You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
v0.8.0-rc0: XInput, SInput touchpad, docs & OS compatibility overhaul #346
This PR builds on top of the excellent work by @LeeNX on the original SInput implementation (PR #341). Their research, pull requests, and ongoing help with issues were instrumental in getting SInput off the ground — huge thanks for all of that.
This is quite a large PR overall, so a heads-up that there's been some refactoring behind the scenes to make it easier to add new controller types going forward. The core sendReport() was split into mode-specific paths (sendSInputReport(), sendXInputReport(), sendGenericReport()), and the config auto-corrects its defaults based on the selected mode. The idea is that adding something like XInput (included here) or potentially a future DualSense mode doesn't require bolting unrelated code onto existing paths — each mode gets its own clean implementation while sharing the common BLE/HID infrastructure.
Summary of what's in this PR:
XInput mode — Xbox One S / Series X emulation with rumble
Just a few notes on docs, we should point out how Generic vs XInput vs SInput is found by host OS, I am not sure everybody would know that it is via VID:PID and that changing that might not work on a host. I did not see if we throw any warnings if changed.
Second, please hold off on merging and releasing. I am about to put up a PR that gives basic hardware-in-the-loop testing. I will donate a Raspberry Pi 3B+ and two MCUs (ESP32 and ESP32-C3) for basic CI/Unit type testing, that will run in my homeLab and will be contactable in CI workflow over TailScale VPN node. Just want to finish testing in my fork and see what I need to put out for other people test in a similar way.
Point 1
I've now updated the docs to reflect your advice and put a check in the code if the vid/pid are changed on SInput/XInput.
I could just override the ability for that to happen. That would be easy enough
Point 2
Sounds great. You've ben talking about doing it for years!
A lot of love shown for the library over the past week or so...
Point 1 I've now updated the docs to reflect your advice and put a check in the code if the vid/pid are changed on SInput/XInput. I could just override the ability for that to happen. That would be easy enough
Just something I think could trip up a new user. I remember that SDL had some env overrides to enable VID:PID for a subsystem, but I did not see this for SInput, at least not yet. A warning or docs should be enough.
Point 2 Sounds great. You've ben talking about doing it for years! A lot of love shown for the library over the past week or so...
Another project might have use for some of these parts, so they worth putting in place.
I don't think Dual Sense offers anything that SInput or XInput currently does not support. If somebody else wants to it and it does not add too much code or size, maybe , but I think there is plenty to work though at the moment.
I think XInput and SInput support will close out most of the issues/requests people had with the library.
Totally agree.
Thoughts?
Comments quotes above.
Can you think of anything further to do?
I think you covered everything that I can think of.
SInput now actually works - Power level, Player Led, rumble tested etc working - more testing needed
XInput now compiles, but is being detected as generic pad - More tweaking and testing needed <- Now fixed and working
Let me try and finish off the HIL tester and see if we can get that in with a stable release today. Just looking to merge #347 with any changes need for the CI to run on real hardware.
Activity
lemmingDev commentedon Aug 30, 2026
This PR builds on top of the excellent work by @LeeNX on the original SInput implementation (PR #341). Their research, pull requests, and ongoing help with issues were instrumental in getting SInput off the ground — huge thanks for all of that.
This is quite a large PR overall, so a heads-up that there's been some refactoring behind the scenes to make it easier to add new controller types going forward. The core sendReport() was split into mode-specific paths (sendSInputReport(), sendXInputReport(), sendGenericReport()), and the config auto-corrects its defaults based on the selected mode. The idea is that adding something like XInput (included here) or potentially a future DualSense mode doesn't require bolting unrelated code onto existing paths — each mode gets its own clean implementation while sharing the common BLE/HID infrastructure.
Summary of what's in this PR:
Would appreciate testing feedback — especially on Steam (both SInput and XInput), macOS via GCController, and Linux with xpad/SDL3. Thanks!
lemmingDev commentedon Aug 30, 2026
Updated: Examples rewrite + axes default change
All examples have been rewritten and the axes default has been restored to -32768 to 32767 (was changed to 0-32767 in v5, restored in v0.8.0).
New SInput examples (8 focused per-feature)
Each example demonstrates ONE feature with thorough comments, a minimal button heartbeat, and hardware wiring notes:
Updated XInput examples
Both XInputGamepad and XInputSeriesX have been rewritten with thorough inline comments covering:
Compile results
All 10 new/updated examples compile with 0 warnings on ESP32:
Generic examples also compile clean (TestFeatureReports has a pre-existing FEATURE_REPORT_LEN redefinition warning unrelated to these changes).
Branch: multi-gamepad (force-pushed)
LeeNX commentedon Aug 30, 2026
Wow @lemmingDev !
That is awesome!
Just a few notes on docs, we should point out how Generic vs XInput vs SInput is found by host OS, I am not sure everybody would know that it is via VID:PID and that changing that might not work on a host. I did not see if we throw any warnings if changed.
Second, please hold off on merging and releasing. I am about to put up a PR that gives basic hardware-in-the-loop testing. I will donate a Raspberry Pi 3B+ and two MCUs (ESP32 and ESP32-C3) for basic CI/Unit type testing, that will run in my homeLab and will be contactable in CI workflow over TailScale VPN node. Just want to finish testing in my fork and see what I need to put out for other people test in a similar way.
lemmingDev commentedon Aug 30, 2026
Thanks for the feedback @LeeNX
Point 1
I've now updated the docs to reflect your advice and put a check in the code if the vid/pid are changed on SInput/XInput.
I could just override the ability for that to happen. That would be easy enough
Point 2
Sounds great. You've ben talking about doing it for years!
A lot of love shown for the library over the past week or so...
Scope
I could add Dual Sense in, but I don't really think its worth doing since https://github.com/Mystfit/ESP32-BLE-CompositeHID already does it
I think XInput and SInput support will close out most of the issues/requests people had with the library.
Thoughts?
Can you think of anything further to do?
LeeNX commentedon Aug 30, 2026
Totally welcome.
Just something I think could trip up a new user. I remember that SDL had some env overrides to enable VID:PID for a subsystem, but I did not see this for SInput, at least not yet. A warning or docs should be enough.
Another project might have use for some of these parts, so they worth putting in place.
I don't think Dual Sense offers anything that SInput or XInput currently does not support. If somebody else wants to it and it does not add too much code or size, maybe , but I think there is plenty to work though at the moment.
Totally agree.
Comments quotes above.
I think you covered everything that I can think of.
lemmingDev commentedon Aug 30, 2026
This branch has a few issues. AM working through to resolve the SInput and XInput stuff
I'll post when they're sorted
lemmingDev commentedon Aug 31, 2026
SInput now actually works - Power level, Player Led, rumble tested etc working - more testing needed
XInput now compiles, but is being detected as generic pad - More tweaking and testing needed <- Now fixed and working
lemmingDev commentedon Sep 1, 2026
So - it's mostly all working now - need testing
LeeNX commentedon Sep 2, 2026
Let me try and finish off the HIL tester and see if we can get that in with a stable release today. Just looking to merge #347 with any changes need for the CI to run on real hardware.
lemmingDev commentedon Sep 4, 2026
Yeah - best to get that sorted before these thousands of new lines are thrown into the mix
lemmingDev commentedon Sep 11, 2026
this branch is superseded by the new NuS brach described here: #352