Re: Low Latency User Input

And CC +public-games@w3.org <public-games@w3.org> for more awareness of
work to improve user input latency.

On Wed, Jan 17, 2024 at 9:45 AM Vincent Scheib <scheib@google.com> wrote:

> Chiming in with my support for this!
>
> On Mon, Jan 15, 2024 at 4:48 PM Marcos Caceres <caceres_m@apple.com>
> wrote:
>
>> Hi Carl,
>>
>> Thanks again for taking the time to outline the issues. I’ve proposed
>> this as a topic for the upcoming W3C Breakout day (12th of March):
>> https://github.com/w3c/breakouts-day-2024/issues/2
>>
>> Details for the W3C Breakout day are here:
>> https://github.com/w3c/breakouts-day-2024
>>
>> Hopefully we can get the right folks in a virtual room to discuss some
>> options and a path forward.
>>
>> Kind regards,
>> Marcos
>>
>> > On Dec 9, 2023, at 13:51, Carl Younger <carl.joseph.younger@gmail.com>
>> wrote:
>> >
>> > Hey guys,
>> >
>> > Despite many advances in recent years, the Web remains a poor choice of
>> platform for any application that depends on responding to user input with
>> low latency.
>> >
>> > The APIs that handle user input (keyboard, pointer, gamepad, HID, USB,
>> MIDI, Bluetooth, serial etc) generally require registering handlers on the
>> main thread. As the main thread is often busy, handlers cannot (reliably
>> and consistently) respond to input with low enough latency for certain
>> applications.
>> >
>> > Affected applications include any video game that demands precise
>> inputs (which is a large subset), pretty much all DAWs and synthesizers, as
>> well as art programs that use touch gestures to emulate brushstrokes etc.
>> >
>> > The situation might be improved somewhat by extending input APIs to
>> include timestamps, but exposing the APIs to Web Workers is a much better
>> solution.
>> >
>> > Some of the input APIs can extend their spec to support workers
>> relatively easily, and there has been some progress there already
>> (including some implementation). Other APIs are more problematic. For
>> example, keyboard and pointer events are part of the DOM API, which marries
>> them to the main thread.
>> >
>> > I recently discussed all this with Marcos Cáceres (and others). Marcos
>> advised me to contact this mailing list, as I feel that these issues will
>> require a macroscopic perspective (beyond the scope of any specific API)
>> and a substantial community effort to address.
>> >
>> > Thank you for taking the time to consider my post.
>> > Much appreciated -- Carl
>> >
>> > The original conversation: github.com/w3c/gamepad/issues/37
>>
>>
>>