Confirm a serious CSS bug in IE 7 (infinite loop)

[=?ISO-8859-1?Q?G=E9rard_Talbot?=]

[followup-to set: comp.infosystems.www.authoring.stylesheets]

Dear fellow CSS colleagues and web authors in alt.html discussion forum,

I would like to ask you to help me confirm that there is a serious bug
in IE 7 final release build 5730.11 under XP Pro SP2 with all the
patches up-to-date.

Please visit:

http://www.gtalbot.org/BrowserBugsSe...teLoopIE7.html

or

bug #42 at

http://www.gtalbot.org/BrowserBugsSection/MSIE7Bugs/

Just load the page.

Actual results (external signs of infinite loop at rendering time):

a) the CPU maxed and maintained at high percentage in the Task manager and

b) the incomplete (too long scrollbar thumb) rendering of the vertical
scrollbar in the Subject Index Pane.

If you move the scrollbar thumb, then it will get resized to what its
size should have been in the first place and then then cpu %tage will
disappear, will shortly get back to zero.

Another discovery: if you zoom in the webpage, the actual results
(external signs of infinite loop at rendering time)

Please help me confirm all this and/or report your findings. Thank you
for your time and cooperation here,

Gérard

[followup-to set: comp.infosystems.www.authoring.stylesheets]
--
Using Web Standards in your Web Pages (Updated Dec. 2006)
http://developer.mozilla.org/en/docs...your_Web_Pages

[Els]

Gérard Talbot wrote:

[followup-to set: comp.infosystems.www.authoring.stylesheets]

Dear fellow CSS colleagues and web authors in alt.html discussion forum,

I would like to ask you to help me confirm that there is a serious bug
in IE 7 final release build 5730.11 under XP Pro SP2 with all the
patches up-to-date.

Check.

http://www.gtalbot.org/BrowserBugsSe...teLoopIE7.html

Just load the page.

Done.
Loads quickly, no problems.

Actual results (external signs of infinite loop at rendering time):

a) the CPU maxed and maintained at high percentage in the Task manager and

Nope, CPU barely notices the page.

b) the incomplete (too long scrollbar thumb) rendering of the vertical
scrollbar in the Subject Index Pane.

Yes, confirmed.

If you move the scrollbar thumb, then it will get resized to what its
size should have been in the first place

Yup.

and then then cpu %tage will
disappear, will shortly get back to zero.

Not sure if I understood that sentence, but touching the scrollbar has
no effect on the CPU, but then again, it wasn't maxing out in the
first place.

Another discovery: if you zoom in the webpage, the actual results
(external signs of infinite loop at rendering time)

Missing a verb here. What should the actual results be doing if I zoom
in the webpage? What it does here, is it does raise the CPU, max at
34% for IE, but goes down again on its own.

Please help me confirm all this and/or report your findings. Thank you
for your time and cooperation here,

You're welcome :-)
--
Els http://locusmeus.com/
accessible web design: http://locusoptimus.com/

[=?UTF-8?B?R8OpcmFyZCBUYWxib3Q=?=]

Els wrote :

Gérard Talbot wrote:

>[followup-to set: comp.infosystems.www.authoring.stylesheets]

Dear fellow CSS colleagues and web authors in alt.html discussion forum,

I would like to ask you to help me confirm that there is a serious bug
in IE 7 final release build 5730.11 under XP Pro SP2 with all the
patches up-to-date.

Check.

>http://www.gtalbot.org/BrowserBugsSe...teLoopIE7.html

>Just load the page.

Done.
Loads quickly, no problems.

>Actual results (external signs of infinite loop at rendering time):

a) the CPU maxed and maintained at high percentage in the Task manager and

Nope, CPU barely notices the page.

There must be something, an element of some sort, in the initial
conditions that I am missing in the description. Argh... Over here, I
get a cpu maxed 100% of the times when loading the page, immediately
after loading the page.

I notice that if my browser window is not maximized to begin with, the
cpu will not get maxed when loading the page. If my browser window is
maximized to begin with, then the cpu will get maxed after loading the page.

>b) the incomplete (too long scrollbar thumb) rendering of the vertical
scrollbar in the Subject Index Pane.

Yes, confirmed.

>If you move the scrollbar thumb, then it will get resized to what its
size should have been in the first place

Yup.

>and then then cpu %tage will
disappear, will shortly get back to zero.

Not sure if I understood that sentence, but touching the scrollbar has
no effect on the CPU, but then again, it wasn't maxing out in the
first place.

Over here, the cpu gets back to normal %tage (close to 0% or 0%) once I
drag the scrollbar thumb all the way down.

>Another discovery: if you zoom in the webpage, the actual results
(external signs of infinite loop at rendering time)

Missing a verb here. What should the actual results be doing if I zoom
in the webpage? What it does here, is it does raise the CPU, max at
34% for IE, but goes down again on its own.

It should have read:
Another discovery: if you zoom in (bottom right corner of IE7 browser,
in the status bar) the webpage, the actual results
(external signs of infinite loop at rendering time) will reappear: the
cpu will get maxed and will remain maxed.

Els, thank you for your time :)

Gérard
--
Using Web Standards in your Web Pages (Updated Dec. 2006)
http://developer.mozilla.org/en/docs...your_Web_Pages

[BootNic]

Gérard Talbot <ne***********@gtalbot.orgwrote:

news: 12************@corp.supernews.com

[snip]

I would like to ask you to help me confirm that there is a serious bug
in IE 7 final release build 5730.11 under XP Pro SP2 with all the
patches up-to-date.

build 5730.11 under XP Home SP2

[snip]

http://www.gtalbot.org/BrowserBugsSe...teLoopIE7.html [snip]

Actual results (external signs of infinite loop at rendering time):

a) the CPU maxed and maintained at high percentage in the Task
manager and

confirmed

b) the incomplete (too long scrollbar thumb) rendering of the vertical
scrollbar in the Subject Index Pane.

confirmed

If you move the scrollbar thumb, then it will get resized to what its
size should have been in the first place and then then cpu %tage will
disappear, will shortly get back to zero.

confirmed

Another discovery: if you zoom in the webpage, the actual results
(external signs of infinite loop at rendering time)

Any change to window size or zoom.
[snip]

--
BootNic Sunday, March 11, 2007 7:21 PM

"I suppose they are vicious rascals, but it scarcely matters what
they are. I'm after what they know."
*Gibson-Sterling, The Difference Engine*

[=?UTF-8?B?R8OpcmFyZCBUYWxib3Q=?=]

Els wrote :

You're welcome :-)

Els, one last request. Please try this webpage:

http://www.charlescooke.plus.com/ugextract.html

Just load it with MSIE7 and see how the cpu behaves. That webpage is the
original webpage from which I constructed that demo page.

Thank you for your time and cooperation; this is greatly appreciated!

Gérard
--
Using Web Standards in your Web Pages (Updated Dec. 2006)
http://developer.mozilla.org/en/docs...your_Web_Pages

[BobaBird]

On Sun, 11 Mar 2007 17:52:23 -0400, Gérard Talbot
<ne***********@gtalbot.orgwrote:

>I would like to ask you to help me confirm that there is a serious bug
in IE 7 final release build 5730.11 under XP Pro SP2 with all the
patches up-to-date.

I get similar results using the same IE7 build under XP Home SP2.

>Please visit:
http://www.gtalbot.org/BrowserBugsSe...teLoopIE7.html

Just load the page.

Actual results (external signs of infinite loop at rendering time):

a) the CPU maxed and maintained at high percentage in the Task manager and

Mine stayed in the 90s with an AMD Athlon 1.2. This is with 3 tabs
and a pop-up open. Closing a tab dropped CPU usage to the 70s, and
closing a second let it dip into the 60s though there is a lot of
fluctuation from 54-83.

>b) the incomplete (too long scrollbar thumb) rendering of the vertical
scrollbar in the Subject Index Pane.

If you move the scrollbar thumb, then it will get resized to what its
size should have been in the first place and then then cpu %tage will
disappear, will shortly get back to zero.

I have to drag it repeatedly to reach the bottom. Each move has a
corresponding re-size of the thumb until you reach reach the bottom at
which point it is sized properly to reflect how much of the DIV is
visible.

Side note: I can only use the mouse's scroll wheel when hovered over
the scrollbar; it has no effect when hovering over the content.

>Another discovery: if you zoom in (bottom right corner of IE7 browser,
in the status bar) the webpage, the actual results
(external signs of infinite loop at rendering time) will reappear: the
cpu will get maxed and will remain maxed.

Usage raises to and remains in the 90s (1 tab and 1 pop-up open) until
I manage to get the scrollbar dragged to the bottom. It then remains
at the prior level.

Also, while scrolling to reveal all the content and get the thumb
properly sized, some of the content will appear outside of
#SubjectIndexPane's border (above when scrolling down, below when
scrolling up).
--

Charles

[Steven Saunderson]

On Sun, 11 Mar 2007 17:52:23 -0400, Gérard Talbot
<ne***********@gtalbot.orgwrote:

Using IE7 5730.11 on XP Home (not Pro), all updates, with Process
Explorer, IE window maximised

a) the CPU maxed and maintained at high percentage in the Task manager and

No such problem here. CPU is negligible.

b) the incomplete (too long scrollbar thumb) rendering of the vertical
scrollbar in the Subject Index Pane.

If you move the scrollbar thumb, then it will get resized to what its
size should have been in the first place and then then cpu %tage will
disappear, will shortly get back to zero.

I see this. The scrollbar gets reduced gradually as I pull it down.

Another discovery: if you zoom in the webpage, the actual results
(external signs of infinite loop at rendering time)

No major CPU load but the box text does continue under the box when zoom
is not 100%.

--
Steven

[Els]

Gérard Talbot wrote:

Els wrote :

>You're welcome :-)

Having read your other post, I maximised the IE window, and I can
confirm that *your* testpage does cause the CPU to keep going (albeit
only at 31%) until I get that scrollbar resized.

Els, one last request. Please try this webpage:

http://www.charlescooke.plus.com/ugextract.html

On that page however, maximising the window means the list of words
doesn't need a scrollbar at all, hence, no CPU usage of any
significance.

Just load it with MSIE7 and see how the cpu behaves. That webpage is the
original webpage from which I constructed that demo page.

So, I made the window shorter (enough to show only half the list,
causing a scrollbar), and then refreshed the page. CPU peaked at 137%
(hundred and thirty seven!), and then kept maximised (with IE taking
between 52 and 130% somehow), until I alt-tabbed back to the window
I'm typing this post in. IE now keeps the CPU at about 50%.

Then I'm resizing the scrollbar, and notice that it doesn't want to go
all the way to the bottom. At some window heights it does, at others
it stops at "Fixed", and at others at "Position". This irregularity
doesn't happen on your test page though.

Thank you for your time and cooperation; this is greatly appreciated!

No problem :-)

--
Els http://locusmeus.com/
accessible web design: http://locusoptimus.com/

[=?UTF-8?B?R8OpcmFyZCBUYWxib3Q=?=]

Els wrote :

Gérard Talbot wrote:

>Els wrote :

>>You're welcome :-)

Having read your other post, I maximised the IE window, and I can
confirm that *your* testpage does cause the CPU to keep going (albeit
only at 31%) until I get that scrollbar resized.

>Els, one last request. Please try this webpage:

http://www.charlescooke.plus.com/ugextract.html

On that page however, maximising the window means the list of words
doesn't need a scrollbar at all, hence, no CPU usage of any
significance.

>Just load it with MSIE7 and see how the cpu behaves. That webpage is the
original webpage from which I constructed that demo page.

So, I made the window shorter (enough to show only half the list,

That's the element in the initial conditions of the test that I didn't
get. The "Subject Index Pane" must only show a part of the whole list.

causing a scrollbar), and then refreshed the page. CPU peaked at 137%
(hundred and thirty seven!), and then kept maximised (with IE taking
between 52 and 130% somehow), until I alt-tabbed back to the window
I'm typing this post in. IE now keeps the CPU at about 50%.

Then I'm resizing the scrollbar, and notice that it doesn't want to go
all the way to the bottom. At some window heights it does, at others
it stops at "Fixed", and at others at "Position". This irregularity
doesn't happen on your test page though.

>Thank you for your time and cooperation; this is greatly appreciated!

No problem :-)

Thank you for your time, Els. I appreciate your feedback on this. I have
report this issue to Microsoft and I'll notify them that it's a
confirmed bug and a serious one.

Gérard
--
Using Web Standards in your Web Pages (Updated Dec. 2006)
http://developer.mozilla.org/en/docs...your_Web_Pages

[=?ISO-8859-1?Q?G=E9rard_Talbot?=]

Steven Saunderson wrote :

On Sun, 11 Mar 2007 17:52:23 -0400, Gérard Talbot
<ne***********@gtalbot.orgwrote:

Using IE7 5730.11 on XP Home (not Pro), all updates, with Process
Explorer, IE window maximised

>a) the CPU maxed and maintained at high percentage in the Task manager and

No such problem here. CPU is negligible.

Can you try again with various browser window size.

>b) the incomplete (too long scrollbar thumb) rendering of the vertical
scrollbar in the Subject Index Pane.

If you move the scrollbar thumb, then it will get resized to what its
size should have been in the first place and then then cpu %tage will
disappear, will shortly get back to zero.

I see this. The scrollbar gets reduced gradually as I pull it down.

Yeah.. that's a minor side effect of the bug, I'd say.

>Another discovery: if you zoom in the webpage, the actual results
(external signs of infinite loop at rendering time)

No major CPU load but the box text does continue under the box when zoom
is not 100%.

The thing with zooming is that the document height gets longer creating
a document overflow within the browser window content area, rendering
area. It forces a reflow and re-rendering of the page.

Gérard
--
Using Web Standards in your Web Pages (Updated Dec. 2006)
http://developer.mozilla.org/en/docs...your_Web_Pages

[Steven Saunderson]

On Sun, 11 Mar 2007 21:02:15 -0400, Gérard Talbot
<ne***********@gtalbot.orgwrote:

No such problem here. CPU is negligible.

Can you try again with various browser window size.

I just tried various window widths on my 800x600 display but can't get
even a noticeable spike on my ProcExp display. I will test some more
and report if I find the infinite loop.

The thing with zooming is that the document height gets longer creating
a document overflow within the browser window content area, rendering
area. It forces a reflow and re-rendering of the page.

I've noticed that IE zooming is different to Opera. If I have a 600px
wide window in Opera and zoom to 200%, Opera treats it as a 300px window
and renders accordingly. With the same site ( http://phelum.net ) in
IE7 the page is always rendered for my 600px window but only a portion
is displayed when the zoom is 100%. Your scrollable div is different
because it is always displayed at the right-hand side of the window
which is not necessarily the right-hand edge of the page. Presumably
this is due to position:fixed. IE seems wrong though because the
contents of this div appear outside the div at zooms %100. I don't
think I like IE's implementation of zooming at all.

--
Steven