33387 - Mac: Looooong delay when loading this site (bgColor)

Status: RESOLVED WORKSFORME

(Core :: CSS Parsing and Computation, defect, P3)

Description

[ssym]

DESCRIPTION

This site takes a Looooooooong time to finally show anything on my 400MHz PIII. 
 And Mozilla seems to "freeze" whilst it is retrieving the page - no buttons 
work, exhibits all other classic symptoms of frozen apps.

HOW TO REPLICATE

Go to http://crash.to/DaPsychoZone

BUILD & PLATFORM

Using the 24/3/00 Build on Win95 ver. 4.00.1111

ADDITIONAL INFO

I am going through a proxy server...

Comment 1

[troy]

Somewhere in the page there is a script that sets the bgcolor. That 
causes the style system to reresolve all of the style contexts. That 
seems to happen at least twice that I saw, and it causes horrible 
performance problems.

Here's a stack trace that shows the problem:

FrameManager::ReResolveStyleContext(nsIPresContext * 0x00d5ac50, nsIFrame * 
0x01618d18, nsIStyleContext * 0x01697720, nsIContent * 0x0162812c, int 0, 
nsIAtom * 0x00000000, nsStyleChangeList & {...}, int 3, int & 3) line 1336
FrameManager::ReResolveStyleContext(nsIPresContext * 0x00d5ac50, nsIFrame * 
0x01618cbc, nsIStyleContext * 0x01776b98, nsIContent * 0x016280b0, int 0, 
nsIAtom * 0x00000000, nsStyleChangeList & {...}, int 3, int & 3) line 1336
FrameManager::ReResolveStyleContext(nsIPresContext * 0x00d5ac50, nsIFrame * 
0x01618c74, nsIStyleContext * 0x01776648, nsIContent * 0x01624e50, int 0, 
nsIAtom * 0x00000000, nsStyleChangeList & {...}, int 3, int & 3) line 1336
FrameManager::ReResolveStyleContext(nsIPresContext * 0x00d5ac50, nsIFrame * 
0x01618324, nsIStyleContext * 0x01725788, nsIContent * 0x01624aa8, int 0, 
nsIAtom * 0x00000000, nsStyleChangeList & {...}, int 3, int & 3) line 1336
FrameManager::ReResolveStyleContext(nsIPresContext * 0x00d5ac50, nsIFrame * 
0x016182b8, nsIStyleContext * 0x0168ff10, nsIContent * 0x01624aa8, int 0, 
nsIAtom * 0x00000000, nsStyleChangeList & {...}, int 3, int & 3) line 1336
FrameManager::ReResolveStyleContext(nsIPresContext * 0x00d5ac50, nsIFrame * 
0x01618264, nsIStyleContext * 0x0168f9c0, nsIContent * 0x0161bcd0, int 0, 
nsIAtom * 0x00000000, nsStyleChangeList & {...}, int 3, int & 3) line 1336
FrameManager::ReResolveStyleContext(nsIPresContext * 0x00d5ac50, nsIFrame * 
0x0161821c, nsIStyleContext * 0x0161b528, nsIContent * 0x015f4c00, int 0, 
nsIAtom * 0x00000000, nsStyleChangeList & {...}, int 3, int & 3) line 1336
FrameManager::ReResolveStyleContext(nsIPresContext * 0x00d5ac50, nsIFrame * 
0x01618148, nsIStyleContext * 0x0161a318, nsIContent * 0x00000000, int 0, 
nsIAtom * 0x0152ac90, nsStyleChangeList & {...}, int 3, int & 3) line 1336
FrameManager::ComputeStyleChangeFor(FrameManager * const 0x014e9c58, 
nsIPresContext * 0x00d5ac50, nsIFrame * 0x01618148, int 0, nsIAtom * 0x0152ac90, 
nsStyleChangeList & {...}, int 3, int & 0) line 1371
nsCSSFrameConstructor::AttributeChanged(nsCSSFrameConstructor * const 
0x014e92f0, nsIPresContext * 0x00d5ac50, nsIContent * 0x015f4c00, int 0, nsIAtom 
* 0x0152ac90, int 3) line 8394
StyleSetImpl::AttributeChanged(StyleSetImpl * const 0x014e9220, nsIPresContext * 
0x00d5ac50, nsIContent * 0x015f4c00, int 0, nsIAtom * 0x0152ac90, int 3) line 
1014
PresShell::AttributeChanged(PresShell * const 0x014e93d8, nsIDocument * 
0x00db9090, nsIContent * 0x015f4c00, int 0, nsIAtom * 0x0152ac90, int 3) line 
2748 + 57 bytes
nsDocument::AttributeChanged(nsDocument * const 0x00db9090, nsIContent * 
0x015f4c00, int 0, nsIAtom * 0x0152ac90, int 3) line 1739 + 32 bytes
nsGenericHTMLElement::SetHTMLAttribute(nsIAtom * 0x0152ac90, const nsHTMLValue & 
{...}, int 1) line 1140
nsGenericHTMLElement::SetAttribute(int 0, nsIAtom * 0x0152ac90, const nsString & 
{...}, int 1) line 1024 + 20 bytes
nsHTMLBodyElement::SetBgColor(nsHTMLBodyElement * const 0x015f4bf8, const 
nsString & {...}) line 655
nsHTMLDocument::SetBgColor(nsHTMLDocument * const 0x00db9178, const nsString & 
{...}) line 2240 + 19 bytes
SetHTMLDocumentProperty(JSContext * 0x014530d0, JSObject * 0x01613690, long -29, 
long * 0x0012f0ac) line 568
js_Interpret(JSContext * 0x014530d0, long * 0x0012f2b0) line 2326 + 954 bytes
js_Execute(JSContext * 0x014530d0, JSObject * 0x01612798, JSScript * 0x016870c0, 
JSFunction * 0x00000000, JSStackFrame * 0x00000000, unsigned int 0, long * 
0x0012f2b0) line 850 + 13 bytes
JS_EvaluateUCScriptForPrincipals(JSContext * 0x014530d0, JSObject * 0x01612798, 
JSPrincipals * 0x01641910, const unsigned short * 0x01649028, unsigned int 1247, 
const char * 0x01641998, unsigned int 61, long * 0x0012f2b0) line 2744 + 27 
bytes
nsJSContext::EvaluateString(nsJSContext * const 0x01453058, const nsString & 
{...}, void * 0x01612798, nsIPrincipal * 0x0164190c, const char * 0x01641998, 
unsigned int 61, const char * 0x004eb468, nsString & {...}, int * 0x0012f30c) 
line 404 + 55 bytes
HTMLContentSink::EvaluateScript(nsString & {...}, nsIURI * 0x014a6168, int 61, 
const char * 0x004eb468) line 4305
HTMLContentSink::ProcessSCRIPTTag(const nsIParserNode & {...}) line 4553
HTMLContentSink::AddLeaf(HTMLContentSink * const 0x01519810, const nsIParserNode 
& {...}) line 2971 + 12 bytes
CNavDTD::AddLeaf(const nsIParserNode * 0x01615290) line 3308 + 22 bytes
CNavDTD::HandleScriptToken(const nsIParserNode * 0x01615290) line 1890 + 12 
bytes

Comment 2

[Marc Attinasi]

I tested this page with the background color changes in my local tree. It does 
take a relatively long time to display, but during that time the background is 
getting progressively lighter in color (very nice!). When it is finally white, 
the page shows pretty quickly.

I'm not sure what the problem is here, but that's what I see. The changes I made 
for the bgcolor will be in this weekend (see bug 11491 for completion 
indication).

Comment 3

[Marc Attinasi]

I was just re-checking this page and I noticed that the UI is pretty much locked
-up during the fade-in that lasts about 45 seconds.

Also, I compared it the the PR1 behavior and it is much better now - in PR1
there was no visible fade from black since the canvas was not being invalidated
and redrawn when the color was changed, so all the time that it now takes to do
the fade looked like nothing was happening. I'm guessing this can be marked
fixed or invalid now - take a look.

Comment 4

[Pierre Saslawsky]

The fading works on Windows, not on Mac. It's a nice visual effect but not 
critical for this release. Marked PP / M19.

Comment 5

[Pierre Saslawsky]

Reassigned to dcone to quickly look into it or just mark "Future".

Comment 6

[dcone (gone)]

This bug has been marked future because we have determined that it is not 
critical for netscape 6.0.  If you feel this is an error, or if it blocks your 
work in some way -- please attach your concern to the bug for reconsideration.

Comment 7

[ssym]

Why is this not critical?  Surely it reveals a JavaScript bug not present in 
other browsers?  Is this not cause for concern??

Rhetorical questions aside, YES, it is a concern, since the site in question is 
visited by me on a regular basis.  A 30 second or more delay is not acceptable 
each time I want to access the site!  In addition, the maintainer of the site 
(Emily Fernandez) has it as her homepage, which means that delaying this until 
after initial release will mean a long delay each time her browser starts up.

The Windows 2000-07-12 nightly build still displays these problems.  If for no 
other reason, consider the script-kiddie risk posed once people find out that 
they can effectively freeze the entire browser with a very simple script?

On another note, since this problem shows on a PC running Win95, why is the 
Platform shown as "Macintosh" with the OS being "Mac System 9.0"?

Comment 8

[Kevin McCluskey (gone)]

I'm not seeing any performance problems with todays build, 12/13/2000 NT build.
Marking WORKSFORME. This may have been fixed by speeding up the tiling of
background images.
 

Comment 9

[Hixie (not reading bugmail)]

Netscape's standard compliance QA team reorganised itself once again, so taking 
remaining non-tables style bugs. Sorry about the spam. I tried to get this done 
directly at the database level, but apparently that is "not easy because of the 
shadow db", "plus it screws up the audit trail", so no can do...