104627 - Loads very slowely (large page of UL and LI)

Status: RESOLVED DUPLICATE of bug 84466

(Core :: Layout, defect, P3)

Description

[rg1244]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:0.9.5) Gecko/20011011
BuildID:    2001101117

This page, and similar message board sites, take a very long time (about 15
seconds) to load the entire page.  In contrast, IE6, Netscape 4.78 and Opear
5.12 load this page in a matter of 3 or 4 seconds.  Netscape 6.1 also loads this
page very slowly. Furthermore, after the enitre page loads, when you click to
read a particular message, especially a message further down the page, and then
hit the back button to return to the main forum, it takes an extremely long time
to load the page and return to the old spot.

Reproducible: Always
Steps to Reproduce:
1. Navigate to the page.
2. Choose a particular message to read (somewhere in the middle of the page.)
3. Click the back button to return the main forum.

Actual Results:  Very long time to load the page and return to old spot.

Expected Results:  Hopefully, Mozilla can display the page much quicker.

To work around this delay, I open the individual message I want to read, in a
new window so that the main list of forum messages stays open and I don't have
to hit the back button and wait for it to load again.

Comment 1

[Grey Hodge (jX)]

WFM
Loads really fast here. Either was a bad build, bad Internet day, or slow
connection for reporter.

Reporter: Reopen and comment if you still see this on new builds.

Comment 2

[rg1244]

Hi Grey, thanks for your reply.  I have a fast DSL connection and this page has
always been slow to load.  It was slow on mozilla 0.9.4 milestone and now on
0.9.5 milestone which I started using as of last night. Thanks.

Comment 3

[rg1244]

Just to clarify, Mozilla will display the page very quickly, but the status bar
shows "transfering data from timezone.com" and after about 12 to 15 seconds, it
will show "document done: 15 seconds"  Thank you.

Comment 4

[Matthias Versen [:Matti]]

wfm with win2k build 200101013..

Comment 5

[Diego Biurrun]

*** This bug has been marked as a duplicate of 33269 ***

Comment 6

[John Morrison]

Reopening. This is not a dup of "Back/forward performance slow..." (which is 
really just a placeholder to cover checking for "dumb network stuff" when 
going back/forward; that's not to be confused with "this page is slow" bugs,
which are slow in either the back/forward case, or intial load, or reload).

-> waterson, cc: dbaron, attinasi

I don't know. Maybe you already have a bug (or planned task) like this one. 
But, this page is slow, and seems to be dominated by block reflow code. This
page is somewhat unique, in that just resizing the window will look up the
CPU, with no visual update of the window, for about 6 seconds on 500MHz/win2k
(current trunk build).

Comment 7

[John Morrison]

grumble. Really reopening.

Comment 8

[John Morrison]

->

Comment 9

[rg1244]

Hi John, thanks so much for confirming my bug report for this page.  I hope 
there will be a fix for this soon, as this is one on my favorite sites.  As a 
new Mozilla user, it's very nice to have this kind of feedback.  

Comment 10

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

It didn't seem all *that* slow to me, but then again I have a fast machine. 
Nothing in particular stands out in this profile -- the only thing I noticed out
of place was about 1% of the time in nsBlockFrame::UpdateSpaceManager in the
O(N^2) state recovery.

Comment 11

[John Morrison]

Well, in the pantheon of 'large-page-loads-slow', this isn't a major deity.
But, loading just from disk, it is 4.5 times slower to load in mozilla than
in nav4.x on the same win32 system. Although, perhaps, this particular bug 
doesn't show anything unique/unknown.

Comment 12

[Diego Biurrun]

It is all that slower. I just tried loading it from disk on my K6-2-500 with
512MB RAM under Linux and Netscape 4.77 loads it in just under three seconds
while 0.9.5 takes around 15 seconds.

Comment 13

[RC]

20 seconds in Netscape 6.1
6 seconds in IE6
5 seconds in Opera
3 seconds in Lynx

Comment 14

[Diego Biurrun]

I've been seeing this under Linux as well, OS --> ALL

I looked around again and found a very similar bug, the source of both pages
looks the same, an endless waste of <li> tags.

*** This bug has been marked as a duplicate of 84466 ***

Comment 15

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Did you profile to ensure that the problems were really the same?

Comment 16

[Diego Biurrun]

No, I did not profile. The symptoms are identical, though, and page source 
appears quite similar.
What's the alternative? Marking one bug as depending on the other and revisiting 
it once the other gets fixed?