Bugzilla

The message here comes from this marionette log:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/tools/wptrunner/wptrunner/executors/executormarionette.py#894-896
```py
except Exception as e:
    if isinstance(e, errors.JavascriptException) and str(e).startswith("Document was unloaded"):
        message = "Document unloaded; maybe test navigated the top-level-browsing context?"
```

I think that's exactly what the test does -- it sets `window.location.href` here:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/print/crashtests/reload-crash.html#15-16
```html
var updatedUrl = url + '?isFirstLoad=false';
window.location.href = updatedUrl;
```
...and [per MDN](https://developer.mozilla.org/en-US/docs/Web/API/Location/href), "Setting the value of href **navigates to the provided URL**".

jgraham, is the exception indicating that this sort of navigation is disallowed for WPT tests? (If so, perhaps we should file a Chromium bug to get the test fixed; and in the meantime, we can annotate it as skipped-on-Android since that's where this intermittent is happening. I wonder why this is android-specific; maybe the test just gets treated as complete before the setTimeout-based navigation happens on other platforms, since the test doesn't use `test-wait`.)

The test-failure-text here comes from this marionette code:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/tools/wptrunner/wptrunner/executors/executormarionette.py#894-896
```py
except Exception as e:
    if isinstance(e, errors.JavascriptException) and str(e).startswith("Document was unloaded"):
        message = "Document unloaded; maybe test navigated the top-level-browsing context?"
```

I think that's exactly what the test does -- it sets `window.location.href` here:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/print/crashtests/reload-crash.html#15-16
```html
var updatedUrl = url + '?isFirstLoad=false';
window.location.href = updatedUrl;
```
...and [per MDN](https://developer.mozilla.org/en-US/docs/Web/API/Location/href), "Setting the value of href **navigates to the provided URL**".

jgraham, is the exception indicating that this sort of navigation is disallowed for WPT tests? (If so, perhaps we should file a Chromium bug to get the test fixed; and in the meantime, we can annotate it as skipped-on-Android since that's where this intermittent is happening. I wonder why this is android-specific; maybe the test just gets treated as complete before the setTimeout-based navigation happens on other platforms, since the test doesn't use `test-wait`.)

The test-failure-text here comes from this marionette code:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/tools/wptrunner/wptrunner/executors/executormarionette.py#894-896
```py
except Exception as e:
    if isinstance(e, errors.JavascriptException) and str(e).startswith("Document was unloaded"):
        message = "Document unloaded; maybe test navigated the top-level-browsing context?"
```

RE the "maybe" situation here -- in fact I think that's exactly what the test does -- it sets `window.location.href` here:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/print/crashtests/reload-crash.html#15-16
```html
var updatedUrl = url + '?isFirstLoad=false';
window.location.href = updatedUrl;
```
...and [per MDN](https://developer.mozilla.org/en-US/docs/Web/API/Location/href), "Setting the value of href **navigates to the provided URL**".

jgraham, is the exception indicating that this sort of navigation is disallowed for WPT tests? (If so, perhaps we should file a Chromium bug to get the test fixed; and in the meantime, we can annotate it as skipped-on-Android since that's where this intermittent is happening. I wonder why this is android-specific; maybe the test just gets treated as complete before the setTimeout-based navigation happens on other platforms, since the test doesn't use `test-wait`.)

The test-failure-text here comes from this marionette code:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/tools/wptrunner/wptrunner/executors/executormarionette.py#894-896
```py
except Exception as e:
    if isinstance(e, errors.JavascriptException) and str(e).startswith("Document was unloaded"):
        message = "Document unloaded; maybe test navigated the top-level-browsing context?"
```

RE the "maybe test navigation" situation that marionette is alluding to here: in fact I think that's exactly what the test does. The test sets `window.location.href` here:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/print/crashtests/reload-crash.html#15-16
```html
var updatedUrl = url + '?isFirstLoad=false';
window.location.href = updatedUrl;
```
...and [per MDN](https://developer.mozilla.org/en-US/docs/Web/API/Location/href), "Setting the value of href **navigates to the provided URL**".

jgraham, is the exception indicating that this sort of navigation is disallowed for WPT tests? (If so, perhaps we should file a Chromium bug to get the test fixed; and in the meantime, we can annotate it as skipped-on-Android since that's where this intermittent is happening. I wonder why this is android-specific; maybe the test just gets treated as complete before the setTimeout-based navigation happens on other platforms, since the test doesn't use `test-wait`.)

The test-failure-text here comes from this marionette code:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/tools/wptrunner/wptrunner/executors/executormarionette.py#894-896
```py
except Exception as e:
    if isinstance(e, errors.JavascriptException) and str(e).startswith("Document was unloaded"):
        message = "Document unloaded; maybe test navigated the top-level-browsing context?"
```

RE the "maybe test navigated" situation that marionette is alluding to here: in fact I think that's exactly what the test does. The test sets `window.location.href` here:
https://searchfox.org/mozilla-central/rev/41edcdf7fe44678c5913a603a286b1fc3979d540/testing/web-platform/tests/print/crashtests/reload-crash.html#15-16
```html
var updatedUrl = url + '?isFirstLoad=false';
window.location.href = updatedUrl;
```
...and [per MDN](https://developer.mozilla.org/en-US/docs/Web/API/Location/href), "Setting the value of href **navigates to the provided URL**".

jgraham, is the exception indicating that this sort of navigation is disallowed for WPT tests? (If so, perhaps we should file a Chromium bug to get the test fixed; and in the meantime, we can annotate it as skipped-on-Android since that's where this intermittent is happening. I wonder why this is android-specific; maybe the test just gets treated as complete before the setTimeout-based navigation happens on other platforms, since the test doesn't use `test-wait`.)