When using _update60(), btnp() timing is also 2x faster, which makes it much less useful - the time to let go before the press is repeated is halved to 0.12s, and time between repeats also goes down.

This matches the docs (which say frames), but doesn't feel like it was intended - maybe something to be fixed for 0.1.9?

I am also seeing an issue here where my checks for button presses are registering twice when on "non-60fps" machines. Is there some other flag I need to check for (or is this indeed a bug)?

(NOTE: I am seeing this using Zep's override code to force my 60fps optimised code to run in 30fps - so dunno if it's a true test?)

with the override you're calling _update60 twice yourself, so the 'system' does not poll buttons in between. for now this also happens with the html player since it ignores _update60 altogether so you have to use the override trick. no idea about other 'non-60fps-machines'.

there's no documented* way around it short of writing your own version of btnp. as an added benefit you'll have your own (non buggy) repeat delays.

*(edited)

Thanks for the explanation @ultrabrite, makes sense :)

If you're calling _update60() twice yourself, I think you also need to call _update_buttons() an extra time between them.

Thanks @Felice, didn't know you could do that! :D