How I can get 5x keyboadrs and mouses streams from different computers on LSL?

[Tkachenko-Denis]

When I try to run keyboard.exe on different computers, I only see 1 stream in labrecorder. What can I do to get a couple or more keyboard streams in labrecorder?

[cboulay]

Ah, yeah that'll be a problem because each stream will appear nearly identical. I thought the latest version of LabRecorder differentiated streams based on host, so you should get them.

If that's failing, the fastest way to accommodate this use-case is to modify the C++ code so the entry point can accept command-line arguments, and add a command-line argument for Stream Name.

[cboulay]

I just took a quick look at the LabRecorder and I did find the code that uses the host name to help sort. So they should be identified independently in LabRecorder.

Without digging in to the LSL code right now, I suppose it's possible that this is failing because there is no unique source id.

Whether or not that's the case, we should probably add a source id to the stream info instantiation. The source id needs to be unique to each computer so this should probably be generated at runtime based on host id. In other applications I've used a hash for this.

Is this something you can tackle?

[Tkachenko-Denis]

I don't really understand much about the LSL protocol, I'm just learning. I found an old archive that has a program for capturing marks from the keyboard. (2013 if I'm not mistaken). I assumed it was an application conflict, but I couldn't solve it with C++

[dmedine]

These apps should probably be updated to use .NET anyway. That way they could be cross-platform.

[Tkachenko-Denis]

less than 2 weeks have passed since I learned how to compile through VS using CMake. Finally. Thank you very much for your help @cboulay, I'll try to understand what you mean =)

[tstenner]

These apps should probably be updated to use .NET anyway. That way they could be cross-platform.

At least for those Linux / macOS-users that have the correct .NET runtime installed.

Some points against it:

• at least on Linux there's backends (e.g. evdev) with way better temporal resolution, timestamped events, per-device queues and exclusive mode, i.e. you open "keyboard 1" and all keypresses go exclusively to your process without interfering with the UI you already have
• the archive with the two current input apps and liblsl is <1MB, everything (well, except the msvc runtime) included
• no JIT, FFI or GC introducing jitter