andreastt

I believe we have all the information we need here. I’ve filed https://bugzilla.mozilla.org/show_bug.cgi?id=1560113 to track this problem specifically, although the fix is the same as for https://bugzilla.mozilla.org/show_bug.cgi?id=1434313.

I’ve asked over on https://github.com/SeleniumHQ/selenium/issues/7259 if it’s possible to introduce an intermediary API between the .NET clients’ service classes and what the users interact with. The difference with the Python...

Can you supply a trace-level log, please?

@whimboo Can you explain the issue, then?

@vavdagic Thanks for explaining. There’s noting in [the specification] that says the return value from Get Element Property should be serialised, i.e. that elements should be serialised as WebDriver web...

@vavdagic If it is added to the specification, sure. I’m on the fence about complicating the Get Element Property command because I feel it is well complemented by Execute Script,...

I can’t reproduce your test case. In the following WPT test I set a property, `$$prop`, after the document has loaded. Subsequently its value is checked to ensure it is...

Yes, we need a _trace-level_ log. See your client’s documentation or the README on how to crank up the verbosity. The tl;dr is that you need to pass `{"moz:firefoxOptions": {"log":...

Here you are using the Get Element Property command, and not Execute Script which I assumed this was about: ``` 1505132113999 webdriver::server DEBUG -> GET /session/76ac4f07-47f0-43e9-bf95-ceea89ef20c9/element/af7c003b-9e6f-41e8-9adc-ce2c469b94fe/property/$$hashKey ``` The basic problem...

We may be able to use something like the [chan-signal crate](https://github.com/BurntSushi/chan-signal) to receive the handleable signals and call `drop(temp_dir)` explicitly. We might have to implement the `Drop` trait for rust_mozprofile....