Re: [html-tests] HTML: test coverage 7.1.5: `_top`, `_self`, existing contexts (#5233)

@zcorpan 

> I can't reproduce this failure in Firefox (for me it passes):

> /html/browsers/windows/browsing-context-names/choose-default-001.html

> Do you have a clue?

I have a clue...ish. This is one of the tests that I inherited, and I had observed the behavior you're seeing, too. For posterity, here's the test, which is _super_ basic:

```
//This test must be run when the current browsing context's name is not set
test(t => {
  assert_equals(window.name, "");
}, "A browsing context has an empty-string default name");
```

(Comments pre-exist my changes, also).

This test will pass when run directly and individually in a browser, but will fail when the test is run in a window opened (and ostensibly named) by a test runner. e.g. if I run the whole suite of tests in `browsing-context-names` in a browser (any browser, in my experience, not just FF):

![image](https://cloud.githubusercontent.com/assets/439947/24457921/88947d76-1465-11e7-80ed-0bdd75d28b02.png)

it will fail.

I'm not sure of the best course of action here. I don't think the test adds a whole lot of value, and the default-window-name-is-empty-string notion is tested elsewhere (e.g. line 13 of the same test file). So part of me wants to yank it altogether.

Another part of me thinks that removing it is just patching over the fact that I'm not sure how to get it to work correctly in the test harnesses we've got, which is not ideal :).

So I'm not sure! Thoughts?

View on GitHub: https://github.com/w3c/web-platform-tests/pull/5233#issuecomment-290097472

Received on Wednesday, 29 March 2017 13:55:55 UTC