-----------------------------------------
Ensure that we set bShouldBeIgnored AFTER we retain the state of old triggers and modifiers. Otherwise we would never actually update the state of the mapping's bShouldBeIgnored flag and it would always be false. I believe that this issue was introduced with 24566913. Once we got the right behavior for that comparison operator, then the mapping would always be converted instead of the old behavior where it would not. This resulted in it overriding the bShouldBeIgnored flag to whatever the previous state was.
#jira UE-196939
#jira UE-206413
#rnx
[FYI] Richard.Graham
[CL 31354440 by ben hoffman in ue5-main branch]
Ensure that we set bShouldBeIgnored AFTER we retain the state of old triggers and modifiers. Otherwise we would never actually update the state of the mapping's bShouldBeIgnored flag and it would always be false. I believe that this issue was introduced with 24566913. Once we got the right behavior for that comparison operator, then the mapping would always be converted instead of the old behavior where it would not. This resulted in it overriding the bShouldBeIgnored flag to whatever the previous state was.
#jira UE-196939
#rb dan.oconnor
#changelist validated
#virtualized
[CL 31227708 by ben hoffman in ue5-main branch]
FMockedEnhancedInputSubsystem is now a UCLASS so it can properly implement the managed map for the interface.
#rb ben.hoffman
#tests PIE touch input simulation, ran input automation tests
[CL 30241210 by oscar burga in ue5-main branch]
Split unreachable clusters destruction to DissolveUnreachableClusters (non-incremental as it's very fast)
Flipping MaybeUnreachable -> Unreachable flag is now a single Swap() function call (MaybeUnreachable / Unreachable flags are deprecated and replaced with interchangable Reachability flags)
GatherUnreachableObjects will now only gather unreachable objects for NotifyUnreachableObjects call and perf gains during unhashing and purging all objects.
Enable with gc.AllowIncrementalGather=1
#rb Johan.Torp, PJ.Kack
[CL 30033914 by robert manuszewski in ue5-main branch]
It isn't technically incorrect for an IMC to contain a mapping to the invalid key, but it can be hard to track down which mapping is empty if you are working at a large scale. This check will hopefully make it easier to find.
#rb Phillip.Kavan
[CL 29243815 by ben hoffman in ue5-main branch]
This was resulting in crashes in Android and PIE with touch input simulation due to touch input modifiers becoming unreachable / deleted while they were still in use.
#rb ben.hoffman
[CL 28926504 by tiago binz in ue5-main branch]
The use case for this is mobile UI where you want to use an input trigger for something like a gesture and you want the input trigger to be created once for a thumbstick and a touch screen.
Copying the values of triggers and modifiers in the ContinousInjection so that we can evaluate a new state instead of using the old one
#jira UE-194659
#rb [at]Ben.Hoffman
[CL 28038276 by marius ursu in ue5-main branch]
Backing this out because it caused any existing BP modifier implementations to have cook errors. Instead of removing the const we would need to properly deprecate it and make a new version to support old content
#jira UE-195954
[FYI] ben.hoffman
Original CL Desc
-----------------------------------------------------------------
PR #10879: Make UInputModifier::ModifyRaw non-const
#rb me
#jira UE-195489
[CL 28038090 by LtRandolph in ue5-main branch]
It provides a normalized smoothed difference between the current and last input value, which options on how the value is calculated for different behavior.
For boolean Input Values, it just returns the boolean as is.
#jira UE-190264
#rb me
[CL 26893500 by najmm01 in ue5-main branch]
