Re: Intent to ship: MediaRecorder.{audio|video}BitsPerSecond
On Fri, Oct 4, 2019 at 6:54 AM Tom Ritter wrote: > It's a bit hard for me to tell from the description - are these values > dependent on a user's hardware, performance of the user's computer, or > a user-chosen setting? If so we would want to support > resistFingerprinting. > The spec allows the UA to base them on anything it sees fit, if the application leaves it to the UA to decide. Our implementation uses fixed and application-provided values, per below. There are three different modes of configuring these values for an application, and here's how we set them: 1) Application did no configuration at all: we use hardcoded values: 2.5Mbps for video and 128kbps for audio 2) Application configured audio and video respectively: we use the provided values 3) Application configured a total bitrate: Our implementation has a fixed algorithm for this -- its only inputs is the number of tracks and kinds of tracks, and the total bitrate configured is distributed according to a weighting where a video track gets 20 times as high a bitrate as audio, and they're capped within fixed limits. (see https://phabricator.services.mozilla.com/D41585#C1663069NL43 for the limits, and https://phabricator.services.mozilla.com/D41585#C1663069NL212 for the algorithm) Given this, I don't see the need for explicit resistFingerprinting support. -tom > > On Thu, Oct 3, 2019 at 9:54 PM Andreas Pehrson > wrote: > > > > As of Oct 4th I intend to land and ship the audioBitsPerSecond and > > videoBitsPerSecond attributes of MediaRecorder on all platforms. We've > > shipped MediaRecorder since Firefox 29. > > > > *Summary*: The audioBitsPerSecond and videoBitsPerSecond attributes of > > MediaRecorder reflect what an application has configured the > MediaRecorder > > to use (and which we already ship support for doing). In case it was > > configured with a total (audio+video) bitrate, or if it was not > configured > > at all, these attributes reflect what the UA has chosen to configure the > > MediaRecorder with. > > > > After starting a recording, these are again updated to reflect the > current > > configuration of the MediaRecorder, since tracks are now known and the UA > > may have recalculated the chosen bitrates based on the number and kind of > > tracks that are being recorded. > > > > At the same time I'm making our MediaRecorder implementation largely > > spec-compliant. There's work in bug 1514158 as well as bug 1512175, where > > the former is focusing on these attributes and the latter focuses on mime > > type handling. We landed spec clarifications for both of these recently. > > > > *Bug*: https://bugzilla.mozilla.org/show_bug.cgi?id=1514158 > > > > *Standard*: > > > https://w3c.github.io/mediacapture-record/MediaRecorder.html#dom-mediarecorder-videobitspersecond > > Clarifications were made to the spec to make this more interoperable. > These > > were approved by Google. > https://github.com/w3c/mediacapture-record/pull/175 > > > > *Platform coverage*: All > > > > *Preference*: There is no special pref for these attributes > > > > *DevTools bug*: There is no generic devtools support for MediaRecorder > > > > *Other browsers*: > > Chrome shipped (since 49, bug: > > https://bugs.chromium.org/p/chromium/issues/detail?id=262211), > > Safari shipped MediaRecorder but no attributes yet (behind an > experimental > > feature flag since TP73, bug: > https://bugs.webkit.org/show_bug.cgi?id=85851) > > > > *web-platform-tests*: I'm adding one as part of bug 1514158: > > https://phabricator.services.mozilla.com/D41584 > > > > *Secure contexts*: No, MediaRecorder is historically not restricted to > > secure contexts. However some APIs for getting access to the > > MediaStreamTracks needed to create a recorder are. > > > > > > Andreas Pehrson > > ___ > > dev-platform mailing list > > dev-platform@lists.mozilla.org > > https://lists.mozilla.org/listinfo/dev-platform > ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
Re: Intent to ship: MediaRecorder.{audio|video}BitsPerSecond
It's a bit hard for me to tell from the description - are these values dependent on a user's hardware, performance of the user's computer, or a user-chosen setting? If so we would want to support resistFingerprinting. -tom On Thu, Oct 3, 2019 at 9:54 PM Andreas Pehrson wrote: > > As of Oct 4th I intend to land and ship the audioBitsPerSecond and > videoBitsPerSecond attributes of MediaRecorder on all platforms. We've > shipped MediaRecorder since Firefox 29. > > *Summary*: The audioBitsPerSecond and videoBitsPerSecond attributes of > MediaRecorder reflect what an application has configured the MediaRecorder > to use (and which we already ship support for doing). In case it was > configured with a total (audio+video) bitrate, or if it was not configured > at all, these attributes reflect what the UA has chosen to configure the > MediaRecorder with. > > After starting a recording, these are again updated to reflect the current > configuration of the MediaRecorder, since tracks are now known and the UA > may have recalculated the chosen bitrates based on the number and kind of > tracks that are being recorded. > > At the same time I'm making our MediaRecorder implementation largely > spec-compliant. There's work in bug 1514158 as well as bug 1512175, where > the former is focusing on these attributes and the latter focuses on mime > type handling. We landed spec clarifications for both of these recently. > > *Bug*: https://bugzilla.mozilla.org/show_bug.cgi?id=1514158 > > *Standard*: > https://w3c.github.io/mediacapture-record/MediaRecorder.html#dom-mediarecorder-videobitspersecond > Clarifications were made to the spec to make this more interoperable. These > were approved by Google. https://github.com/w3c/mediacapture-record/pull/175 > > *Platform coverage*: All > > *Preference*: There is no special pref for these attributes > > *DevTools bug*: There is no generic devtools support for MediaRecorder > > *Other browsers*: > Chrome shipped (since 49, bug: > https://bugs.chromium.org/p/chromium/issues/detail?id=262211), > Safari shipped MediaRecorder but no attributes yet (behind an experimental > feature flag since TP73, bug: https://bugs.webkit.org/show_bug.cgi?id=85851) > > *web-platform-tests*: I'm adding one as part of bug 1514158: > https://phabricator.services.mozilla.com/D41584 > > *Secure contexts*: No, MediaRecorder is historically not restricted to > secure contexts. However some APIs for getting access to the > MediaStreamTracks needed to create a recorder are. > > > Andreas Pehrson > ___ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-platform ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
Intent to ship: MediaRecorder.{audio|video}BitsPerSecond
As of Oct 4th I intend to land and ship the audioBitsPerSecond and videoBitsPerSecond attributes of MediaRecorder on all platforms. We've shipped MediaRecorder since Firefox 29. *Summary*: The audioBitsPerSecond and videoBitsPerSecond attributes of MediaRecorder reflect what an application has configured the MediaRecorder to use (and which we already ship support for doing). In case it was configured with a total (audio+video) bitrate, or if it was not configured at all, these attributes reflect what the UA has chosen to configure the MediaRecorder with. After starting a recording, these are again updated to reflect the current configuration of the MediaRecorder, since tracks are now known and the UA may have recalculated the chosen bitrates based on the number and kind of tracks that are being recorded. At the same time I'm making our MediaRecorder implementation largely spec-compliant. There's work in bug 1514158 as well as bug 1512175, where the former is focusing on these attributes and the latter focuses on mime type handling. We landed spec clarifications for both of these recently. *Bug*: https://bugzilla.mozilla.org/show_bug.cgi?id=1514158 *Standard*: https://w3c.github.io/mediacapture-record/MediaRecorder.html#dom-mediarecorder-videobitspersecond Clarifications were made to the spec to make this more interoperable. These were approved by Google. https://github.com/w3c/mediacapture-record/pull/175 *Platform coverage*: All *Preference*: There is no special pref for these attributes *DevTools bug*: There is no generic devtools support for MediaRecorder *Other browsers*: Chrome shipped (since 49, bug: https://bugs.chromium.org/p/chromium/issues/detail?id=262211), Safari shipped MediaRecorder but no attributes yet (behind an experimental feature flag since TP73, bug: https://bugs.webkit.org/show_bug.cgi?id=85851) *web-platform-tests*: I'm adding one as part of bug 1514158: https://phabricator.services.mozilla.com/D41584 *Secure contexts*: No, MediaRecorder is historically not restricted to secure contexts. However some APIs for getting access to the MediaStreamTracks needed to create a recorder are. Andreas Pehrson ___ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
