JS Stops working in IE6

VK wrote:

On May 18, 12:03 am, sheldonlg <sheldonlg> wrote:
Gregor Kofler wrote:
sheldonlg meinte:
sheldonlg wrote:
Here is an app: www.sheldonlg.com/JSstops.htm
I have the entire menuing removed, tags and css stuff, in
www.sheldonlg.com/JSstops2.htm.
This expands three levels, but not the fourth. Removing things
causes all four levels to appear.
Well I suppose it doesn't work in *any* browser. And "refreshView is
not defined". Apart from that: nuttin' to click.
In order to clear up confusion, the new app is atwww.sheldonlg.com/JSstop1.com

That one has no buttons or dropdown list. All it gas that is clickable
is the plus sign to expand the nested tables. With this app, it stops
after one expansion in IE6, but expands all three times in FF.

There is not JSstop1.html on your server. I see
http://www.sheldonlg.com/JSstops.htm
http://www.sheldonlg.com/JSstops2.htm
there and both seem working as intended with IE6

I also just test the JSstops1.htm on IE7. It expands two levels but not
the third. FF expands them all.

sheldonlg wrote:

VK wrote:
It works for me under IE6 / Windows XP SP2
What configuration are you using for test?

I am running XP SP2 and I have IE6 on it. It is version
6.0.2900.xpsp_sp2)070227-2254
cipher Strenght: 128-bit
Update Versions:; SP2;

I just called my son and had him try it on his machine that has IE6.
His expanded twice (mine does only once), but not the third time (to get
to the last, fourth, line). His version differs from mine in that after
the sp2 he has "_qfe" while I have "_gdr" (I just not is that my
previous reply was in error. It should have said:
6.0.2900.xpsp_sp2_gdr.070227-2254 )

On May 18, 12:34 am, sheldonlg <sheldonlg> wrote:

Is that a general rule to not have IE specific conditionals in CSS?

Of course not. IE specific conditionals in CSS are wrong not because
they are "theoretically wrong" or "should not be used" etc. They
simply do not work if they are not on HTML page. So your comments in
CSS file are not parsed and do nothing.

http://msdn.microsoft.com/en-us/library/bb250496.aspx

If you want IE-specific adjustments right in CSS table, use
expressions:

someElement {
width: 20em; /* default value */
width: expression(javascript code here) /* IE adjustment */
}

Thanks for the tip.

you are welcome

On May 18, 12:40 am, sheldonlg <sheldonlg> wrote:

I also just test the JSstops1.htm on IE7. It expands two levels but not
the third. FF expands them all.

OK, I think I know the problem. The table default display style
differs by browsers: "block", "table-block" and crazy mess. This is
why one _never_ _ever_ does two things with tables:

1) you do not override default display style.
2) you do not set it to any explicit values. Instead you set it to
"" (empty string) if needed to reset to default.

So in your code all initially visible tables must have no display rule
at all: carefully check CSS and scripting.

All initially hidden tables have display:none

To show a table you set display to empty string:
tableReference.style.display = '';

To hide it again you set to none:
tableReference.style.display = 'none';

VK wrote:

On May 18, 12:40 am, sheldonlg <sheldonlg> wrote:
I also just test the JSstops1.htm on IE7. It expands two levels but not
the third. FF expands them all.

OK, I think I know the problem. The table default display style
differs by browsers: "block", "table-block" and crazy mess. This is
why one _never_ _ever_ does two things with tables:

1) you do not override default display style.
2) you do not set it to any explicit values. Instead you set it to
"" (empty string) if needed to reset to default.

So in your code all initially visible tables must have no display rule
at all: carefully check CSS and scripting.

All initially hidden tables have display:none

To show a table you set display to empty string:
tableReference.style.display = '';

To hide it again you set to none:
tableReference.style.display = 'none';

I made that change and it didn't change anything.

On May 18, 1:33 am, sheldonlg <sheldonlg> wrote:

VK wrote:
On May 18, 12:40 am, sheldonlg <sheldonlg> wrote:
I also just test the JSstops1.htm on IE7. It expands two levels but not
the third. FF expands them all.

OK, I think I know the problem. The table default display style
differs by browsers: "block", "table-block" and crazy mess. This is
why one _never_ _ever_ does two things with tables:

1) you do not override default display style.
2) you do not set it to any explicit values. Instead you set it to
"" (empty string) if needed to reset to default.

So in your code all initially visible tables must have no display rule
at all: carefully check CSS and scripting.

All initially hidden tables have display:none

To show a table you set display to empty string:
tableReference.style.display = '';

To hide it again you set to none:
tableReference.style.display = 'none';

I made that change and it didn't change anything.

I really need to go now, sorry. I will take another look today evening.

VK wrote:

On May 18, 1:33 am, sheldonlg <sheldonlg> wrote:
VK wrote:
On May 18, 12:40 am, sheldonlg <sheldonlg> wrote:
I also just test the JSstops1.htm on IE7. It expands two levels but not
the third. FF expands them all.
OK, I think I know the problem. The table default display style
differs by browsers: "block", "table-block" and crazy mess. This is
why one _never_ _ever_ does two things with tables:
1) you do not override default display style.
2) you do not set it to any explicit values. Instead you set it to
"" (empty string) if needed to reset to default.
So in your code all initially visible tables must have no display rule
at all: carefully check CSS and scripting.
All initially hidden tables have display:none
To show a table you set display to empty string:
tableReference.style.display = '';
To hide it again you set to none:
tableReference.style.display = 'none';
I made that change and it didn't change anything.

I really need to go now, sorry. I will take another look today evening.

Thank you.

On May 18, 1:54 am, sheldonlg <sheldonlg> wrote:

VK wrote:
On May 18, 1:33 am, sheldonlg <sheldonlg> wrote:
VK wrote:
On May 18, 12:40 am, sheldonlg <sheldonlg> wrote:
I also just test the JSstops1.htm on IE7. It expands two levels but not
the third. FF expands them all.
OK, I think I know the problem. The table default display style
differs by browsers: "block", "table-block" and crazy mess. This is
why one _never_ _ever_ does two things with tables:
1) you do not override default display style.
2) you do not set it to any explicit values. Instead you set it to
"" (empty string) if needed to reset to default.
So in your code all initially visible tables must have no display rule
at all: carefully check CSS and scripting.
All initially hidden tables have display:none
To show a table you set display to empty string:
tableReference.style.display = '';
To hide it again you set to none:
tableReference.style.display = 'none';
I made that change and it didn't change anything.

I really need to go now, sorry. I will take another look today evening.

Seems like some broken table glitch on IE6. As you may notice the
third table is blocked for user interaction, see the cursor style or
try to set a simple alert on image click.
If we replace

<td colspan="15">
<table id="contract_900135_01-Jan-2008_both_all" ...

to

<td>
<table id="contract_900135_01-Jan-2008_both_all" ...

then the things are coming back to life right away.
Are you sure that there are indeed exactly 15 cells to span?

VK wrote:

On May 18, 1:54 am, sheldonlg <sheldonlg> wrote:
VK wrote:
On May 18, 1:33 am, sheldonlg <sheldonlg> wrote:
VK wrote:
On May 18, 12:40 am, sheldonlg <sheldonlg> wrote:
I also just test the JSstops1.htm on IE7. It expands two levels but not
the third. FF expands them all.
OK, I think I know the problem. The table default display style
differs by browsers: "block", "table-block" and crazy mess. This is
why one _never_ _ever_ does two things with tables:
1) you do not override default display style.
2) you do not set it to any explicit values. Instead you set it to
"" (empty string) if needed to reset to default.
So in your code all initially visible tables must have no display rule
at all: carefully check CSS and scripting.
All initially hidden tables have display:none
To show a table you set display to empty string:
tableReference.style.display = '';
To hide it again you set to none:
tableReference.style.display = 'none';
I made that change and it didn't change anything.
I really need to go now, sorry. I will take another look today evening.

Seems like some broken table glitch on IE6. As you may notice the
third table is blocked for user interaction, see the cursor style or
try to set a simple alert on image click.
If we replace

td colspan="15"
table id="contract_900135_01-Jan-2008_both_all" ...

to

td
table id="contract_900135_01-Jan-2008_both_all" ...

then the things are coming back to life right away.
Are you sure that there are indeed exactly 15 cells to span?

No, there aren't. I have always used a large number when I wanted to
span all columns of a table and it has always worked in the past. Using
the large number avoided the necessity of going back and changing things
if a column was added or removed later on in the development cycle.

I changed it to the exact number of columns in the containing table, and
it made no difference. I can't simply remove it (as you indicated
above) because that breaks the formatting and tries to put the inner
table into the first <td> position of the containing table. However, as
you said above, doing that causes all the expansions to occur.

By the way, in the past (I have been wrestling with this for weeks now)
I did simply put an alert there as you suggested. It didn't alert.
Also, when I did did that, as now, the cursor style didn't change. It
is not recognizing the mouse over.

As you may have noticed, what I am trying to do with this sample app is
not to "get the sample app to work". Rather, I am trying to get at the
root cause of the problem so that I can get my real app to work. That
is why I have left the sample app at the minimum I can get to and still
have it exhibit the broken behavior. Note that if I remove the menuing
lines and all the menuing information from the CSS, that the tables
expand one additional level, but it still breaks as it does not expand
to the last (pink background, no "+") table. So, it is not the menuing.
Rather, I has to be something, as you also have surmised, in the table
stuff.

Are you sure that there are indeed exactly 15 cells to span?

No, there aren't. I have always used a large number when I wanted to
span all columns of a table and it has always worked in the past.

That's a nasty way of doing things, really. But whatever.

As you may have noticed, what I am trying to do with this sample app is
not to "get the sample app to work". Rather, I am trying to get at the
root cause of the problem so that I can get my real app to work. That
is why I have left the sample app at the minimum I can get to and still
have it exhibit the broken behavior.

No, it is not a minimum case: there is still plenty of extra stuff
floating around. For example do not use the current CSS file, just
load the page w/o styling: and the problem goes away. So either
something sneaky in your CSS rules, or some particular CSS rule acting
strange for IE6 with malformed tables. Keep cleaning up. For the
starter make a page loading an empty CSS file and then start adding
ruleset by ruleset from the current CSS checking the behavior after
each modification.