FYI, the TAG finished our review with https://github.com/w3ctag/design-reviews/issues/1038#issuecomment-2853142041. We were generally happy with the design decisions that Evan and the WG have made, but we were still concerned that "ondevice-only" excludes some choices that future UAs might reasonably want to explore. We listed 5 kinds of locations that a user might want to run speech recognition (or heavy workloads in general), and we thought the WG should look at the concrete websites that want to adopt this, figure out which locations they're ok with, and pick a name based on that. We didn't think Google Meet's described use case for "ondevice-only" <https://github.com/w3ctag/design-reviews/issues/1038#issuecomment-2837046998> was even about recognition location, but it might also indicate a feature the WG might want to add.
Jeffrey On Fri, Apr 18, 2025 at 11:31 AM Evan Liu <ev...@google.com> wrote: > Hi all, > > We discussed the TAG feedback at the Audio Working Group meeting yesterday > and I've posted our response here: > https://github.com/w3ctag/design-reviews/issues/1038#issuecomment-2815982645 > > Please let me know if anyone has any questions/comments/concerns. > > I don't think there's any particular reason to unprefix before shipping >> on-device, is there? > > Also to answer your question, Rick, I don't think there's any reason to > unprefix before shipping on-device, so we might as well lump it together as > a bundle :). > > Thanks! > Evan > > On Wed, Apr 16, 2025 at 10:54 AM Brian Kardell <bkard...@gmail.com> wrote: > >> Just linking this up as I see there are some questions, but the opening >> post seems to suggest there are positive signals from WebKit... >> >> https://github.com/WebKit/standards-positions/issues/443 >> >> On Wednesday, April 16, 2025 at 10:55:52 AM UTC-4 Rick Byers wrote: >> >>> On Tue, Apr 15, 2025 at 8:14 PM Evan Liu <ev...@google.com> wrote: >>> >>>> Thanks for the detailed feedback, Jeffrey! We'll discuss this at the >>>> Audio Working Group meeting this week and I'll update this thread >>>> afterwards. >>>> >>>> Thanks, >>>> Evan >>>> >>>> On Mon, Apr 14, 2025 at 9:08 PM Jeffrey Yasskin <jyas...@chromium.org> >>>> wrote: >>>> >>>>> FYI, the TAG left comments at >>>>> https://github.com/w3ctag/design-reviews/issues/1038#issuecomment-2803693504 >>>>> . >>>>> >>>>> On Fri, Apr 4, 2025 at 10:22 AM Evan Liu <ev...@google.com> wrote: >>>>> >>>>>> Are you thinking it might be reasonable to ship in M128 (decide by >>>>>>> branch on Apr 28, plan to merge any required changes before May 21)? >>>>>> >>>>>> That sounds like a reasonable target, assuming TAG doesn't propose >>>>>> any significant changes. >>>>>> >>>>>> That said, if you want to, I'm supportive of shipping the unprefixing >>>>>>> alone >>>>>>> <https://chromium-review.googlesource.com/c/chromium/src/+/6422562> now, >>>>>>> since you already proved to us that the unprefixed API is not an >>>>>>> opportunity to make any breaking API changes. Do you prefer to decouple >>>>>>> that, or just wait and get the whole bundle approved to ship together? >>>>>>> >>>>>> Either is fine with me! Would decoupling just be a matter of making >>>>>> the changes, or would I need to create a separate Chrome Status entry, >>>>>> get >>>>>> position statements, all of the approvals, etc.? If it's the former, we >>>>>> might as well make the change now. Otherwise it might just be easier to >>>>>> bundle everything together. >>>>>> >>>>> >>> I'm OK with just shipping the unprefixing under this same intent without >>> the extra paperwork, but also it's a bit simpler if we just keep it all >>> lumped together as a bundle. I don't think there's any particular reason to >>> unprefix before shipping on-device, is there? >>> >>> >>>> >>>>>> Thanks, >>>>>> Evan >>>>>> >>>>>> On Fri, Apr 4, 2025 at 6:51 AM Thomas Steiner <to...@google.com> >>>>>> wrote: >>>>>> >>>>>>> This all looks great to me! Are you thinking it might be reasonable >>>>>>>> to ship in M128 (decide by branch on Apr 28, plan to merge any required >>>>>>>> changes before May 21)? >>>>>>>> >>>>>>> >>>>>>> Off by one, classic. I think you meant 1*3*8 here. I know it's >>>>>>> obvious now, but someone might once look back at this in ten years from >>>>>>> now >>>>>>> and wonder… >>>>>>> >>>>>> >>> Whoops, yes of course - thank you :-). >>> >>> >>>> >>>>>>> >>>>>> -- You received this message because you are subscribed to the Google Groups "blink-dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+unsubscr...@chromium.org. To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CANh-dXkPXU2VHOPVOMotGkSPUePf8RHjRE58z%3DOr8y9dHHCTEw%40mail.gmail.com.
