Game: a full forward diagonal prevents rolling with a controller #88

New issue

Open

opened 2026-11-13 02:27:27 +00:00 by mooki · 0 comments

mooki commented

2026-11-13 02:27:27 +00:00

Longshot version

Carried from Hookshot v2.1.0. This remains an unresolved controller report, not a newly reproduced failure on Longshot main 6b2dafefe.

What kind of problem

Controls

What happened

AvailableMovie reported that tilting the left stick fully forward-left or forward-right and pressing the right face button makes Link keep running instead of rolling. Other directions roll normally. Their two logs identify an SDL3 device named PS4 Controller but do not record its active mapping, stick readings, layout preset or dead zone.

I took this on and reported the negative reproduction: all 43 tested actions across three complete Linux sessions rolled, including full diagonals, circular-limit diagonals, nearby angles and normal keyboard bindings supplied by scripted input. Both game-time modes were exercised. This does not rule out the Windows/physical-controller report.

Longshot inherits the inner-dead-zone repair. It preserves full diagonal values and is not a demonstrated fix for this case. No causal input-range defect was established, and no speculative extra clamp was added. A physical controller and Windows gameplay remain untested here.

The next useful evidence is a current-build retest with the same pad, compared with W+A/W+D and the bound Action key: exact model, wired/Bluetooth connection, Layout preset, Stick dead zone, mapping log and a short video. A failure witness from the actual input path is needed before changing its policy.

Where in the game

The reporter says anywhere. My completed comparisons used flat ground in Hyrule Field.

How often

Every time in the original report; not reproduced in the 43 completed scripted actions here.

Graphics card and driver version

Reporter: RTX 5080, NVIDIA 621.14. My earlier tests: RTX 4090, NVIDIA 615.78.08.

CPU, RAM and operating system

Reporter: Ryzen 7 5700X, 32 GB RAM, Windows 11. My tests used Arch Linux and a Core i9-13900K.

Changes from the defaults

Reporter did not supply input settings. My tests used a 60 cap and both Real time and Per frame. No physical key or controller press is claimed.

Other software hooked into the game

Reporter says none. Original logs remain linked from the report.

Reported on GitHub as NotRexed/Hookshot#2776; taken on there by mooki.

### Longshot version Carried from Hookshot v2.1.0. This remains an unresolved controller report, not a newly reproduced failure on Longshot main `6b2dafefe`. ### What kind of problem Controls ### What happened [AvailableMovie reported](https://github.com/NotRexed/Hookshot/issues/2776) that tilting the left stick fully forward-left or forward-right and pressing the right face button makes Link keep running instead of rolling. Other directions roll normally. Their two logs identify an SDL3 device named PS4 Controller but do not record its active mapping, stick readings, layout preset or dead zone. I took this on and [reported the negative reproduction](https://github.com/NotRexed/Hookshot/issues/2776#issuecomment-6832475909): all 43 tested actions across three complete Linux sessions rolled, including full diagonals, circular-limit diagonals, nearby angles and normal keyboard bindings supplied by scripted input. Both game-time modes were exercised. This does not rule out the Windows/physical-controller report. Longshot inherits [the inner-dead-zone repair](https://git.magicbean.dev/MagicBeanDev/Longshot/commit/284c912123cc06fae39ba2061ce3beadf2e9c212). It preserves full diagonal values and is not a demonstrated fix for this case. No causal input-range defect was established, and no speculative extra clamp was added. A physical controller and Windows gameplay remain untested here. The next useful evidence is a current-build retest with the same pad, compared with W+A/W+D and the bound Action key: exact model, wired/Bluetooth connection, Layout preset, Stick dead zone, mapping log and a short video. A failure witness from the actual input path is needed before changing its policy. ### Where in the game The reporter says anywhere. My completed comparisons used flat ground in Hyrule Field. ### How often Every time in the original report; not reproduced in the 43 completed scripted actions here. ### Graphics card and driver version Reporter: RTX 5080, NVIDIA 621.14. My earlier tests: RTX 4090, NVIDIA 615.78.08. ### CPU, RAM and operating system Reporter: Ryzen 7 5700X, 32 GB RAM, Windows 11. My tests used Arch Linux and a Core i9-13900K. ### Changes from the defaults Reporter did not supply input settings. My tests used a 60 cap and both Real time and Per frame. No physical key or controller press is claimed. ### Other software hooked into the game Reporter says none. Original logs remain linked from the report. Reported on GitHub as NotRexed/Hookshot#2776; taken on there by mooki.