FWIW, I also dug into it in windbg, and it appears that the compiled script was getting serialized as part of IE's "fast-back" functionality, probably as a consequence of the fact that the script elements were created via doc.write rather than being fetched through the browser's normal mechanism.
On Tue, Jul 7, 2009 at 4:09 PM, Joel Webber <j...@google.com> wrote: > Yup, the XHR-based IFrameTemplate seems to have been causing the HUGE delay > you and I were both seeing when setting the url #hash for History.newItem(). > I'm afraid I'm going to have to revert back to the more-poorly-parallelized > version of the selection script to get this fixed (at least for now). Thank > god we didn't put this in the 1.6 release branch. > > > On Tue, Jul 7, 2009 at 2:29 PM, Joel Webber <j...@google.com> wrote: > >> On Tue, Jul 7, 2009 at 1:40 PM, Amir Kashani <amirkash...@gmail.com>wrote: >> >>> Opened issue 3808 and attached a patch to it that reverts the change and >>> fixes the issue for me. >>> It's interesting you mentioned IE6/7. I didn't test them as thoroughly as >>> IE8, but I did notice a HUGE added delay in newToken calls at one point, in >>> the range of 10-20 seconds. I'm not sure if this was directly related to the >>> XHR change or any of the number of hacks I was trying out at the time >>> though. >>> >> >> I'm glad to hear you say that, as I'm about to run a test on my end to see >> whether the XHR change is causing that issue. I'll keep everyone posted. >> >> > --~--~---------~--~----~------------~-------~--~----~ http://groups.google.com/group/Google-Web-Toolkit-Contributors -~----------~----~----~----~------~----~------~--~---