[Webkit-unassigned] [Bug 32383] New: Selecting article text at sfgate.com results in spurious back/forward entries

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=32383

           Summary: Selecting article text at sfgate.com results in
                    spurious back/forward entries
           Product: WebKit
           Version: 528+ (Nightly build)
          Platform: All
               URL: http://www.sfgate.com
        OS/Version: All
            Status: NEW
          Keywords: InRadar
          Severity: Normal
          Priority: P2
         Component: History
        AssignedTo: webkit-unassigned at lists.webkit.org
        ReportedBy: beidson at apple.com


Selecting article text at sfgate.com results in spurious back/forward entries.

Steps:
- Visit http://www.sfgate.com
- Click any article
- Click and hold back button, note no extraneous entries in back/forward list
- Select portions of the article a few times.
- Notice each time you do this, Safari seems to go through a page load (based
on the loading indicator in the location field)
- Click and hold back button again, note extraneous entries in back/forward
list
- Back no longer takes you out of the article, due to all the extra entries

What's going on is that sfgate.com is doing some sort of tracking on mouse up
after a selection by doing the following:
-Creating an iframe element with some forms
-Appending the iframe to the DOM
-Submitting the form
...then later:
-Removing the iframe from the DOM

The current heuristics WebKit has to not add a back/forward entry don't
preclude this iframe form submission, so an entry gets added for the
"navigation"

Without too much difficult, when a form is programmatically submitted we can
note that:
A - The frame was *not* created by the parser - that is, it was dynamically
created by script.
B - The form submission was not the result of a user event - that is, the
submission was also done by script.

In this case, we can lock the back/forward list.

I can't think of a reasonable site this would break, and will know if this
approach passes layout tests shortly.  At which point I'll post my patch.

This problem is in radar as <rdar://problem/7342725>

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