its checks. - `tests/test_hud_layout.py` against broken fixes, each failing: the held layout's panes not noted, the column moved the wrong way, moved on wider screens too, and no rule for it. - `tests/test_hud_layout.py` and `tests/test_wide_ui.py` (the two checks that read `hos_wide_ui.c`) pass. - `host/hos_wide_ui.c` compiled: whole, for Windows, in the pack's build (llvm-mingw, no `warning:` in its log); on Linux with `gcc -std=c11 -fsyntax-only -Wall -Wextra` and the build's generated headers, no warning or error on this branch or on main, and the checks above compile the whole file's body with gcc on Linux. - `python -I -B tests/run_all.py` on Linux (WSL, Ubuntu 22.04): 565 passed, 26 failed (25 checks) on main @ ac778751; on this branch the same 25 fail and nothing else, the WSL machine lacking clang, the Vulkan headers and some Python modules for them (102 skipped): `nx2/test_nup_zstd_heads`, `test_bug_report_defaults`, `test_check_third_party`, `test_cpp_preflight`, `test_crash_record`, `test_crt_options`, `test_crt_shaders`, `test_dlss_feature_size`, `test_failed_compile_keeps_objects`, `test_guest_imported`, `test_lift_lookups`, `test_link_skip`, `test_linux_update`, `test_longshot_saves`, `test_make_game_cache`, `test_missing_nup`, `test_mods_size_table`, `test_pending_guest_import`, `test_prof_clock`, `test_prompt_skip`, `test_run_all_changed`, `test_scene_tonemap_pass`, `test_vk_occlusion_queries`, `test_walk`, `test_walk_key`. `test_bug_report.py` and `test_hardware_report.py` failed in the branch's first run only because `../Tooling` was not cloned yet; with it they pass on both. That run was on the branch over ac778751; the last rebase, onto 623a0ca3, brought one commit (the menu's Frame generation and CRT rows, #1106), which touches none of this change's files; the two layout checks and the tested-runs check pass on it. **Not tested:** - A real 16:10 or 4:3 monitor in fullscreen or a window: the runs set the output size (`HOS_OUTPUT_RES`) and drew offscreen; the plan reads only the picture's shape, which the log shows. - A full Linux build and run: the change is plain C in one portable file, with no platform code. - Conversations other than the gate guard's, and the text box in menus: every choice prompt uses the same layout, so the same column. **Must be tested:** - None. ## Risks **What could break:** the text box's panes are now kept in the plan's table (about 80, plus their parts'), out of 65,536 slots; the planner walks none of them. A layout of another name with choices at the canvas's edge would still be cut: none was seen. **Turning it off:** revert the commit (`HOS_WIDE_UI=0` turns off the HUD's layout as a whole). **Compatibility:** Nothing changes: no save, option, cache or interface (`MKWVK_API_VERSION`) is touched. ## Issues Fixes #1050 See #95 ## Checklist ### Everyone - [x] One topic. - [x] The title is a Conventional Commit (`type(scope): description`) of the status line's type. - [x] Why gives the reason plainly and links the issues that fit (the tracker searched). - [x] Measurements has a verdict, or one line saying why there is nothing to measure. - [x] Testing names what was tested, what wasn't, and what must still be tested, where and how. - [x] Rebased on current `main`. - [x] `python -I -B tests/run_all.py` passes (any failure that also fails on `main` is named). - [x] No game files, dumps, build output or generated files. - [x] `third_party/` untouched (a new folder with a `PROVENANCE.md` instead). - [x] The license is OK: GPL-3.0-or-later, nothing incompatible and nothing decompiled. ### When it applies - [x] AI assistance recorded with `Assisted-by:` / `Co-Authored-By:` trailers. - [x] Screenshots of the same spot, captioned like the Measurements columns, for a change that shows on screen. - [x] A file in `docs/tested/` for each new measured run (format in `docs/tested-configurations.md`). - [x] A game-free test added for the fix, failing on `main` and passing here. - [x] A check under `tests/` that reads changed source updated in this PR, not weakened. - [x] Host code builds on both Windows (llvm-mingw) and Linux, with UTF-8 paths. - [x] Docs updated. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01VxeaZib8ryJL5zYJuGgxcr
tests/test_hud_layout.py models MessageDialog_00 as the game lays it out: the box across the middle, and the
choices' column (N_BtnNum_00) at x 1460 with boxes 940 units wide, reaching x 1930 of the 16:9 canvas. It checks the
column and its boxes at 21:9 (held beside the box), 16:10 and 4:3 (in to the screen's right side, as far from its edge
as at 16:9) and at every shape from 5:4 to 48:9 (never further past the edge than at 16:9).

On main the column is held where the game puts it, so at 16:10 and 4:3 its boxes' right ends are off the screen: the
check at 16:10 fails (Longshot #1050).

Assisted-by: Claude Code
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01VxeaZib8ryJL5zYJuGgxcr
On a screen narrower than 16:9 the HUD keeps its 16:9 size down to 3:2, so its canvas is wider than the screen and
its side panes move in. The text box (MessageDialog_00) is held where the game puts it, choices included, and the
choices' column reaches the canvas's right edge: at 16:10 the boxes' right 200 units and their longer texts were off
the screen.

The held layout's panes are now noted, and its choices' column (N_BtnNum_00, RULE_HELD_RIGHT) moves in to the screen's
right side where the canvas is wider than the screen, as far from the edge as at 16:9. The box stays where it is, and
nothing moves on a wider screen. The check in tests/test_hud_layout.py from the commit before now passes.

Fixes #1050

Assisted-by: Claude Code
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01VxeaZib8ryJL5zYJuGgxcr
docs(tested): the text box's choices at 16:10, 4:3 and 21:9, main against the fix
Some checks failed
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Successful in 4s
CI / Conventional Commit title (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been cancelled
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been cancelled
CI / Workflow lint (actionlint) (pull_request) Has been cancelled
Guide screenshots / Guide images and local UI source review (pull_request) Has been cancelled
Third-party parts / As their notes say (pull_request) Has been cancelled
Third-party parts / The same as upstream (pull_request) Has been cancelled
Website / Website on Cloudflare (pull_request) Has been cancelled
81afb62aa3
Assisted-by: Claude Code
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01VxeaZib8ryJL5zYJuGgxcr
Author
Contributor

Tested in game: Windows 10 22H2, RTX 4090, NVIDIA 617.42, headless through the control bridge: the Market gate guard's three choices at 1920x1200 (16:10) and 1440x1080 (4:3) are cut at the right edge on main e090cc48 and whole with this change (the plan logs N_BtnNum_00 moved left 192 and 300); at 2560x1080 (21:9) they stay where main puts them.

Tested in game: Windows 10 22H2, RTX 4090, NVIDIA 617.42, headless through the control bridge: the Market gate guard's three choices at 1920x1200 (16:10) and 1440x1080 (4:3) are cut at the right edge on main e090cc48 and whole with this change (the plan logs N_BtnNum_00 moved left 192 and 300); at 2560x1080 (21:9) they stay where main puts them.
Author
Contributor

Screenshots: the Market gate guard's choices, cropped to the right of the picture. Left: Before: main @ e090cc48. Right: After: this PR @ 6f9b566c (the same change as b4dcb377). Windows 10, RTX 4090, headless at the output size, pt-BR.

1920x1200 (16:10): on main the boxes are cut at the right edge and the scroll hint (↑) is off the screen; with this PR they are whole, the hint beside them.

16:10, main left, this PR right

1440x1080 (4:3): on main the texts are cut too ("Onde fica o caste").

4:3, main left, this PR right

2560x1080 (21:9): the same place on both (the column is moved only where the screen is narrower than 16:9).

21:9, main left, this PR right

Screenshots: the Market gate guard's choices, cropped to the right of the picture. Left: **Before: main @ e090cc48**. Right: **After: this PR @ 6f9b566c** (the same change as b4dcb377). Windows 10, RTX 4090, headless at the output size, pt-BR. **1920x1200 (16:10)**: on main the boxes are cut at the right edge and the scroll hint (↑) is off the screen; with this PR they are whole, the hint beside them. ![16:10, main left, this PR right](https://git.magicbean.dev/attachments/f4217a21-6f12-4e1f-b4a5-3c1310b2bbce) **1440x1080 (4:3)**: on main the texts are cut too ("Onde fica o caste"). ![4:3, main left, this PR right](https://git.magicbean.dev/attachments/a238a9ac-c5a4-41ec-958d-b7dfd2fe5f8c) **2560x1080 (21:9)**: the same place on both (the column is moved only where the screen is narrower than 16:9). ![21:9, main left, this PR right](https://git.magicbean.dev/attachments/f789179c-db24-477e-87ce-ec8aaad4b267)
Author
Contributor

Description: the screenshots linked, a Baseline line saying why the runs on main e090cc48 hold on 623a0ca3 (hos_wide_ui.c unchanged in between; the other commits on the text box's path are #1191's speech-bubble site and audio/options changes), and the Docs box for the docs/tested files.

Description: the screenshots linked, a Baseline line saying why the runs on main e090cc48 hold on 623a0ca3 (hos_wide_ui.c unchanged in between; the other commits on the text box's path are #1191's speech-bubble site and audio/options changes), and the Docs box for the docs/tested files.
cuecuemoar changed title from WIP: fix(ui): the text box's choices stay on a 16:10 or narrower screen to fix(ui): the text box's choices stay on a 16:10 or narrower screen 2026-11-16 21:02:27 +00:00
cuecuemoar force-pushed fix/held-choices-narrow from 81afb62aa3
Some checks failed
Website / Website on Cloudflare (pull_request_target) Has been skipped
PR description / Description follows the template (pull_request_target) Successful in 4s
CI / Conventional Commit title (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Has been cancelled
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Has been cancelled
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Has been cancelled
CI / Workflow lint (actionlint) (pull_request) Has been cancelled
Guide screenshots / Guide images and local UI source review (pull_request) Has been cancelled
Third-party parts / As their notes say (pull_request) Has been cancelled
Third-party parts / The same as upstream (pull_request) Has been cancelled
Website / Website on Cloudflare (pull_request) Has been cancelled
to c55b75f221
Some checks are pending
CI / Conventional Commit title (pull_request) Blocked by required conditions
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Blocked by required conditions
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Blocked by required conditions
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Blocked by required conditions
CI / Workflow lint (actionlint) (pull_request) Blocked by required conditions
Guide screenshots / Guide images and local UI source review (pull_request) Blocked by required conditions
Third-party parts / As their notes say (pull_request) Blocked by required conditions
Third-party parts / The same as upstream (pull_request) Blocked by required conditions
Website / Website on Cloudflare (pull_request) Blocked by required conditions
PR description / Description follows the template (pull_request_target) Successful in 5s
Website / Website on Cloudflare (pull_request_target) Has been skipped
beanstalk/approval approved by llun (reviewed)
beanstalk/any 0/1 pass, 0 running, 0 queued
2026-11-16 21:07:03 +00:00
Compare
Author
Contributor

Force-pushed: the three commits' author and committer changed to cuecuemoar [email protected]; nothing else. The test commit is now 5fe2f5a6 and the fix cc8cc945 (the screenshots comment's b4dcb377 is cc8cc945); the change's patch ID is still d0d83118.

Force-pushed: the three commits' author and committer changed to cuecuemoar <[email protected]>; nothing else. The test commit is now 5fe2f5a6 and the fix cc8cc945 (the screenshots comment's b4dcb377 is cc8cc945); the change's patch ID is still d0d83118.
Member

@Beanstalk Please test with any machine.

@Beanstalk Please test with any machine.
Member

Beanstalk test farm

Commit c55b75f · tier reviewed · approved by llun · dashboard

What will be tested · default plan (AI failed: the plan was refused: probe[0]: "at_s" is required)

The AI plan failed: the plan was refused: probe[0]: "at_s" is required; using the default plan.

  • Where: any (any PC, 1 report).
  • title: from the title screen, then measure title for 20 s (after 5 s warm-up); 2 runs on each side.
  • hyrule-field: from a new save, then teleport to Hyrule Field, measure field for 60 s (after 10 s warm-up); 2 runs on each side.
  • Measured: crashes, boot time, frame times, stutters and hitches, VRAM use, the game's log.
  • The PR fails if: it crashes; or peak VRAM is more than 10% above main; or frame time (p99) is more than 15% above main.
  • Time: about 51 min of game runs per tester (8 game starts), plus the two builds.

Done. Comment @Beanstalk to run it again.

Target Reports Crash Frame p99 (ms) vs main VRAM peak (MB) vs main Result
any 0/1 — — — queued
<!-- beanstalk:sticky --> ### Beanstalk test farm Commit `c55b75f` · tier **reviewed** · approved by llun · [dashboard](https://beanstalk.magicbean.dev/prs/1233) **What will be tested** · default plan (AI failed: the plan was refused: probe[0]: "at_s" is required) The AI plan failed: the plan was refused: probe[0]: "at_s" is required; using the default plan. - **Where:** any (any PC, 1 report). - **title:** from the title screen, then measure title for 20 s (after 5 s warm-up); 2 runs on each side. - **hyrule-field:** from a new save, then teleport to Hyrule Field, measure field for 60 s (after 10 s warm-up); 2 runs on each side. - **Measured:** crashes, boot time, frame times, stutters and hitches, VRAM use, the game's log. - **The PR fails if:** it crashes; or peak VRAM is more than 10% above main; or frame time (p99) is more than 15% above main. - **Time:** about 51 min of game runs per tester (8 game starts), plus the two builds. Done. Comment `@Beanstalk` to run it again. | Target | Reports | Crash | Frame p99 (ms) vs main | VRAM peak (MB) vs main | Result | | --- | --- | --- | --- | --- | --- | | any | 0/1 | — | — | — | queued |
HumanBean deleted branch fix/held-choices-narrow 2026-11-16 21:58:35 +00:00
Some checks are pending
CI / Conventional Commit title (pull_request) Blocked by required conditions
CI / Python source checks (ubuntu-24.04, part 1/2) (pull_request) Blocked by required conditions
CI / Python source checks (ubuntu-24.04, part 2/2) (pull_request) Blocked by required conditions
CI / SPIR-V fixtures (global narrow load/store, direct surface atomics, F2I immediate) (pull_request) Blocked by required conditions
CI / Workflow lint (actionlint) (pull_request) Blocked by required conditions
Guide screenshots / Guide images and local UI source review (pull_request) Blocked by required conditions
Third-party parts / As their notes say (pull_request) Blocked by required conditions
Third-party parts / The same as upstream (pull_request) Blocked by required conditions
Website / Website on Cloudflare (pull_request) Blocked by required conditions
PR description / Description follows the template (pull_request_target) Successful in 5s
Website / Website on Cloudflare (pull_request_target) Has been skipped
beanstalk/approval approved by llun (reviewed)
beanstalk/any 0/1 pass, 0 running, 0 queued
Some workflows are waiting to be reviewed.
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
3 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!1233
No description provided.