Intent to Remove: Prefixed Media Source API

76 views
Skip to first unread message

Aaron Colwell

unread,
Apr 14, 2014, 12:37:02 PM4/14/14
to blink-dev

Primary eng (and PM) emails

acol...@chromium.org


Link to “Intent to Deprecate” thread


https://groups.google.com/a/chromium.org/forum/#!searchin/blink-dev/Prefixed$20Media$20Source/blink-dev/6CEtU6nhcS4/2fzhpRi8d98J


Summary

Remove the prefixed MediaSource API (aka WebkitMediaSource) and related objects.


Motivation

An unprefixed version of the API has been shipping for some time and is a superset of the functionality in the prefixed API. Removing the prefixed API will also allow us to cleanup duplicate code and greatly simplify the Chromium code for the unprefixed API.


Usage information from UseCounter

How much of the web are you going to break?

Barely anything. The usage is so low it shows up as 0% on metrics dashboard!

http://www.chromestatus.com/metrics/feature/timeline/popularity/242


Looking at internal numbers show miniscule usage but that is likely from folks basing their code on old examples published on the Internet.


Entry on chromestatus.com

http://www.chromestatus.com/features/4563797888991232


Compatibility Risk

Very low risk since the unprefixed API has been on by default for almost 7 months and provides equivalent functionality to the prefixed implementation. The deprecation messages have been displayed for at least 3 months now and given people plenty of time to migrate.


Aaron


Eric Seidel

unread,
Apr 14, 2014, 12:44:28 PM4/14/14
to Aaron Colwell, blink-dev
Assuming your UseCounter data is from the Stable channel (M34), then LGTM.

Aaron Colwell

unread,
Apr 14, 2014, 1:11:38 PM4/14/14
to Eric Seidel, blink-dev
Yes. Looking at internal M34 data we are talking about usage in the 1/10000th of a percent range here.

Adam Barth

unread,
Apr 14, 2014, 5:24:03 PM4/14/14
to Aaron Colwell, Eric Seidel, blink-dev
LGTM

Ojan Vafai

unread,
Apr 14, 2014, 7:00:53 PM4/14/14
to Adam Barth, Aaron Colwell, Eric Seidel, blink-dev
LGTM3

PhistucK

unread,
Apr 14, 2014, 7:03:12 PM4/14/14
to Adam Barth, Aaron Colwell, Eric Seidel, blink-dev
What about the Android specific usage?


PhistucK


LGTM
To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.

Aaron Colwell

unread,
Apr 14, 2014, 7:22:45 PM4/14/14
to PhistucK, Adam Barth, Eric Seidel, blink-dev
For some reason there only appears to be 1 data point for Android and it indicates zero usage. Unfortunately more Android specific usage data does not appear to be available, but it would likely be WAY smaller than the desktop numbers because MSE has only been available on Android for a few versions of Chrome and only works on Jelly Bean or higher. The 2 main consumers of the MediaSource API, YouTube and Netflix, have Android apps which also significantly deminishes the likelihood of any major usage on that platform.

Aaron

Ojan Vafai

unread,
Apr 14, 2014, 7:46:42 PM4/14/14
to Aaron Colwell, PhistucK, Adam Barth, Eric Seidel, blink-dev
Not sure what exactly you're looking at. When I look at Android stable channel data, I see plenty of data points and zero usage of PrefixedMediaSourceOpen. I see 0.0000075% on Windows stable.

In either case, LGTM still stands. This is many orders of magnitude below what we consider a safe threshold for removing an API.

Aaron Colwell

unread,
Apr 14, 2014, 8:01:59 PM4/14/14
to Ojan Vafai, PhistucK, Adam Barth, Eric Seidel, blink-dev
Perhaps I am doing something wrong. I was looking at the UMA stat timeline for 
Metric: WebCore.FeatureObserver
Bucket: PrefixedMediaSourceOpen
Platform: android-all.

Thanks for the LGTM's. I'm looking forward to the wonderful red-filled patch set being uploaded shortly. :)

Aaron
Reply all
Reply to author
Forward
0 new messages