Issue 38 - Cursor constraint

[suhasHere]

Following up on the discussions in the issue, this PR adds cursor constraint that enumerates the conditions for capturing the cursor.

[martinthomson]

Probably better to change the authors in a separate PR, but this looks fine.

Do you think that there is any point in mentioning that the browser needs to determine how long to render the cursor for after movement stops?

[martinthomson]

Missing "the"

[suhasHere]

thanks @martinthomson for the review.

Sure adding a statement to remove the cursor when there is no movement might be handy. Would something like this work ?

"The captured cursor is removed when there is no further movement of the pointer for certain period of time (as determined by the user agent) to ensure smoother user experience."

[martinthomson]

Yeah, that works. I would drop the tail: "The captured cursor is removed when there is no further movement of the pointer for certain period of time, as determined by the <a>user agent</a>."

[juberti]

a -> an

[suhasHere]

@juberti my bad .. I will get this fixed.

[juberti]

This sentence seems wrong - we do want the app to change the constraints, not just observe them.

[martinthomson]

Good catch.

@suhasHere, we need to reframe this section. Rather than do the obvious, and split the section on constraints into two, we should make it clear that though constraints don't affect selection (as much as people might like them to), they still affect the shape of the stream that is produced.

[suhasHere]

@juberti @martinthomson I was planning to do constraints related work in a separate PR that also deals with bringing back few constraints as decided in the last meeting. Would it be ok, if I address the general constraints ideas in there ?

[juberti]

Think this one still needs a bit more work.

[guest271314]

@juberti FYI Tried .getDisplayMedia({video: {cursor: "never"}}), videoTrack.applyConstraints({cursor: "never"}) and videoTrack.cursor = "never" at Chromium 72. Neither removed the cursor entirely from the captured media stream

After setting the constraints videoTrack.getSettings()

https://developer.mozilla.org/en-US/docs/Web/API/MediaStreamTrack/getConstraints#Return_value

To get the currently active settings for all constrainable properties, you should instead call getSettings().

the resulting object still had cursor set to "always"

{aspectRatio: <value>
cursor: "always" // after applying constraints to the (video) `MediaStreamTrack`
deviceId: <value>
displaySurface: <value>
frameRate: <value>
height: <value>
logicalSurface: <value>
resizeMode: <value>
videoKind: <value>
width: <value>}

Outside of using .pointerLock() CSS element {cursor:none} was the code that resulted in the cursor not being rendered as part of the media stream.

The screen-capture directory at wpt there does not appear to include a test for cursor.

[guest271314]

https://bugs.chromium.org/p/chromium/issues/detail?id=959651

[CapOM]

Hi, can you confirm that when a capture is started through .getDisplayMedia({video: {cursor: "always"}}) then later on a call to videoTrack.applyConstraints({cursor: "never"}) should make the cursor disappear on the fly, without the need to re-select the source ? I.e. further calls to never and always on the track must be supported to allow developers to show and hide cursor when needed without requiring to restart the capture pipeline.
Refs: https://w3c.github.io/mediacapture-main/#dfn-applyconstraints-algorithm and https://w3c.github.io/mediacapture-main/#dfn-selectsettings .
It looks to me that the spec allows that but it is up to the implementation to decide whether to succeed or fail.
I wonder if it would be possible to make it a requirement or at least have a way to query if the implementation supports that.
Thx!