In reply to @asdil12:im.f3l.de

I didn't really use it because the hid input only makes sense with video capturing working, but the current implementation works hardly at all (if it works, there is a huge delay measured in minutes). I suspect due to the PPM frame handling in perl.

So the following is implemented (using a rpi pico to emulate a keyboard) but not really used due to the above issue.

For my hid input, I just need to send http requests - with strings or keyboard shortcuts in the same syntax as qemu (and therefore the testapi) is using - as a get param.
So as long as that is possible, I'm fine.

re delay - I have some, but definitely in seconds (single digit), not minutes; I switched some workers to ustreamer instead of ffmpeg, but didn't improved things much

re input protocol - I want to introduce some confirmation for each action, as right now it may happen several actions (especially mouse) can queue in the stdin pipe, which throws off delays that openqa assumes (between keys, after clicks etc); so, yes, converting to http would still be possible, but the converting script (if you have it already) will need adjustment to print something (probably just "ok") after sending the request (and receiving the response)

In reply to @okurz:matrix.org
element.reaffirm: see room topic: Please direct general openSUSE support questions to https://matrix.to/#/#support:opensuse.org

In reply to @foursixnine_:matrix.org
marmarek: mind sharing some details (if you can) of how that works? I'd like to add some testing for these, and intel (laptops and desktop systems) but I'm in the research state at the moment