SourceBuffer(List) / MediaSource: update event entries

[Elchi3]

Update for new event structure.

Content update: mdn/content#11176

[Elchi3]

Content PR merged. This is ready.

[foolip]

Chrome 53 is when onabort and lots of other event handler properties were added:
https://storage.googleapis.com/chromium-find-releases-static/000.html#00003ed19f12e054784ee59b95c863b380361ef4

I would suggest also including MediaSource in this PR, to sort it all out at once.

Unfortunately, we'll have to find the versions when each event was supported.

It also looks like it's the same situation for Edge. Firefox and Safari match the parent feature version though, so those don't need any research.

[foolip]

@Elchi3 it will be a lot of work to sort this out, and really any time that the event's version doesn't match the parent feature I'd be suspicious and want to check if it's yet another case of the event handler property being missing originally.

[foolip]

@Elchi3 so my question is, how do you want to handle this? Doing the research as part of PR review is going to slow things down, but can you maintain a list of likely incorrect data in an issue for later review?

[Elchi3]

I've added MediaSource, so it is all together in this PR now.

For all events, the question is:

Is it Chrome 53 or Chrome 31 (Chrome 25 with webkit)?
Is it Edge 17 or Edge 12?

SourceBuffer.abort_event
SourceBuffer.error_event
SourceBuffer.update_event
SourceBuffer.updateend_event
SourceBuffer.updatestart_event
SourceBufferList.addsourcebuffer_event
SourceBufferList.removesourcebuffer_event
MediaSource.sourceclose_event
MediaSource.sourceend_event
MediaSource.sourceopen_event

It currently states the later versions.

[foolip]

@Elchi3 the event handler properties were added in Chromium 53:
https://storage.googleapis.com/chromium-find-releases-static/000.html#00003ed19f12e054784ee59b95c863b380361ef4

So we'll need partial_implementation ranges for the events before that.

The events themselves look complicated. Some of the events themselves were originally prefixed, with https://trac.webkit.org/changeset/123522/webkit being an example of this. (But note how no code was added to actually fire "webkitaddsourcebuffer" or "webkitremovesourcebuffer" there.)

@Elchi3 would you like to capture the prefixed events so that we have data going back to when the interfaces were added? This will require digging in WebKit source for each event. I can show you how I do it if you'd like.

[Elchi3]

I'm happy to add ranges like these:

"chrome": [
  {
    "version_added": "53"
  },
  {
    "version_added": "$versionnumber",
    "partial_implementation": true,
    "notes": "The <code>onsourceclose</code> event handler property is not supported."
 }
],

I have no idea how to find $versionnumber. Pointers appreciated.

The events themselves look complicated. Some of the events themselves were originally prefixed,

I don't know how to find the complete history here and I'm not sure it is worth to go back in time so much?

[foolip]

OK, so let's ignore the prefixed events and try to get reasonable versions for when the unprefixed events were supported.

[foolip]

Found some more things wrong when trying to do "final" review.

Note that the IE data is also arguably wrong, I'm pretty sure it was also missing the event handler properties. I won't insist on adding that though.

[foolip]

I started to review again and found more things wrong, but at this point I'm just going to look away :)

[Elchi3]

Ha, thanks for merging! This has been quite an advantage in compat archeology.

[foolip]

For future archeologists: we still have a mix of 41 and 42 versions for Firefox Android, which is almost certainty wrong.