Deal with not fully active documents

[marcoscaceres]

See guide - example using Geolocation, which is not covered in our spec I think.

Cc @rakina for review.

[rakina]

Thanks! I think this is already somewhat implicitly specified by step 4 of "request position", which waits for the document to become visible, and bfcached/non-fully active documents should not be visible. But I think it is better to make it more explicit (wait for the document to become fully active also).

also cc @fergal @domenic

[marcoscaceres]

Ok, so, it might be worth checking if the document is still active as part of acquire position.

I'll verify what browsers do there... if they silently ignore the request or error.

[marcoscaceres]

Did a little digging... both Chromium and WebKit simply do the following check:

if (!frame())
   return 0;

But I couldn't find any other checks for when .watchPosition() is actually running.

It's hard to test if the callbacks are invoked, as even .watchPosition() only fires once.... I'll see if I can try on my phone and move around a bit.

Gecko unfortunately throws. I can probably fix that in Gecko easily enough though.

[marcoscaceres]

@rakina ok, below is a simple test case... I'll covert it to a WPT, but would like your feedback before I do.

The neat thing in all browsers is that removal and reattachment does indeed kill the geolocation service for the frame.

  async function runTest() {
    const iframe = document.createElement("iframe");
    iframe.src = "empty.html";
    document.body.appendChild(iframe);
    console.log("loading iframe");
    await new Promise((r) => {
      iframe.onload = r;
    });

    // steal geolocation
    const geo = win.navigator.geolocation;

    // queue up some position watchers...
    console.log(geo.watchPosition(console.log, console.error));
    console.log(geo.watchPosition(console.log, console.error));
    console.log(geo.watchPosition(console.log, console.error));
    geo.getCurrentPosition(console.log, console.error)
    geo.getCurrentPosition(console.log, console.error)
    geo.getCurrentPosition(console.log, console.error)

    // make doc no longer fully active
    iframe.remove();

    // try to use geolocation
    try {
      console.log("watchPosition", geo.watchPosition(console.log, console.error));
    } catch (err) {
      console.log(err);
    }

    try {
      console.log(
        "getCurrentPosition",
        geo.getCurrentPosition(console.log, console.error)
      );
    } catch (err) {
      console.log(err);
    }
    // re-attach
    console.log("reattach");
    document.body.appendChild(iframe);

    // And we are back! But dead... 
    iframe.contentWindow.navigator.geolocation.getCurrentPosition(
      console.log,
      console.error
    );
    iframe.contentWindow.navigator.geolocation.watchPosition(
      console.log,
      console.error
    );

    console.log("done");
  }

  runTest();

[marcoscaceres]

Filed Gecko bug https://bugzilla.mozilla.org/show_bug.cgi?id=1715482

[marcoscaceres]

Sent Gecko patch: https://phabricator.services.mozilla.com/D117273

[marcoscaceres]

Test file is here: https://marcoscaceres.github.io/playground/geo_test.html

[rakina]

Thanks a lot @marcoscaceres! I think the test looks good, neat that you can test it with just iframes. CC-ing @hiroshige-g who is working on BFCache WPTs so that he can make a similar test but with BFCache involved (he's also currently working on a guide to write BFCache tests) and @fergald who is involved in BFCache WPT stuff as well.