Re: WebXR Device API call agenda, Mar 20th 2018

Hello everyone.

Here are the notes from the call yesterday. Please, as always, feel free to
correct if something is incorrect or needs more context/information.



* - Updates- Mozilla announced that they have a relationship with Magic
Leap but cannot comment much more than that.- Magic Leap made their SDK
public. Cannot do much more than using the simulator. They are very Unity
heavy. It allows to learn a bit about their APIs/approaches. A brief
overview shows that their tech might fit pretty well to the current
approach with WebXR.- There was a presentation of OpenXR during GDC.
https://www.khronos.org/assets/uploads/developers/library/2018-gdc-webgl-and-gltf/OpenXR_A-look-at-OpenXR_Mar18.pdf
<https://www.khronos.org/assets/uploads/developers/library/2018-gdc-webgl-and-gltf/OpenXR_A-look-at-OpenXR_Mar18.pdf>
- Ana: Some activity on the thread about testing. Waiting to hear a bit
more about Mozilla and their approach with the web driver API. There are no
one from Mozilla in the call that could comment on it (either at GDC or off
line). It would be nice to have some comments from MSFT.- Raphael:
Apologies for not making comments yet. Edge seems to have a very advanced
system. The conversation should be focused on fully relying on JS or using
the Web Driver API. How are you expecting to organize the code.- Ana: Right
now running everything in JS but would love to move some parts to the
backend. JS provides flexibility.- Yashar: Is there a IDL proposal?- Ana:
We have been working on one but have not make it public yet.- David: The
idea is to first create a repo and then start making proposals. Thought
that everyone would like to use a JS based API but then got tangled in the
Web Driver conversations.- Raphael: Even with Web Driver a fully JS
approach could be used.- David: The main point is to decide how the
communication will be, just JS or Web Driver based.- Yashar: Try to
coordinate as much as possible not to work on different approaches at the
same time.- Ana: Ideally creating a repo would the best option. We need
consensus on the approach.- David: If people do not object to having a JS
facade in front of Web Driver, then we could move to creating a repo.-
Yashar: MSFT is ok with creating a repo with that approach.- Trevor: No one
at Mozilla seems to have a problem with that approach too.- Brandon: Talk
about the input proposal. MSFT wanted to verify that there are not
incompatibilities with their implementation.- Yashar: The engineer (Luis)
 that was working on validating this has been sick for several weeks so
there has not been much more progress. If this is blocking something would
be important to know.- Brandon: On the Chrome side there is an
implementation working well. Haven’t heard much issues on the approach
(IDL) but regarding the implementation could be different in other
systems/vendors. Would love to know if the group is ok moving the input
propoal to be part of the WebXR explainer. But if there are any concerns,
it would be good to know asap.- Yashar: Reviewing the history, the proposal
was made and Nell proposed to do some testing/implementation to validate.-
Brandon: That is correct. No blockers per se but wanted to validate with an
implementation.- Yashar: When Luis comes back could continue to work on
this.- Brandon: Minor changes is ok, but if the core of the proposal does
not fit, this is the most important aspect to know as soon as possible.-
Trevor: We are fine with moving with the explainer as it is. There might be
some nuances down the line but for now is ok.- Artem: What devices have you
tested with?- Brandon: With daydream and the oculus controllers. Landed in
Canary already. Very happy how it works with both simpler and more complex
controllers.- Artem: What 3D models are being used?- Brandon: Using some
models in the examples. Of course there is a mismatch but it works pretty
well.- Artem: Will the gamepad API still be operative?- Brandon: For now it
will remain but with the idea to turn it off in the future once it is
studied the need for developers. I do not see how the OpenXR input model
can be mapped to the gamepad API. Happy to hear is someone has some
suggestion on how to support it.- Artem: My idea is to use the gamepad API
to identify the controllers so the correct model can be shown.- Brandon:
There are some pros and cons on that approach. Would love to see Apple to
comment on this and the finger printing implications.- Yashar: Would love
to get the implementation done to move the proposal to the spec explainer
if possible if there is no rush.- Brandon: There is no rush but wanted to
ping about it and make sure there is a plan and where things at.- Brandon:
The other topic is the discussion about the session creation. Pity that
Blair is not in the call. A recap on the discussion:- Some concerns about
the simplicity of the current session creation.- How do we enable people to
decide whether or not to support WebXR support in the case there is a
device that does not support all their needs.- Extend support session by
having a promise returning a minimum feature set. - Brandon: How much
should be exposed? Should we allow developers to be very granular? Also
there are many situation where the information could not be accurate like
with controllers in stand by.- Raphael: Have been considerations about
fingerprinting? We think they should be taken into account and run by the
right people.- Brandon: I agree that these considerations should be
completely be taken into consideration. There could be some possibilities
to trigger some permission requests even in the support queries. In my mind
the optimal point to launch permission requests should be once the user
presses a button to do something but there could be some restrictions in
the system that forbid this.- Rephael: Blair and Brandon made very good
points on things I did not think about. Would it be good to compile a list
and go one by one?- Brandon: We have been leaving it pretty high level and
it would be good to work on something more specific. There can be some
limitations that even technically possible are not very likely to happen.
We might need to take these into consideration.- Artem: Will there be a
reason on why the session was rejected?- Brandon: That’s a really good
point. The idea would be not to be very strict in the request but to
provide more information once the session fails or even progresses.*



On Tue, Mar 20, 2018 at 8:59 AM, Brandon Jones <bajones@google.com> wrote:

> Very sorry for the delay getting this email out. My schedule was thrown
> off by GDC.
>
> *Call Agenda Items:*
>
>    - Revisit session creation options and flow (#330)
>    <https://github.com/immersive-web/webxr/issues/330>
>    - Input Proposal pull request (#325
>    <https://github.com/immersive-web/webxr/pull/325>)
>
> Please reply with any additional items you'd like to see addressed.
>
> *Call date:* Tuesday March 20th (and every other Tuesday thereafter)
> *Call time:* 1:00 PM PST for one hour
>
> WebEx call details are posted to the internal-webvr@w3.org mailing list,
> accessible to any community group member.
>