webdriver: Fix outdated timeouts test and remove redundant check#41631
webdriver: Fix outdated timeouts test and remove redundant check#41631yezhizhen merged 9 commits intoservo:mainfrom
Conversation
|
🤖 Opened new upstream WPT pull request (web-platform-tests/wpt#56959) with upstreamable changes. |
|
📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#56959). |
|
📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#56959). |
|
If Chrome and Firefox already behave like this, I think the specification editors typically update the spec instead of the browsers. Do you have a spec issue where browser vendors indicate they agree to update the behaviour? If not, I don't think we should update the test and instead match their behaviour and update the spec. |
No. The test is written 6 years ago and spec got updated. The upstream is being reviewed, and bug is confirmed in bugzilla link above. Such things happen several times before, where the spec is ahead of vendors: e.g. window rect. |
|
I can't find the link to the spec change. The Bugzilla link is for Firefox only. Can you provide the link to the commit/PR that changed the spec? |
|
@TimvdLippe Thanks for taking a look. I found it to be here: w3c/webdriver#1802. The review's been delegated to web-platform-tests/wpt#56959, similar to #39988 before. |
TimvdLippe
left a comment
There was a problem hiding this comment.
Sounds good! I approve this PR, but please don't merge until you have full approval on web-platform-tests/wpt#56959
Definitely! |
Signed-off-by: Euclid Ye <[email protected]>
Signed-off-by: Euclid Ye <[email protected]>
Signed-off-by: Euclid Ye <[email protected]>
Signed-off-by: Euclid Ye <[email protected]>
Signed-off-by: Euclid Ye <[email protected]>
Signed-off-by: Euclid Ye <[email protected]>
Signed-off-by: Euclid Ye <[email protected]>
|
📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#56959). |
| WebDriverError::new(ErrorStatus::InvalidArgument, "Invalid value for {key}") | ||
| })? as u64), | ||
| Value::Number(num) => Some( | ||
| num.as_u64() |
There was a problem hiding this comment.
My bad. It panicked with 2.0, which is valid according to spec.
Signed-off-by: Euclid Ye <[email protected]>
Signed-off-by: Euclid Ye <[email protected]>
|
📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#56959). |
|
🔨 Triggering try run (#20713924804) for Linux (WPT) |
|
Test results for linux-wpt from try job (#20713924804): Flaky unexpected result (1)
|
|
✨ Try run (#20713924804) succeeded. |
Initially the attempt is to validate params according to spec. But it turns out validation is already done at webdriver crate so I reverted the change.
But Chrome/Firefox are not compliant with spec: https://wpt.fyi/results/webdriver/tests/classic/new_session/timeouts.py?label=experimental&label=master&aligned. They only allow
script timeoutsto beNonebut notpageLoadandimplicit, which is how the test is currently written.Testing: Updated the test to be compliant with spec. New failures are because it is validated at webdriver crate.