So.. how do we solve this problem? Should we clear script-set window.name on page traversals? What does IE do, I wonder?
MSIE6/winxpsp2 does the same thing as we do strangely enough.
AFAIK this is the first time this obvious issue has ever been raised, and AFAIK all browsers have had the same problem since, what, 1998? It makes sense to me to set a window's name to the zero-length string any time it changes domains.
I don't know about changing the name of the windows when we load new pages into the window, but maybe we should make changes to window.name only go into the DOM object, and we'd reset that on pageload?
It's unclear to me that this should become a stop ship bug. Since dan and jesse got cc'ed, can someone elaborate if this is a security issue? Time is running out for this bug which seems to have been idle since January. Leaving the nomination alone for right now pending some more information.
not a blocker until someone provides a compelling reason to take this so late in the game.
I believe it's a security bug to allow one site to affect where another site's window.open() calls put content. We have all sorts of checks to prevent just that, and this bug is about a way for a site to completely bypass all those checks. Of course I don't know whether that's a compelling reason, since I'm not clear what our blocker criteria are at this point.