[Webkit-unassigned] [Bug 9629] Safari crashes when AJAX coincides with unloading page

[bugzilla-daemon at opendarwin.org]

http://bugzilla.opendarwin.org/show_bug.cgi?id=9629


ddkilzer at kilzer.net changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ddkilzer at kilzer.net




------- Comment #5 from ddkilzer at kilzer.net  2006-06-28 05:25 PDT -------
(In reply to comment #4)
> We don't see the problem with Safari 2.0.3 (417.9.2).  Does that help re
> providing a hint as to how we might be provoking this crash?

The stack trace clearly shows that it's something related to an XMLHttpRequest.

If you want to find the exact bug (or look at the exact source code triggering
the bug), there are a couple things you can try.

1. Download the earliest nightly build image (http://nightly.webkit.org/) and
test to see if the problem occurs in it.  If it does, then you can do a "binary
search" between the earliest nightly and the latest one to find where it was
fixed.  Once you find two nightlies (one that works and one that doesn't) in
sequential order, you can review the CVS/SVN logs between those revisions to
find the commit that fixed it.  Note that if the issue was fixed in Safari
2.0.3 (417.9.2), then it was probably already fixed before WebKit was released
as an open source project.

2. Check out an old branch of WebKit code from the SVN repository that matches
the version of Safari and WebKit you're using on 10.3.9.  You can use this page
to determine which version of WebKit you want to look at:

http://developer.apple.com/internet/safari/uamatrix.html

Then head over to the anonsvn.opensource.apple.com Subversion repository and
check out a branch (or a tag) corresponding to the version you're running, then
open up the source files listed in the stack trace.

http://anonsvn.opensource.apple.com/svn/

Here are a couple example SVN URLs that you might want to investigate (by
checking out the source):

svn://anonsvn.opensource.apple.com/svn/webkit/branches/Safari-1-3-branch/
svn://anonsvn.opensource.apple.com/svn/webkit/tags/Safari-312/

Looking at the source may give you an idea of how you're triggering the bug.  I
wouldn't expect Apple to fix any bug you found in code this old, though.


-- 
Configure bugmail: http://bugzilla.opendarwin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.