Bug 29668

Summary:

[GTK, WinCairo] REGRESSION: XMLHttpRequest gives bogus status value

Product:

WebKit

Reporter:

Brent Fulgham <bfulgham>

Component:

WebCore Misc.

Assignee:

Nobody <webkit-unassigned>

Status:

RESOLVED DUPLICATE

Severity:

Normal

CC:

jdance, mrowe

Priority:

Version:

528+ (Nightly build)

Hardware:

OS:

Windows XP

Attachments:

Description	Flags
Test case for failing XMLHttpRequest invocation.	none
Expected results (Mac OS X)	none
Failed test results generated on Windows (WinCairo) build.	none

Brent Fulgham

Reported 2009-09-22 21:38:58 PDT

Created attachment 39975 [details] Test case for failing XMLHttpRequest invocation. User reports that XMLHttpRequest under Windows after @r48212. This was known to work properly in @r47083 -- it is not known when the regression occured. The bad behavior was Behavior: On send(), it gives an immediate readyState of 4, and a status of 0. Attached is a simple test which you must open with file protocol so that you don't have to worry about cross domain issues.

Attachments
Test case for failing XMLHttpRequest invocation. (1.37 KB, text/html) 2009-09-22 21:38 PDT, Brent Fulgham	no flags	Details
Expected results (Mac OS X) (342 bytes, text/plain) 2009-09-23 10:07 PDT, Brent Fulgham	no flags	Details
Failed test results generated on Windows (WinCairo) build. (115 bytes, text/plain) 2009-09-23 10:09 PDT, Brent Fulgham	no flags	Details
View All Add attachment proposed patch, testcase, etc.

Mark Rowe (bdash)

Comment 1 2009-09-22 23:39:03 PDT

Which ports does this affect? What is the expected behavior of the attached test case?

Brent Fulgham

Comment 2 2009-09-23 10:07:58 PDT

Created attachment 40003 [details] Expected results (Mac OS X) The attached results are 'good' values generated by running on a Mac OS nightly.

Brent Fulgham

Comment 3 2009-09-23 10:09:31 PDT

Created attachment 40004 [details] Failed test results generated on Windows (WinCairo) build. Example of failed output. This result is generated on the WinCairo port as well as the Gtk/Linux port.

Brent Fulgham

Comment 4 2009-09-23 10:10:06 PDT

Added GTK and WinCairo labels to bug description.

Brent Fulgham

Comment 5 2009-09-23 12:11:34 PDT

Bad behavior seems to have been introduced by checks for actual response/same origin in DocumentThreadableLoader::didReceiveResponse. If the original behavior of calling m_client->didReceiveResponse is used the test passes. This raises a few questions: 1. Why does the Apple port differ here? 2. Why isn't useful error information returned to the XMLHttpRequest object that the load was rejected due to cross-origin checks?

Brent Fulgham

Comment 6 2009-09-23 13:16:16 PDT

Closing bug as duplicate. See Bug 29255. The behavior can be overridden through WebPreferences. *** This bug has been marked as a duplicate of bug 29255 ***

Note You need to log in before you can comment on or make changes to this bug.