User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.3a1pre) Gecko/20091219 Minefield/3.7a1pre Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.3a1pre) Gecko/20091219 Minefield/3.7a1pre ID:20091219042456 Firefox hangs for several minutes trying to load: http://www.whatwg.org/specs/web-apps/current-work/ At first I thought it might be a JavaScript or HTML parsing issue, but after extensive testing, it now appears to be a problem with Firefox's networking back-end. The following are timings for when the page is first requested to when the page finishes loading and CPU utilization drops back down to idle levels (remote - denotes that the page is being requested from the Internet; local - denotes that the page is being loaded from a local copy [attached]): NOTE: My testing was on a P3, so people with more current processors will not see hangs anywhere near as long. FIREFOX 3.7 (new profile): remote: 10 minutes 30 seconds local: 1 minute 15 seconds remote (HTML 5): 10 minutes 15 seconds CHROMIUM 4.0.x.x: remote: 1 min 25 sec local: 1 min 34 sec IE 8 remote: 1 min 30 sec local: crash IE 6 (remote): remote: 2 min 40 sec local: 4 min 14 sec Not sure about the weirdness with the IE timings, but they do report scripting errors. For Firefox, I also did JavaScript profiling (for remote and local), so this can be used to compare, but I doubt this is a JavaScript issue. Reproducible: Always Steps to Reproduce: 1. Create a new profile and launch Firefox 2. Visit http://www.whatwg.org/specs/web-apps/current-work/ and measure time it takes to complete loading (check Windows Task Manager). 3. Close tab or load about:blank; then "Clear Recent History" 4. Compare by loading the attached testcase.

The content of attachment 418521 [details] has been deleted by Reed Loden [:reed] <reed@reedloden.com> who provided the following reason: Removal requested by attachment submitter via IRC. The token used to delete this attachment was generated at 2009-12-19 20:17:39 PST.

Regressed with the 20070227 Trunk build. Regression window: _FAST_ Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a3pre) Gecko/20070226 Minefield/3.0a3pre ID:2007022604 remote: 2 min 3 sec local: 2 min 35 sec _SLOW_ Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a3pre) Gecko/20070227 Minefield/3.0a3pre ID:2007022704 remote: 9 min 13 sec local: 2 min 34 sec

Page loads here in about 6 secs on my 3meg DSL. Hanging, and fails to "Not Responding" when scrolling down the page several times using the mouse-wheel, where I hung at 25% CPU on a 2.3gig Athlon Phenom II x-4 Quad and 8 gig RAM. Had to force close Minefield with the task-manager. Setting to new...

(In reply to comment #7) > Page loads here in about 6 secs on my 3meg DSL. Hanging, and fails to "Not > Responding" when scrolling down the page several times using the mouse-wheel, > where I hung at 25% CPU on a 2.3gig Athlon Phenom II x-4 Quad and 8 gig RAM. > > Had to force close Minefield with the task-manager. > > Setting to new... Forgot to add, running Win7 Home Premium and today's nightly: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.3a1pre) Gecko/20091221 Minefield/3.7a1pre Firefox/3.6 ID:20091221052153

Note: my reference to "networking" in the report is not about Internet bandwidth. This is why I posted results of other browsers for comparison. By the way, when you indicate that you have a "3meg DSL" connection, are you meaning 3 Mb or 3 MB? Factor of 8 difference. The total page load (HTML and ancillary files) is 5,936,867 bytes. If your DSL connection is 3 Mb, your load time would be impossible and would suggest it was only partially loaded in the 6 seconds. At 3 Mb is would take a minimum of 16 seconds, ignoring overhead, to load.

Oops: Of course I'm ignoring that the download may have been compressed, which would work in that case. :-)

The page has always been very slow to load on Gecko 1.9.x.x - Firefox 3.0.x and newer (that is bug 481131). That said, things have gotten worse apparently - and fairly recently. It now hangs on OS X 10.6.2 as soon as I attempt to page down (spacebar or page down key). I wonder if Hixie changed something in the scripts that doesn't sit well with Gecko. It didn't use to hang on Gecko 1.9.0.x (using Camino 2.1a1pre nightly build) nor nightly Minefield builds. Now it does on all those.

regression range based on comment 6: http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=PhoenixTinderbox&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2007-02-26+04%3A00%3A00&maxdate=2007-02-27+04%3A00%3A00&cvsroot=%2Fcvsroot

OK. So there are several things going on with this page: 1) There's the initial page load. 2) There's an onload handler that runs some scripts that asynchronously load more data using XHR and then do some things that are known to be slow. IU, I assume that if you disable script you don't see the long hang? Looking at the range from comment 12, it looks like bug 371536 landed in that range and changed what views are created. Bug 526394 is about a known slow script issue on the HTML5 page due to positioned scrollable elements on the HTML5 page, but I would have thought it would have been just as slow when positioned elements got views... Maybe not if all those kids have a single positioned ancestor with no other kids, though? In any case, there's a definite hang on that page with 1.9.2 in my testing, just like in 1.9.3, so I'm not seeing the regression you're seeing...

@Boris: See comment 6: The regression appeared in 1.9.0a3 -- way before 1.9.2

(In reply to comment #13) > IU, I assume that if you disable script you don't see the long hang? Correct. With JavaScript disabled, both builds (2007022604 and 2007022704) load the page in 60 seconds.

This bug is actually a dup of bug 524436.

Yep, sounds like it. I have no idea why this was affected by the abs pos view thing, though...

So let me get this straight (per bug 481131 comment 29). Local with script disabled takes 1 minute? And local with script enabled takes 1 minute 15 seconds? I have a hard time believing that.

(In reply to comment #18) > I have a hard time believing that. Believe it my friend. Most of that time is spent loading the HTML, which is 5.38 MB. Firefox has never been efficient at loading large text documents -- even IE6 has always been faster in this regard. And I know of non of the major browsers that does not trounce Firefox in that regard. So it doesn't take long to load the scripts. Even after everything is loaded, if I scroll, I get about a 10 sec hang then everything is okay. This is for the local case. The 1 minute is because my system is a P3, but of course, all my measurements are relative, being all taken on the same system. Verify the relative consistency of the measurements for yourself, if you don't believe me. But be aware, if again trying to verify the regression, that it regressed during *1.9.0a3*, per comment 6 and NOT 1.9.3, as it appears you presumed, per comment 13. As I stated, I can confirm that the scripts do get loaded with the local case, and you can confirm this yourself.

Fixed by bug 526394

That is, only the regression noted in this bug. Bug 481131 still needs fixing.