missed XMLHttpRequest and EventSource tests about wpt

[hallvors]

#242 :)

[Yaffle]

@hallvors, thanks
but, may be i was wrong about this test:
http://www.whatwg.org/specs/web-apps/current-work/multipage/history.html#abort-a-document

Cancel any instances of the fetch algorithm in the context of this Document, discarding any tasks queued for
them, and discarding any further data received from the network for them

but, for "unloading" it is hardly to understand what spec says and
AND for XHR:

If the end user cancels the request
This is an abort error.

in any case "error" event should not be fired at xhr, seems, and Firefox is buggy here

see something similar for WebSockets: https://github.com/w3c/web-platform-tests/blob/master/websockets/unload-a-document/001-1.html

[Yaffle]

readyState before navigation/cancellation:

XMLHttpRequest LOADING
WebSocket (active) OPEN
EventSource (active) OPEN
EventSource (reconnecting) CONNECTING

---

"window.stop" OR page navigation OR navigation to a resources with Content-Disposition: attachment:

|                            | InternetExplorer 11 | FireFox 28           | Chrome 33          | Opera 12.16 |
|----------------------------|---------------------|----------------------|--------------------|-------------|
|XMLHttpRequest (active)     |                     |                      |                    |             |
|                events      | -                   | "error", "loadend"   | "abort", "loadend" | -           |
|                readyState  | LOADING             | DONE                 | DONE               | LOADING     |
|                active      | yes                 | no                   | no                 | yes         |
|----------------------------|---------------------|----------------------|--------------------|-------------|
|WebSocket (active)          |                     |                      |                    |             |
|                events      | N/A                 | "close" (code: 1001) | -                  | -           |
|                readyState  | OPEN                | CLOSED               | OPEN               | OPEN        |
|                active      | yes                 | no                   | yes                | yes         |
|----------------------------|---------------------|----------------------|--------------------|-------------|
|EventSource (active)        |                     |                      |                    |             |
|                events      | N/A                 | "error"              | -                  | -           |
|                readyState  | N/A                 | CLOSED               | CLOSED             | OPEN        |
|                active      | N/A                 | no                   | no                 | yes         |
|----------------------------|---------------------|----------------------|--------------------|-------------|
|EventSource (reconnecting)  |                     |                      |                    |             |
|                events      | N/A                 | -                    | -                  | -           |
|                readyState  | N/A                 | CONNECTING           | CONNECTING         | CONNECTING  |
|                active      | N/A                 | yes                  | yes                | yes         |

What about the rule for window.stop: if the fetching is cancelled, the browser must fire an event and change the readyState. ?

Firefox issues:

1. "error" event type instead of "abort" for XMLHttpRequest - https://bugzilla.mozilla.org/show_bug.cgi?id=768596 - If user pressed the ESC button, XMLHttpRequest fires onerror instead of onabort.

Chrome issues:

1. "error" is not fired at EventSource after window.stop usage
2. "close" at WebSocket with code = 1001 - https://code.google.com/p/chromium/issues/detail?id=51687

Issues with Content-Disposition / <a download> and subframe navigation:

https://code.google.com/p/chromium/issues/detail?id=103238 - Requests cancelled after changing window location.
https://code.google.com/p/chromium/issues/detail?id=225654 - EventSource close event doesn't fire when computer put into sleep mode
https://code.google.com/p/chromium/issues/detail?id=153570 - EventSource is closed without notifications on a file download
https://code.google.com/p/chromium/issues/detail?id=260144 - B/f navigation stops ongoing XMLHttpRequests, even if only a subframe is navigated
https://bugzilla.mozilla.org/show_bug.cgi?id=564744 - XMLHttpRequest request cancelled (with status of 0 and no response text) if file downloading window pops up meanwhile
https://bugzilla.mozilla.org/show_bug.cgi?id=896666 - Need to not put EventSource and WebSocket into loadgroups or something

[AFBarstow]

@jungkees @hallvors are these going to be added to your tests?

[hallvors]

Didn't we already add the XHR-related tests in this issue? #242

[Yaffle]

@hallvors , yes , but #285 may also be useful

[hallvors]

@Yaffle : I work only on the XHR part of this test suite, I think the pull request in #285 only affects eventsource tests. I don't know who is the "test facilitator" for event-source, but I'm sure @AFBarstow can help you nag the right person so that these pull requests can be handled! :-)

[AFBarstow]

@hallvors the eventsource (aka Server-Sent Events) test facilitators are @zqzhang and @tinazhao

[hallvors]

ping @zqzhang and @tinazhao - something to merge in here? :-)

[zqzhang]

@hallvors - nothing here. The eventsource test is in #285; and @Yaffle is expected to address the review comments there.

After #285 is merged, this one can be closed :)

[hallvors]

@zqzhang Thanks for clarifying!