Problem is on Trunk Builds If you click on this link http://wiki.http//wiki.mozilla.org/Performance/Status_Meetings/2007-July-18 in Thunderbird it opens a new Tab in Firefox that is greyed out and not usable. You can`t also use this tab for other urls. On Firefox 2.0.0.5 it displays only the "Can`t find page" Net Error and the Tab is usable. Regression between 2007-06-14 and 2007-06-15 (thanks martijn) http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2007-06-14+04&maxdate=2007-06-15+09&cvsroot=%2Fcvsroot So we think its a regression from Bug 371360

This might help diagnose the problem -- Using Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.4) Gecko/20070509 SeaMonkey/1.1.2 with WindowsXP. Preference set to open link from external application in a new window. After attempting and failing to open bad URL, I went to the address area and removed the "//wiki.http". On selecting Enter, I got the desired Web page. Thus, either the problem is limited to Firefox and is not in SeaMonkey, or else the problem is limited to configurations in which links from external applications open new tabs instead of new windows (or both).

The issue is that for external loads we CreateAboutBlankContentViewer before doing the load... which fires the unload event right then and there. If the load then fails in any way, we'll never be able to make use of this docshell again. I think it's reasoanble to just fire unload on the about:blank intemediary when the new page comes in. I checked, and the only other callers of CreateAboutBlankContentViewer are similar code in history loads of javascript: and EnsureContentViewer. This last will never be firing unload anyway, so no behavior change there. I _could_ probably change things so that we unset this boolean when we go to handle errors instead of doing it here. But this seems more robust to me.

Just a note, I just saw this with a page that took a long time to respond, when I hit stop before it loaded or timed out. > The issue is that for external loads we CreateAboutBlankContentViewer before > doing the load... which fires the unload event right then and there. If the > load then fails in any way, we'll never be able to make use of this docshell > again. I guess that means this case is already covered by the patch?

> Just a note, I just saw this with a page that took a long time to respond Loaded from outside the app?

(In reply to comment #5) > Loaded from outside the app? Yes.

Then yeah, this patch should help.

Marking blocker to make sure we get this in.

Comment on attachment 273049 [details] [diff] [review] Fixed it seems really fragile to maintain this as state on the docshell instead of the document or something

Comment on attachment 273049 [details] [diff] [review] Fixed Yeah, true. I was going for minimally invasive here....

Comment on attachment 273049 [details] [diff] [review] Fixed Need this on branches too if bug 371360 lands there.

Fixed.

it'd also be nice if the header had a comment about the purpose of this variable

Checked one in.

Comment on attachment 273049 [details] [diff] [review] Fixed approved for 1.8.1.7 and 1.8.0.14, a=dveditz for release-drivers

Fixed on both branches.

verified fixed 1.8.1.8 using Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.8.1.8pre) Gecko/20070929 BonEcho/2.0.0.8pre ID:2007092904 following my steps to reproduce Also verified fixed for Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.9a9pre) Gecko/2007092705 Minefield/3.0a9pre ID:2007092705 -> adding verified keyword and changing Bug status to verified