Bug 30424 – REGRESSION (r48687): Pages on ucas.com appear blank

• Component: WebCore Misc.
• Status: RESOLVED FIXED

[Fumitoshi Ukai]

WebKit r48687 causes the regression reported on http://crbug.com/24123

Visit the following URLs. It is expected to show login pages, but show only blank page with WebKit r48687 or later.
It works as expected with WebKit r48686.

 https://track.ucas.com/ucastrack/Login.jsp
 https://apply2.ucas.com/appreg2010/SecurityServlet

As investigated Developer Tools in chrome, it doesn't show any body in Element page.
However, view source shows correct html contents.

[Mark Rowe (bdash)]

In the future please describe the bug in the “Summary" field.  If the bug is a regression you should include the REGRESSION tag at the start of the summary, and you should CC the author of the original change on the bug.

[Mark Rowe (bdash)]

<rdar://problem/7308952>

[Brady Eidson]

*** Bug 30704 has been marked as a duplicate of this bug. ***

[Brady Eidson]

In the bugzilla dupes and radar dupes we've seen of this issue, we've seen parsing of the page stop once a schedule redirection is hit.  In a couple cases it's been a history.forward(), and in a couple it's been a <meta> refresh.

That symptom matches quite obviously with the nature of r48687

[Brady Eidson]

Created attachment 41892 [details]
Proposed fix and layout test

[Darin Adler]

Comment on attachment 41892 [details]
Proposed fix and layout test

> +    // http://webkit.org/b/30424 - Invalid history navigations (such as history.forward() during a new load) have
> +    // the side effect of cancelling any scheduled redirects. We also avoid the possibility of cancelling the
> +    // current load by avoiding the scheduled redirection altogether.

I am not a huge fan of including bug URLs in the code. Otherwise, this comment is fine. I'd just leave out the URL.

> +if (window.layoutTestController)
> +    window.layoutTestController.dumpAsText();

The second line can just say "layoutTestController", not "window.layoutTestController".

> +http://webkit.org/b/30424 and <rdar://problem/7308952> - REGRESSION (r48687) - Some pages appear blank.<br>

You need an &lt; here instead of the < sign.

[Brady Eidson]

Committed r50111.
http://trac.webkit.org/changeset/50111

[Brady Eidson]

*** Bug 30746 has been marked as a duplicate of this bug. ***

[Simon Fraser (smfr)]

#webkit is reporting crashes:
http://nopaste.com/p/aeY6lni8fb

[Darin Adler]

Lets put those crashes in a new bug report. Are they reproducible?

[Simon Fraser (smfr)]

They are being tracked via bug 30839.

[Brady Eidson]

There was already a new bugzilla - https://bugs.webkit.org/show_bug.cgi?id=30839

But they are *NOT* fallout from this change.  We have radars for them from weeks ago (<rdar://problem/7313414>)

Sadly, they are not reliably reproducible.

[Justin Hibbits]

Just tested with the nightly 50124, and it works correctly, thanks.