Meeting minutes
A couple of FYIs
Mounir: Happy New Year everyone! Agenda is a bit tiny. Starting with FYIs. One is around Media Capabilities and feedback around Security and Privacy
… possible fingerprinting increase.
… WebCodecs has the same potential issue in principle.
Tess: I'm one of the editors of the questionnaire from the TAG side (joint work with PING)
… The ED has numerous changes over the last few years. Even though it hasn't been published for a while, I would encourage the use of the ED.
Mounir: Right, I would like the editors of the spec to update the existing responses that we had with changes brought to the specs.
… Anyone has a question/comment about that?
… Chris, since you're editing both specs, any comment?
Chris: The context for Media Capabilities is that we're adding encoding/decoding info for WebRTC. We have had for a long time a transmission type
… This was sort of half-baked. We need to fix that.
… As part of fixing that, the WebRTC WG wondered whether we had a privacy review and we do not.
… The fingerprinting surface is not significantly different from what it was in the past.
… Similar story for WebCodecs. The difference is that you can be very granular about e.g. encoding options and hardware capabilities. There is more information here, creating a bit more entropy.
Mounir: OK, moving on to second FYI about adopting WebCodecs as a Media WG deliverable. We quickly discussed that during our last call. We're checking with editors first. Chris gave a thumbs up, we're still waiting for Paul, before we issue a call for consensus to adopt the spec in the group.
… Any comments before we send that CfC?
… Otherwise, that should develop in a week or two.
Should PiP video removed from the DOM leave PiP?
Mounir: Raised by Francois (Beaufort) who is not here today. I talked with him earlier today and his question was whether there was a specific use case for not pausing when you are a picture-in-picture.
Jer: There have been a number of situations where weird playback behavior are triggering issue. I'm not entirely sure that there is a specific problem there. That said, we have seen Web sites trying to use that as a way to prevent picture-in-picture.
Mounir: [missed point]
… It creates a lot of hassle if DOM and PiP double depend on each other.
Jer: The bug I filed against HTML proposes to expose a "is the video element visible to the user" property that can be toggled by other APIs, including PiP in our case.
… That would not create a double dependency.
… I believe that this section of the spec was written at a time where not being in the DOM meant the video was not visible.
… Let's see what they say.
Mounir: Fair enough.
Remove persistent-usage-record from EME
Greg: Simple enough, Netflix decided not to use that feature after all. We're not aware of anyone else using it for now or needing it, so we propose to drop the feature from the Editor's Draft.
Jer: Safari added support for this a long time ago, due to Netflix. Is that going to be removed from the native side as well as from the Web side.
Greg: The answer is yes and no. It will not be required from now on, but won't disappear from older devices.
… We were using it for concurrent stream checks. How long has the license been used? Somebody could start a session and we wouldn't securely know whether they had been using the license for 5mn or 24h.
… It's a fraud detection mechanism at the account level.
Jer: Short renewable duration license is a way to do that without the feature?
Greg: Yes, that's one solution. We thought about it e.g. for suspicious accounts.
MarkW: We have added various ways to detect frauds over the years. Given all the tools that we now have, we believe that we can make do without this one and save implementation time for browsers.
Jer: OK, I would just like to get more clarity in other forum on the native side before we drop the feature from the spec.
Media Session
Jer: We have had requests for controlling WebRTC phone calls. Wondering about whether pick up / hang up actions have been considered.
Mounir: We received similar requests. The problem is that the list grows very fast. I don't think that people in Google would be opposed to a hang up action if needed.
Jer: Also thinking about PiP external controls, such as touchbar, that don't really fit anywhere. We would likely be interested in exploring this in the future.
<Zakim> cyril, you wanted to comment on a different topic
ISOBMFF Byte Stream format for MSE
Cyril: I've started to make modifications to the spec based on my understanding of the spec and on what browsers implement
… When you start reading the ISOBMFF spec, I think that there are things that are wrong, e.g. in section 3.
… Example of assertion that requires browsers to maintain requirements per brand.
… In ISOBMFF, you're supposed to process files if there's a brand you support, this also goes counter to the current statements.
… Very surprising to see that all browsers ignore the "ftype" box, regardless of what you put in it.
<cyril> https://
ChrisC: Matt is not on the call, but I'm sure he would be interested as well.
<Zakim> tidoust, you wanted to note that byte stream format specs have moved to specific repos
https://
Francois: Just note that byte stream format specs have moved to dedicated repos.
Cyril: Noted. Back to the spec, any pushback on the updated?
[None heard]
Misc.
<markw> me Sorry, gtg now
Mounir: On other news, Becca is going to stop editing specs in the group. That affects Media Session. So we'll need to find another editor. Maybe from Apple?
Jer: I'll look into it.
Mounir: On behalf of the Media WG, thank you Becca for the work you've done on the spec!