This might only be an issue in non-secure contexts, but in Firefox, the lower precision of performance.now() seems to be making the poll diff slower, because the timestamp isn't able to properly express that something changed, until the next number rolls over. So I can try to work around this by one way or another, we'll see. Right now, using the timestamp as a diff between Gamepad object changes works really well, and from the docs, it seems to be how one is meant to see if any key changes were made since the last reference.
This might only be an issue in non-secure contexts, but in Firefox, the lower precision of
performance.now()seems to be making the poll diff slower, because the timestamp isn't able to properly express that something changed, until the next number rolls over. So I can try to work around this by one way or another, we'll see. Right now, using the timestamp as a diff betweenGamepadobject changes works really well, and from the docs, it seems to be how one is meant to see if any key changes were made since the last reference.