161633 - dataviz.com - Browser run extremely slow while visiting any page of dataviz.com

Status: RESOLVED FIXED

(Tech Evangelism Graveyard :: English US, defect)

Description

[Son]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.1b) Gecko/20020806
BuildID:    2002080614

Browser run extremely slow while visiting any page of dataviz.com.
Virtual memory usage increases. No bug nor crash, however.

running on a WinXP box, Pentium III 400Mhz, 192MB RAM, virtual mem managed by
the OS.

Reproducible: Always
Steps to Reproduce:
1. Open anypage from http://www.dataviz.com
2. Take a coffee
3. Return back to wait (exagerate)

Comment 1

[Jo Hermans]

Confirmed on NT - build 2002072104 (you can grow your own coffee while waiting
for this page).

It seems a issue with both the menus and the image-rollovers. When you disable
JavaScript, there's no problem.

I'm not sure to which I should dupe this against - but it must be one of the
dependents of bug 113492. 

Comment 2

[Boris Zbarsky [:bzbarsky]]

The page has a wonderful menu system... The relevant part of the code that
builds it is:

var
BV=parseFloat(navigator.appVersion.indexOf("MSIE")>0?navigator.appVersion.split(";")[1].substr(6):navigator.appVersion);
var BN=navigator.appName;
var OP=(navigator.userAgent.indexOf('Opera')!=-1&&BV>=4)?true:false;
var SM=(BN.indexOf('Netscape')!=-1&&(BV>=5)||OP)?true:false;

function xrep(s, f, n) {
 if ((s.length > 100) && SM) {
     while (s.indexOf(f) != -1) {
         s = s.substr(0, s.indexOf(f)) + n + s.substr(s.indexOf(f) + f.length);
     }
 } else {
     s = s.split(f).join(n);
 }
 return s;
}

The expr() function is called thousands of times on a string "s" that ends up
being over 100KB long.  Look at that function carefully. Notice the basic structure:

if (Opera or Netscape) {
  do something really slow and memory-intensive;
} else {
  do something really fast;
}

This is all in the "IE" version of the code which the three share; NS4 gets its
own version of the whole file.

I set "SM" to "false" in a local copy of the page, just to test.  The page came
up instantaneously (instead of the 20-40 seconds it took with the code as-is)
and worked flawlessly.

This is _so_ not a Mozilla bug....

Comment 3

[Son]

Do you plan to report this "Moz-friendly attitude" to DataViz?

Comment 4

[Boris Zbarsky [:bzbarsky]]

That's what the Evangelism product is for... Read the product description at
http://bugzilla.mozilla.org/enter_bug.cgi

Comment 5

[Bob Clary [:bc] (inactive)]

tech evang june 2003 reorg

Comment 6

[Gérard Talbot]

*** Bug 173944 has been marked as a duplicate of this bug. ***

Comment 7

[sjw]

Seems fixed / redesigned. Or Firefox became much faster :D
[closeme]