Depends on https://github.com/lightpanda-io/libdom/pull/36
The spec says this should be a High Definition timestamp. But browsers avoid
that in order to avoid fingerprinting. By default, FireFox rounds to 2ms (which
is what this PR does).
Previously, the timestamp was seconds, so you'd think: isn't that better? Well,
it's pretty far off the spec and what browsers do, but more importantly, it
crashes our WPT test. If you look at `Event-timestamp-safe-resolution.html`
you'll see that it's trying to find the delta between two timestamps, in an
endless loop (without a loop of many iterations). With second-resolution, it
just takes too long (and crashes..memory).
- Continue to reuse the Browser/Env/Isolate, but no start a new session per test.
- Test http server now properly closes the sendFile fd
- Run WPT in ReleaseMode
- Add --quiet option to WPT and some commented out debug code for dumping v8
memory stats
Like Chrome, the NetworkIdle and NetworkAlmostIdle will only be sent if the
condition (no network requests / <= 2 network requests) holds for at least 500ms
Also merged runHighPriority and runLowPriority as they are now always run
together (but we still only block/wait for high priority tasks).
We previously ignored tasks scheduled more than 5 seconds away. These tasks are
now scheduled on the low priority queue. This means that they won't stop a
page.wait for returning, but they'll still [eventually] be run if page.wait is
called multiple times.
Practically, this means that they'll never be run in `fetch` mode, but they
might be run from CDP if the driver waits.
Make queue names consistent, primary => high_priority, secondary => low_priority
(the same names used by the page)
According to MDN, inline script tags should not have defer/async attributes. But
some do. This ignores those attributes for inline script tags.
(Previously, we were only half ignoring them. We were treating them as inline,
but flagging them as deferred or async, which was causing issues with cleanup)
Fixes: https://github.com/lightpanda-io/browser/issues/1014
