FIX: uum 138143 button press points for stickcontrol and vector2 add functionality by Darren-Kelly-Unity · Pull Request #2411 · Unity-Technologies/InputSystem

Darren-Kelly-Unity · 2026-04-29T15:06:02Z

Purpose of this PR

Fixes UUM-138143 — InputAction.IsPressed() / in-progress style checks could fire at the wrong effective magnitude for Vector2 / stick bindings when developers set pressPoint on the control or on PressInteraction, instead of honoring those thresholds consistently.

Introduces IActuationPressPoint, implements it on ButtonControl, Vector2Control, and (via Vector2Control) StickControl, and threads pressPoint / pressPointOrDefault through InputAction, InputActionState, and InputControlExtensions so press-style polling matches the documented interaction defaults. Adds ActuationPressPointTests and refreshes RespondingToActions.md.

Release Notes

UUM-138143 — InputAction.IsPressed() (and related press / in-progress behavior) for actions bound to Vector2Control and StickControl now respects per-control pressPoint and binding PressInteraction.pressPoint, aligning with ButtonControl and fixing incorrect early triggers when rewriting default interaction thresholds.

Functional Testing status

Automated: new ActuationPressPointTests and updates in CoreTests_Actions / CoreTests_Controls cover press-point actuation paths.
Manual: repro project IN-136917 / scene Input Bug from the ticket; verify console no longer shows IsPressed before processed magnitude reaches the configured press threshold for vector / stick cases.

Performance Testing Status

I have added performance testing for the hot path methods that Hakan flagged and I don't see any major issues when comparing. There is a slight increase in cost of performance of about 0.1ms for 1000 iterations on the GetActuationPressThreshold() method that was asked to be checked.

No new per-frame allocations identified in scope.

Overall Product Risks

Technical Risk: 2 — Touches InputActionState and widely used press APIs; edge cases around magnitude, deadzones, and interactions need regression confidence.

Halo Effect: 2 — Any title using IsPressed / WasPressedThisFrame on Vector2 / stick actions may see timing changes until bindings are tuned; behavior is intended to match documented pressPoint semantics.

Class diagram

Git range analyzed: origin/develop...HEAD (4 commits).

Mermaid class diagram (GitHub)

classDiagram
  direction TB

  class IActuationPressPoint {
    <<interface>>
    +pressPoint
    +pressPointOrDefault
  }

  class ButtonControl {
    +pressPoint
    +pressPointOrDefault
  }

  class Vector2Control {
    +pressPoint
    +pressPointOrDefault
  }

  class StickControl {
  }

  class InputAction {
    +IsPressed()
    +WasPressedThisFrame()
    +WasReleasedThisFrame()
  }

  class InputActionState {
  }

  class InputControlExtensions {
    <<static>>
  }

  IActuationPressPoint <|.. ButtonControl
  IActuationPressPoint <|.. Vector2Control
  Vector2Control <|-- StickControl
  InputAction --> InputActionState : uses
  InputControlExtensions ..> IActuationPressPoint : pressPointOrDefault

Documentation Impact

Changes analyzed: origin/develop...HEAD (4 commits).

User-facing: New public IActuationPressPoint; InputAction press-helper remarks/behavior; ButtonControl, Vector2Control, StickControl docs around pressPoint; RespondingToActions.md updated for polling vs pressPoint.

Documentation updates in this PR

RespondingToActions.md — Clarifies IsPressed / frame helpers vs interactions and pressPoint; review wording and cross-refs after merge.

Follow-up (optional)

If QA/docs want a dedicated IActuationPressPoint callout in a control reference page, add a short subsection and link from Responding to Actions.

Jira (fetched via MCP): Bug, In Progress, TBD priority, assignee Darren Kelly, label Input-Actions. Ticket summary: InputAction.IsPressed() and IsInProgress() triggers on wrong values when rewriting the default values of Vector2 interactions.

…ld used for stick controls and Vector2. Test

…at a user declared as an interaction.

u-pr · 2026-04-29T15:06:20Z

⚠️ Review skipped — this PR is too large to process.

The diff contains approximately 325,668 tokens (~977,005 characters), which exceeds the maximum limit of 150,000 tokens.

Please consider splitting this PR into smaller, focused changes.

_{🤖 Helpful? 👍/👎}

codecov-github-com · 2026-04-29T15:17:50Z

Codecov Report

Attention: Patch coverage is 72.14137% with 134 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...sets/Tests/InputSystem/ActuationPressPointTests.cs	57.46%	131 Missing ⚠️
Assets/Tests/InputSystem/CoreTests_Actions.cs	97.24%	3 Missing ⚠️

@@             Coverage Diff             @@
##           develop    #2411      +/-   ##
===========================================
+ Coverage    78.13%   78.22%   +0.09%     
===========================================
  Files          483      762     +279     
  Lines        98770   140555   +41785     
===========================================
+ Hits         77169   109946   +32777     
- Misses       21601    30609    +9008

Flag	Coverage Δ
inputsystem_MacOS_2022.3_project	`75.31% <70.28%> (-0.10%)`	⬇️
inputsystem_MacOS_6000.0_project	`77.19% <70.28%> (-0.11%)`	⬇️
inputsystem_MacOS_6000.3_project	`77.19% <70.28%> (-0.08%)`	⬇️
inputsystem_MacOS_6000.4_project	`77.20% <70.28%> (-0.08%)`	⬇️
inputsystem_MacOS_6000.5_project	`77.23% <70.28%> (-0.09%)`	⬇️
inputsystem_MacOS_6000.6_project	`77.23% <70.28%> (-0.08%)`	⬇️
inputsystem_Ubuntu_2022.3_project	`75.21% <70.28%> (-0.10%)`	⬇️
inputsystem_Ubuntu_6000.0_project	`77.10% <70.28%> (-0.11%)`	⬇️
inputsystem_Ubuntu_6000.3_project	`77.10% <70.28%> (-0.08%)`	⬇️
inputsystem_Ubuntu_6000.4_project	`77.11% <70.28%> (-0.08%)`	⬇️
inputsystem_Ubuntu_6000.5_project	`77.14% <70.28%> (-0.09%)`	⬇️
inputsystem_Ubuntu_6000.6_project	`77.14% <70.28%> (-0.08%)`	⬇️
inputsystem_Windows_2022.3_project	`75.43% <70.28%> (-0.10%)`	⬇️
inputsystem_Windows_6000.0_project	`77.32% <70.28%> (-0.11%)`	⬇️
inputsystem_Windows_6000.3_project	`77.32% <70.28%> (-0.08%)`	⬇️
inputsystem_Windows_6000.4_project	`77.32% <70.28%> (-0.08%)`	⬇️
inputsystem_Windows_6000.5_project	`77.37% <70.28%> (-0.08%)`	⬇️
inputsystem_Windows_6000.6_project	`77.37% <70.28%> (-0.08%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
Assets/Tests/InputSystem/CoreTests_Controls.cs	`99.80% <100.00%> (-0.02%)`	⬇️
...tsystem/InputSystem/Runtime/Actions/InputAction.cs	`92.12% <100.00%> (ø)`
...em/InputSystem/Runtime/Actions/InputActionState.cs	`92.42% <100.00%> (ø)`
...m/Runtime/Actions/Interactions/PressInteraction.cs	`88.60% <ø> (ø)`
...stem/InputSystem/Runtime/Controls/ButtonControl.cs	`88.37% <100.00%> (ø)`
...tSystem/Runtime/Controls/InputControlExtensions.cs	`76.12% <100.00%> (ø)`
...ystem/InputSystem/Runtime/Controls/StickControl.cs	`100.00% <ø> (ø)`
...tem/InputSystem/Runtime/Controls/Vector2Control.cs	`100.00% <100.00%> (ø)`
Assets/Tests/InputSystem/CoreTests_Actions.cs	`98.89% <97.24%> (+0.48%)`	⬆️
...sets/Tests/InputSystem/ActuationPressPointTests.cs	`57.46% <57.46%> (ø)`

... and 242 files with indirect coverage changes

ℹ️ Need help interpreting these results?

…-stickcontrol-and-vector2-add-functionality # Conflicts: # Packages/com.unity.inputsystem/InputSystem/Runtime/Controls/InputControlExtensions.cs

ekcoh

@Darren-Kelly-Unity Would it be possible to get rid of all the whitespace diffs on this PR? It would be much easier to review if this was corrected.

ekcoh · 2026-05-06T12:58:31Z

        /// </summary>
        public float pressPointOrDefault => pressPoint > 0 ? pressPoint : s_GlobalDefaultButtonPressPoint;

+        float IActuationPressPoint.pressPoint => pressPoint;


To me it look odd that presspoint (new and previous is modelled on the control instead of on the interaction), what happens when two interactions with different press points bind to the same button?

Confirmed this with AI but you can see in InputActionState.GetActuationPressThreshold() that:

There is an order, it is control-first for device semantics, interaction when the control does not override, then the default settings value.

AI Brief answer:

"Press threshold on the control models the physical control’s default; each binding’s interaction can still use its own press point when the control does not fix one, and separate bindings to the same button keep separate interaction state and thresholds unless the control’s own pressPoint is set, in which case that value wins for unified resolution."

There is also a longer text/ investigation done if you want to see that too?

…-stickcontrol-and-vector2-add-functionality

… it was implicitly implemented.

ekcoh

Reviewed, I had quite a lot of questions and some remaining from last review so leaving this as comment-only for now.

ekcoh · 2026-05-12T08:44:24Z

+        /// <summary>
+        /// Effective press threshold: <see cref="pressPoint"/> when set, otherwise the global default.
+        /// </summary>
+        float pressPointOrDefault { get; }


Why do we need two? Would it be better to have one, e.g. pressPoint that returns default if not having a layout-configured press threshold? Since both go through an interface anyway I wouldn't expect it to be an optimisation?

This is just following the format that was already there for now, not sure about removing one as it was publicly available before.

ekcoh · 2026-05-12T08:46:48Z

                    if (interactionState.totalTimeoutCompletionTimeRemaining > 0)
                    {
-                        return (interactionState.totalTimeoutCompletionDone + timerCompletion * interactionState.timerDuration)  /
+                        return (interactionState.totalTimeoutCompletionDone + timerCompletion * interactionState.timerDuration) /


Looks like pure formatting diff and could be reverted

I have left a few of the formatting changes in for now as they are correct. This one also looks ok I think?

ekcoh · 2026-05-12T08:55:19Z

+            if (count == 0)
+                return null;
+            var start = bindingStateForInteractions->interactionStartIndex;
+            for (var i = 0; i < count; ++i)


Corner case: I am confused around this code, it looks like if there are multiple PressInteractions it will return the pressPoint of the first one? While this is probably OK, it might be good documenting it on PressInteraction or similar?

Yes, It was already this way for buttonControl's previously. I have added documentation for this now.

ekcoh · 2026-05-12T08:57:27Z

+        private float GetActuationPressThreshold(InputControl control, BindingState* bindingStatePtr)
+        {
+            var bindingForInteractions = GetBindingStateForInteractionParameters(bindingStatePtr);
+            var explicitPress = TryGetExplicitPressInteractionPressPoint(bindingForInteractions);


This cost is paid also when the first return is triggered. Did you compare any performance benchmark for this to rule out that these changes do not regress performance for various press configurations? Might be worthwhile to check since we have a combination here of virtual calls, fixed overhead evaluation, type checks etc.

ekcoh · 2026-05-12T09:00:20Z

            {
-                if (control is ButtonControl button)
-                    buttonPressPoint = button.pressPointOrDefault;
+                if (control is IActuationPressPoint actuation)


Beware that this type check might be more expensive than the previous. Previously checking against a specific type, now checking for multiple types implementing interface. Its probably negligible but might we worth checking since hot-path.

I added performance tests for this vs the legacy system, we can maybe chat about it in a call if that makes sense but I don't see much change here.

…ctor2Control.cs Co-authored-by: Håkan Sidenvall <hakan.sidenvall@unity3d.com>

…ves." This reverts commit bce3776.

…coverage shown.

…es test coverage shown." This reverts commit 1243bc4.

…t improves test coverage shown."" This reverts commit db81e01.

…see if it improves test coverage shown.""" This reverts commit 8ff8746.

Darren-Kelly-Unity added 4 commits April 28, 2026 17:36

Update documentation for press point usage & default settings thresho…

ede1b9f

…ld used for stick controls and Vector2. Test

Add fix for Vector2Controls & StickControls not using pressPoint's th…

4786635

…at a user declared as an interaction.

Add documentation for changes.

85e63df

Separate and add tests for press point actuation.

9255610

Darren-Kelly-Unity changed the title ~~Bugfix/uum 138143 button press points for stickcontrol and vector2 add functionality~~ FIX: uum 138143 button press points for stickcontrol and vector2 add functionality Apr 29, 2026

Darren-Kelly-Unity added 2 commits May 5, 2026 12:02

Merge branch 'develop' into bugfix/uum-138143-button-press-points-for…

3a30637

…-stickcontrol-and-vector2-add-functionality # Conflicts: # Packages/com.unity.inputsystem/InputSystem/Runtime/Controls/InputControlExtensions.cs

Remove old comment in test.

1edbc32

Darren-Kelly-Unity requested review from MorganHoarau and ekcoh May 5, 2026 10:23

Add formatting!

f5500dc

ekcoh reviewed May 6, 2026

View reviewed changes

Darren-Kelly-Unity and others added 15 commits May 6, 2026 16:15

Merge branch 'develop' into bugfix/uum-138143-button-press-points-for…

6b3acd1

…-stickcontrol-and-vector2-add-functionality

Merge branch 'develop' into bugfix/uum-138143-button-press-points-for…

89c29e4

…-stickcontrol-and-vector2-add-functionality

Add test format commit!

509fd55

Add missing meta files.

04e7d31

Add more meta files. Not sure why they show up though.

fb11b47

Remove redundant IActuationPressPoint.pressPointOrDefault property as…

2e84fa9

… it was implicitly implemented.

Move positioning of press point forwarding properties.

577945b

Revert weird meta file changes.

e09e2c1

Update whitespace changes for InputAction.cs

24300ba

Fix more white space formatting issues.

d392a56

Fix StickControl white space.

b1fbc4f

Fix whitespace in Vector2Control.cs

0e622e4

Fix more whitespace.

f323b65

Remove fomratting issues with InputControlExtensions.cs

e4f0290

Add another batch of whitespace fixing.

bb4e745

ekcoh reviewed May 12, 2026

View reviewed changes

Update Packages/com.unity.inputsystem/InputSystem/Runtime/Controls/Ve…

773a44a

…ctor2Control.cs Co-authored-by: Håkan Sidenvall <hakan.sidenvall@unity3d.com>

Darren-Kelly-Unity and others added 8 commits May 12, 2026 12:35

Update Packages/com.unity.inputsystem/InputSystem/Runtime/Controls/Ve…

6a08e2f

…ctor2Control.cs Co-authored-by: Håkan Sidenvall <hakan.sidenvall@unity3d.com>

Fix more whitespace issues.

ceb494c

Fix compiler error.

2b72b4e

Make IActuationPressPoint.cs internal.

ef60059

Refactor and add documentation for PR review.

d7d7762

Refactor for code review.

c1cf34d

Add performance tests addressing Hakan's worrys about hot path issues.

26e92f4

Test updating codecov.yml to see if code coverage shown improves.

bce3776

Darren-Kelly-Unity requested a review from Pauliusd01 May 12, 2026 14:40

Darren-Kelly-Unity added 6 commits May 12, 2026 16:45

Revert "Test updating codecov.yml to see if code coverage shown impro…

7209d0c

…ves." This reverts commit bce3776.

Update documentation.

674c649

Test yamato configuration to exlude tests to see if it improves test …

1243bc4

…coverage shown.

Revert "Test yamato configuration to exlude tests to see if it improv…

db81e01

…es test coverage shown." This reverts commit 1243bc4.

Revert "Revert "Test yamato configuration to exlude tests to see if i…

8ff8746

…t improves test coverage shown."" This reverts commit db81e01.

Revert "Revert "Revert "Test yamato configuration to exlude tests to …

a9facb6

…see if it improves test coverage shown.""" This reverts commit 8ff8746.

Conversation

Darren-Kelly-Unity commented Apr 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Purpose of this PR

Release Notes

Functional Testing status

Performance Testing Status

Overall Product Risks

Class diagram

Documentation Impact

Documentation updates in this PR

Follow-up (optional)

Uh oh!

u-pr Bot commented Apr 29, 2026

Uh oh!

codecov-github-com Bot commented Apr 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

ekcoh left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

ekcoh left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Darren-Kelly-Unity commented Apr 29, 2026 •

edited

Loading

codecov-github-com Bot commented Apr 29, 2026 •

edited

Loading