...http://www.amnet.com.au/ --it's my ISP

To reproduce this, wave the cursor around over the page, then at top left click "Residential". This is essentially reloading the page. KM160b25 will freeze solid on the first or second click, but Task Manager shows it as "running", and Resource Monitor shows elevated CPU and Memory usage.

I have not noticed any problems with any other part of the Amnet site.

The only way to recover is to kill the process with TM, but when I restart KM160b25 it does not complain about being terminated improperly or ask if I want to use the previous session.

This is under W7HPx64 SP1, Classic Start Menu 3.6.1 based on Windows Classic Desktop, and SuRun 1.2.1.0.

Gordon.

____________________
Understanding the scope of the problem is the first step on the path to true panic. [Florence Ambrose, "Freefall" 01372 January 22, 2007 http://freefall.purrsia.com/ff1400/fv01372.htm]

It must be your setup. I just went to the site and had no problem. I am away from home and my laptop has XP SP3.

Yah. I thought so too. Except it doesn't reproduce at all with b2.0.

Having said that, I'm sure the two setups are different. I created brand new profiles for b2.5, but did try to get the same functionality, so I could run them side-by-side to suss whether b2.5 was worth the effort.

Right now I have no idea what could trigger this behaviour, so I can't think of any other sites to try it out.

Gordon.

____________________
Understanding the scope of the problem is the first step on the path to true panic. [Florence Ambrose, "Freefall" 01372 January 22, 2007 http://freefall.purrsia.com/ff1400/fv01372.htm]

But, does it happen every time?

Because the behavior you describe is quite common and quite random.

If it doesn't happen every time, as I said, is a "normal" behavior.

I suffer it sometimes myself. Usually with pages with high load of CSS and Javascript together.

I am back home now and using my desktop to view the site in question. I have had no trouble seeing and using the site. I have been able to click on items and go to other pages. I am now on the "let's get your number" page. I can freely move between tabs. I am using beta 2.4 but it has the same engine as 2.5 The only difference between the two versions is the method for calling spellcheck.

I do see the memory for KM constantly changing in a range from 37,000 to 40,000. On my machine that is not a problem as memory usage is only 29%.

There seems to be a lot of errors in the error console. I don't worry much about CSS but there are sever and XML errors there.

I do not know why your setup will not run the page. It ran on my laptop with only 1 gig of memory in XP. I loaded and ran a beta 2.5 there to test.

Hanlon’s razor is an eponymous adage named after Robert J. Hanlon that states: “Never attribute to malice that which is adequately explained by stupidity.”

JamesD

Quote
JohnHell
But, does it happen every time?

Yes, every time I click (first or second time) that link with the b2.5 version. So it's "not normal".

The page itself is the residential page, so clicking the "Residential" link was unnecessary, which is why I have not seen this on any other site. Also, K-M of any version has never frozen like this before--I would expect a {car+tree=pile-up} crash rather than hanging. I also expect K-M to complain bitterly on restart...

Quote
JohnHell
Because the behavior you describe is quite common and quite random.

I have never noticed this behaviour before, and I doubt I'll see it again, because I don't know what indicators to look for. I did run the page thru the W3C validator, and like most pages, it's a coding catastrophe and CSS disaster--fairly normal in other words. I know K-M in some settings is error-intolerant, but if this behaviour is "quite common", I would be disturbed. Random behaviour... OK, it can seem that way, if we cannot pin down the cause.

At the moment, I'll go with James D on my b2.5 settings. K-M is so configurable that we can easy lose sight of apparently inconsequential tweaks, even forget we made them. And it's very difficult to keep a list of changes away from bog-standard.

Gordon.

____________________
Understanding the scope of the problem is the first step on the path to true panic. [Florence Ambrose, "Freefall" 01372 January 22, 2007 http://freefall.purrsia.com/ff1400/fv01372.htm]

I think I isolated, in my system at least, to cookies behavior, that, believe it or not, at least with my config (ask for cookie permissions), it is quite common too.

If you disable javascript (and therefore cookies) or block cookies, the site loads without problems.

If you set cookies to accept cookies without asking (I tried session), it loads without problems.

I'm suffering the behavior I told time ago because of the crazy way K-meleon 1.6x manage cookies.

(not from your site)

Asks for all together instead go one by one, creating a window for each cookie and close it before create the next.

I'm not saying it is your problem (obviously, it isn't), but it is very common to see the behavior you described, and, as I said, I suffer it a lot.

Hope you can fix it.

I think Johnhell is right. It is a problem with certain settings regarding cookies. I had not myself ever use the permissions for each cookie before. I tried with that on and the second cookie stopped KM. I had to use task manager to get out. I did get the message about KM not ending right upon my next startup.

I am not sure if I can do anything about the problem. I did not write the code. I just put pieces together as I found them. I will have a look and see if I can modify any behavior regarding cookies when asking for permission for each cookie.

Hanlon’s razor is an eponymous adage named after Robert J. Hanlon that states: “Never attribute to malice that which is adequately explained by stupidity.”

JamesD

To be fair with the "bug", sometimes, after an uncertain amount of time, K-meleon unfreezes and you can recover the control, but are few, few, few times. And I'm talking of a few minutes and after trying to close the origin window without close the whole K-meleon.

I don't know if you could, I don't bet for it, to be fair, but this is one of the annoyances I talked about when I was switching from 1.1x to 1.6x (I didn't find the topic before post the previous message):

http://kmeleonbrowser.org/forum/read.php?1,118723,118960#msg-118960

I think this was a problem with the changes from 1.1x and 1.5x to the renewed 1.6x, so must be in the C code. (And I don't know if happened to 1.5x, because I never used, beyond test)

If you can without doing any further base code edit and compilation, you'll fix the ultimate crazy "bug".

And I'll thank you a lot, because it's the only thing, beyond render issues for the reason we all know, I suffer as a nightmare :/

Quote
JamesD
It is a problem with certain settings regarding cookies.

Well, I know K-M roars to a halt if you try "Tools>Privacy>View Data>View Cookies" and one or more cookies has an expiry date longer than 10 digits... Dunno what upper value a 10-digit date can have, eg: 9999999999? None of mine have much larger than 4100000000.

SQLite 2.0 b1 is a wonderful program!

I gave up trying to set individual permissions for cookies, too much trouble ¬_¬ and I'm too lazy I simply fix it in hosts. eDexter answers on behalf of the blocked site, with a paw-print Most of my sites I always accept cookies and keep them until they time-expire. Only Opera has good cookie mangement--it's too bad I don't like Opera!

Gordon.

____________________
Understanding the scope of the problem is the first step on the path to true panic. [Florence Ambrose, "Freefall" 01372 January 22, 2007 http://freefall.purrsia.com/ff1400/fv01372.htm]

Quote
gordon451
Well, I know K-M roars to a halt if you try 'Tools>Privacy>View Data>View Cookies' and one or more cookies has an expiry date longer than 10 digits... Dunno what upper value a 10-digit date can have, eg: 9999999999? None of mine have much larger than 4100000000.

Thought it was already fixed. Line 70 in sources should be changed.
Date values > 9999999999 are invalid and cause 'roaring' overflows.
This should be changed if someone orders re-compiled binaries.

Quote
deadlock
Thought it was already fixed. Line 70 in sources should be changed.

We-e-ell, "Fri Feb 24 17:55:34 2006 UTC (7 years, 2 months ago) by boisso"...

    Line 71:       char fDate[128];

Maybe not.

What should it be changed to? (Not that I'm about to do it I gave up programming decades ago )

Gordon.

____________________
Understanding the scope of the problem is the first step on the path to true panic. [Florence Ambrose, "Freefall" 01372 January 22, 2007 http://freefall.purrsia.com/ff1400/fv01372.htm]

@gordon451
ndebord's cookies.sqlite sample file with large dates is no longer online, but it did work with a version using this code in line 70:

if ((expires>0)&&(expires<9999999999)){

Did you find a web page that sends bad cookies?
Since 160b25 still uses old 1.6b2 binaries, you should replace them with last years update and test again.

Quote
deadlock
using this code in line 70: if ((expires>0)&&(expires<9999999999)){

Oops. :O Please don't ask me what Freudian impulse led me to Line 71. :s

Unfortunately, I no longer have any web pages with bad cookies. I cleaned the cookies file a couple of days ago...

Ummm... with KMeleon1.6.7z, do I extract to a folder and run the KM exe? Or simply replace files in my current folder?

Gordon.

____________________
Understanding the scope of the problem is the first step on the path to true panic. [Florence Ambrose, "Freefall" 01372 January 22, 2007 http://freefall.purrsia.com/ff1400/fv01372.htm]

Quote
gordon451
Unfortunately, I no longer have any web pages with bad cookies. I cleaned the cookies file a couple of days ago...

Ummm... with KMeleon1.6.7z, do I extract to a folder

You should not change or update anything.
Just wait for the next bad cookie to continue testing.

The update came without information,
but you will find out how to use it.

Quote
deadlock

Quote
gordon451
Unfortunately, I no longer have any web pages with bad cookies.

Just wait for the next bad cookie to continue testing.

OK, I got one now. Download it from here in Dropbox. Problem, it doesn't crash K-M, b2 or b2.5 :s

Check out record 176, .sitepoint.com, 11 characters: 10008781749

Its expiry date is Wednesday 2 March 2287 @17:09...

Gordon.

____________________
Understanding the scope of the problem is the first step on the path to true panic. [Florence Ambrose, "Freefall" 01372 January 22, 2007 http://freefall.purrsia.com/ff1400/fv01372.htm]

Quote
gordon451
OK, I got one now.

Great, but you've got to continue searching,
because _gaos still uses a valid date.

Sat 20 Nov 2286 18:46 == 9999999999
Wed 02 Mar 2287 10:09 == ? (the _gaos cookie)
Tue 30 Jul 2318 21:33 == 11000000000

9999999999 is a valid date.
99999999999 is not a valid date.

I had chosen 9999999999 because it is already big enough.

You'll need a cookie that crashes KM 1.6b2,
to find out if it was fixed in the update.

Quote
deadlock
Great, but you've got to continue searching,
because _gaos still uses a valid date.

I had chosen 9999999999 because it is already big enough.

Ummm... After a binary progression upwards from a known safe number, I stopped at 32535193210, which KM tells me is 01-01-3001@01:40 I reckon that the cut-off is 31-12-3000@23:59 or maybe 01-01-3001@00:00 :cool: Off the top of my head, I'd say KM 1.6b2 has not been fixed.

I'm too lazy to work out the date-number system in use, and a life is calling me, so I'll leave it there at least for a day or so.

Gordon.

____________________
Understanding the scope of the problem is the first step on the path to true panic. [Florence Ambrose, "Freefall" 01372 January 22, 2007 http://freefall.purrsia.com/ff1400/fv01372.htm]



Edited 1 time(s). Last edit at 05/19/2013 06:57AM by gordon451.

@gordon451
That was close.
The longest accepted cookie expiration date is:
32535244799 == Thu 01 Jan 3001 08:59

Found ndebord's cookies.sqlite file. It contains two cookies using
253402300801 in expiration date.

This modification in Line 70 of Cookies.h protects the browser:
if ((expires>0)&&(expires<=32535244799))



Edited 1 time(s). Last edit at 05/19/2013 06:15PM by deadlock.