[Webkit-unassigned] [Bug 125840] Have XHR.getResponseHeader() return null in initial ready states

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=125840





--- Comment #16 from youenn fablet <youennf at gmail.com>  2014-01-10 02:35:16 PST ---
(In reply to comment #15)
> What do other browsers do? The spec changed many times over the last few years, and it can't be taken very seriously.
> 

Firefox is aligned with the patch and latest spec.
Blink recently aligned to it as well.

> > Source/WebCore/xml/XMLHttpRequest.cpp:1013
> > +    if (m_state < HEADERS_RECEIVED || m_error)
> 
> Why was "|| m_error" added here? This is not mentioned in ChangeLog.
> 

It makes it clearer that the code is aligned with the spec.
The code is also made more robust.

> If this is a behavior change, it should be landed in a separate patch to facilitate regression testing when a site breaks. If there is no behavior change, why is the code changed? I'm not sure if there is any test coverage for this.

I can add a test to cover the "m_error" check (e.g. xhr.send();xhr.abort();).
The test would pass today without the patch: the current code is already aligned with the spec in most cases, m_response is cleared when m_error is set to true.

There may be corner cases though but I am not aware of any right now.
Filing a separate bug may therefore be safer?

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.