Closed Bug 146851 Opened 23 years ago Closed 22 years ago

aftonbladet.se - Text sometimes crosses the border on the page and a reload can cure it

Categories

(Tech Evangelism Graveyard :: Other, defect, P3)

defect

Tracking

(Not tracked)

VERIFIED FIXED
Future

People

(Reporter: andre.bugs2, Unassigned)

References

()

Details

(Keywords: testcase)

Attachments

(9 files, 4 obsolete files)

The subject is very vague, but that is because this bug only happens sometimes, and it is therefore hard to figure out what is wrong. Sometimes, but far from always, the main page at aftonbladet.se is layed out badly, with text crossing the border that surrounds it. On these occasions it often helps to reload the page, or to leave it and then go back to it right after. I will attach two screenshot that shows first how the bad layout can look, and how the page looks after a reload just seconds later.
I should add that I have been seeing this bug on this page for as long as I can remember with mozilla. The screenshot was taken with a build from 2002-05-23-01.
I can also reproduce this if I save the page locally and load it in mozilla.
repeatedly reported - and repeatedly resolved as wfm or even invalid (?), this is a bug that comes and goes: layout problem at same site reported in: bug 62091, bug 100473, bug 134091, bug 137927
> repeatedly reported - and repeatedly resolved as wfm or even invalid (?), That's interesting. It seams like the problem has never really been investigated throughly, just resolved because a few people can't see the problem immediately. Like I said: this happens just once in a while, but is NOT related to a change on the website itself. A reload often cures it. There IS something fishy in Mozilla that causes this. To reproduce it you basically need to surf to aftonbladet.se once in a while, and eventually it will happen to you. Aftonbladet.se is the most visited site in sweden with 3.6 million unique hits per month (25 million hits in total), so this is a rather major site. When this happens it looks as if mozilla doesn't do that last bit of correction to the layout that it normally does. It normally moves things around a bit as it renderes a page, but on aftonbladet.se it sometimes doesn't do that last correction. A reproducable testcase is basically impossible to find. Some developer just needs to place aftonbladet.se in their daily browse path:-)
Priority: -- → P3
Changing QA Contact
QA Contact: petersen → moied
I can confirm this bug. It must be Mozilla:s fault because it works on reload. Here's another site with the same problem which looks **** everytime unless you press "reload". http://www.d.kth.se/arskurs/d01/
*** Bug 148568 has been marked as a duplicate of this bug. ***
I can confirm that this also happens on http://ps2.ign.com/ and I've even managed to see the problem on http://www.mozillaquest.com (not that I ever visit that site!) This is *definitely* a bug in Mozilla, not site specific. It seems to be related to tables.
-> Karnaze
Assignee: attinasi → karnaze
Target Milestone: --- → Future
Also seing this on Windows 2000 and Mac OS 9. Platform and OS should change to All/All. Today the actually have a page were they talk about Mozilla (for those who speaks swedish...): http://www.aftonbladet.se/vss/it/story/0,2789,172683,00.html This works fine in Netscape 4. Add 4xp in keywords?
all/all.
OS: Linux → All
Hardware: PC → All
It's not uncommon that it renderes this badly. This time it doesn't actually help to reload the page, and it has been rendered this badly for at least two days for me.
*** Bug 153205 has been marked as a duplicate of this bug. ***
I do not know which code that is used in different version's of the netscape/mozilla browsers. I have this problem in mozilla 1.0 also (not as bad as in 1.1) . Nescape 6.1 6.2 did similar error's on www.kuriren.nu (also a swedish paper) Mozilla handles kuriren.nu correct.
Can someone please create a minimized testcase that shows the problem ?
I think this bug is fixed in Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.1a+) Gecko/2002062508 . Reloaded 2002061104 and the bug is there but it is not present in the 2002062508 version. I also had a similar bug with the columns on http://java.sun.com/j2se/1.4/download.html that exists in the older vesion but not in the newer. I would like to change the status to resolved on this bug.
S-O Westberg, please see comment 5 in this bug. Lets leave this bug open for quiet a while now, and let people truly verify that it does not appear, ever, for say a month. This bug is *extremely* random, and I haven't seen it with a bulid from the 1.0 branch for a few days now. But from comment 14 you can see that I did see this with a build from the 1.0 branch just a few weeks ago. I will be very happy to resolve this worksforme myself if this continues for everyone for a while, but based on this bugs history I'm not at all confident it's fixed yet.
Undoubtly, this looks very promising. I now tried a build from 2002-06-10 on Linux, and the bug is visible. Re-did the test with a current built on the 1.0 branch and the bug is not there. I'm convinced that this bug is indeed fixed, but rest assured I will re-open it if this ever comes back again:-) Thanks to the unknown mozilla coder who fixed this! I will now, out of pure curiosity, look in bonsai to see if I can see a recent checkin that might have fixed this.
Status: NEW → RESOLVED
Closed: 23 years ago
Resolution: --- → WORKSFORME
Bug visible using build 20020625 on Windows 2000. Arent this bug reports for the trunk and not the branch? Resolving this as WFM will not make this bug to be fixed on the trunk, which it should be. Suggesting changing status to REOPENED.
I figured this would happen...
Status: RESOLVED → REOPENED
Resolution: WORKSFORME → ---
To comment 20: Please let us know if you find any checkin that is related to this. I find it surprising that such a strange bug just get fixed without anyone knowing who did it :)
It is not all that odd bugs get fixed without that anyone know why. I think this bug is an "overwrite of a variable/constant" or an uninitilised variable. This types of bugs comes and goes and are very hard to trackdown. To test if this is the case I would change the linking order to see what happens add some dummy variables and see if the error is changed.
This bug is now worse than ever. Using build 2002072408 (Win 2000) and now the pages looks totally weird. They are *extremely* long, you can scroll forever. Still looks completely normal in IE. Is this a regression of some sort??
This problem is still here in mozilla 1.1 (linux build 20020827).
Attached file Testcase (obsolete) —
Testcase extracted from aftonbladet.se
I've created a testcase that seems to show the problem every time. On the other hand, this is an evil bug that have been here for ever so i'm not surprised if it works for some of you :-) For me the problem is only visible the first time when you load the page, so close down mozilla and open the testcase. This is a bit strange since I frequently see the bug at http://www.aftonbladet.se and I never close down mozilla. On the webpage a reload does not always fix the problem while on the testcase it seems like a reload always works. I've tried it on mozilla 1.1 using linux, I hope this testcase works for you other too (and most important for a mozilla developer). Please try.
Very good testcase! I am able to reproduce it every time. Hopefully, we are now able to fix this problem.
About 95% sure this is a dup, please verify when bug 159363 is checked in. *** This bug has been marked as a duplicate of 159363 ***
Status: REOPENED → RESOLVED
Closed: 23 years ago23 years ago
Resolution: --- → DUPLICATE
Status: RESOLVED → REOPENED
Resolution: DUPLICATE → ---
I applied the patch in attachment 98341 [details] [diff] [review] in bug 159363 and the problem is unfortunately still present at aftonbladet.se. REOPENING.
*** Bug 148568 has been marked as a duplicate of this bug. ***
assign it to me.
Assignee: karnaze → jerry.tan
Status: REOPENED → NEW
I believe this is a dup of bug 156629. also an image with align value inside table. *** This bug has been marked as a duplicate of 156629 ***
Status: NEW → RESOLVED
Closed: 23 years ago23 years ago
Resolution: --- → DUPLICATE
I do not think it is a duplicate of bug 156629. There is one big difference: The URL in bug 156629 is OK when loaded, reload makes it wrong, shift reload makes it right again. In my testing this was 100% reproducible. The URL in this bug has the opposite behaviour, wrong layout on first reload and after shift-reload, fine with reload. I think this bug should be reopened and kept open until a fix is checked in for bug 156629, if this also fixes this bug then make it a duplicate.
I compared the two bugs's testcase, they are very samilar. they both have one image inside one table, that image has align attribute, they both give different layout when reloading and shift-reloading. if remove align attri, they both works fine. So I think they are dup. as which layout is good or wrong, it depends the html designer. Anyway, anyone who fix bug 156629 in the future should check its dup bugs to make sure they are dup.
I'm going to reopen this and add a dependency on bug 156629. This bug has been resolved before and a similar bug has always been filed on a later date, so I would like to keep this open just to avoid any further duplicates. I will make sure to dupe this one against 156629 if the checkin for that bug cures this one as well.
Status: RESOLVED → REOPENED
Depends on: 156629
Resolution: DUPLICATE → ---
*** Bug 166831 has been marked as a duplicate of this bug. ***
Attached file another testcase
more precise testcase, there are two table in it, one is ok, another is bad, the only difference is that one set image's width 1px less than another.
block 0412A080 r=1 a=4452,UC c=4452,UC cnt=2191 text 04162890 r=2 a=UC,UC c=UC,UC cnt=2192 text 04162890 d=0,0 me=0 place 0409C5E8 r=2 a=UC,UC c=UC,UC cnt=2193 place 0409C5E8 d=0,0 me=0 img 0409C528 r=4 a=UC,UC c=3744,UC cnt=2194 img 0409C528 d=3744,361 me=3744 m=3744 text 0412A1EC r=2 a=UC,361 c=UC,UC cnt=2195 text 0412A1EC d=732,228 me=684 text 04162890 r=2 a=4452,UC c=UC,UC cnt=2196 text 04162890 d=0,0 place 0409C5E8 r=2 a=4452,UC c=UC,UC cnt=2197 place 0409C5E8 d=0,0 img 0409C528 r=2 a=4452,UC c=3744,UC cnt=2198 img 0409C528 d=3744,361 text 0412A1EC r=2 a=672,361 c=UC,UC cnt=2199 text 0412A1EC d=684,228 status=1 text 0412A1EC r=2 a=672,UC c=UC,UC cnt=2200 text 0412A1EC d=684,228 me=684 text 0412A1EC r=2 a=4452,UC c=UC,UC cnt=2201 text 0412A1EC d=732,228 me=684 block 0412A080 d=4464,601 me=4464 m=684 during incremental reflow, block get d=4464 bigger than 4452,this is the problem. change component to block and inline
Component: Layout → Layout: Block & Inline
add nsbeta1 keyword
Keywords: nsbeta1
This also happends with IE 6 SP1 sometimes. Maybe it is Aftonbladets HTML code that sucks?
It's both. There is a bug in Gecko though.
I've found a page (which I've stripped down a bit) that has this bug every time. Steps to reproduce: 1. Load the testcase. Notice how the page sucks. 2. Press Ctrl++ to zoom in, and then Ctrl+0 to restore zoom level 3. Repeat step 2 a few times and notice how the right "menu" on the page moves around.
Based on the testcases I believe this is a duplicate of bug 186593. Could anyone confirm that this bug no longer appears in builds on 2003-02-23 or later? If so, could you mark it a duplicate of bug 186593?
Depends on: 186593
I can no longer reproduce my 100% reproducible testcase in comment 44! That's a very good sign.
Not fixed in 20030310-1.3, but seems to be fixed in 2003031108-trunk.
Now my testcase renders even worse results. The length of the document now differs greatly when following the steps to reproduce: Steps to reproduce: 1. http://bugzilla.mozilla.org/attachment.cgi?id=114886&action=view 2. Press Ctrl++ to zoom in, and then Ctrl+0 to restore zoom level 3. Repeat step 2 a few times and notice how the right "menu" on the page moves around AND how the length of the document changes.
I have the same problem on another page, I can solve it by doing a CTRL + + and then CTRL + 0, but when I refresh the page again, the problem is back... and I have to repeat the fix-procedure again.... I actually produced this bug on a page by just keep pressing F5 (was debugging) and suddenly the bug was on that page, and now I can't get rid of it... Using: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4a) Gecko/20030320 Phoenix/0.5
Shouldn't this bug be marked fixed? The site renders fine now since bug bug 186593 was fixed. The issue mentioned in comment 48 - 49 is not really the same issue.
No, I've still seen this bug happen, although it didn't look as bad as before. The normal "reload fixed it" applied.
I still have the problem, *nothing* is fixed here... Although my problem is not the fontsize, it is based on the same problem, the page loads like it's an end-less loop going on, hundreds of pages containging generated because of some element being extended vertically... it is fixable, like i mentioned in Comment 49, by using CTRL + + | CTRL + 0, but still very VERY annoying Should I file this a a seperate bug, I know David Tenser has the same problem as I do, and so he directed my attention to this bug. This is defenetly something that could make me drop Phoenix, or maybe just stay with an older build, since the new ones contain this bug that makes some sites un-viewable. Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4a) Gecko/20030321 Phoenix/0.5
I can confirm Comment 48, since it is bassically my problem, the page becomes huge in height... should this be transfered to another bug report? Or is it already reported? This is a huge bug in Gecko
Also visible on http://home.earthlink.net/~nilsoncain sometimes. I use no tables in this site; all layout is done using valid css and XHTML 1.0 Strict, so not a coding problem on my site. I am using March 19th Phoenix build.
aftonbladet.se locks fine to me since bug 186593 was fixed. Since this is not a meta bug for reload-related layout bugs, I am marking this as a dupe. Problems with other pages than aftonbladet.se should be filed as separate bugs, but search bugzilla throughly first, several such bugs already exist. Please only reopen if this problem reapears on aftonbladet.se. *** This bug has been marked as a duplicate of 186593 ***
Status: REOPENED → RESOLVED
Closed: 23 years ago23 years ago
Resolution: --- → DUPLICATE
My comment 51 was seen on Aftonbladet.
Status: RESOLVED → REOPENED
Resolution: DUPLICATE → ---
If you see the problem, please attach a screenshot of what the problem is in current builds.
Since the image is to large to be attached, I will post the URL http://pub.tsn.dk/bugs/phoenix_146851_01.jpg Please note the HUGE scroll bar. What you see in the bottom (the simple yellow-black-white background) goes on-and-on for hundreds of pages, until finally it displays the bottom of the site, all other HTML is rendered normally Produced as described in Comment 48
I dont know if its me being blind, but that screenshot has nothing to do with this bug. This bug is about the text overlapping other text in the middle section. Since bug 186593 was checked in I have not seen this issue again. But now this bug seems to have become a meta bug for various bugs (scrollbar issue, maximizing text etc).
José is right, please try to keep this bug about the original layout problem. I have on one occasion seen a mild form of this layout bug happen again on a build which had the fix checked in, which is why I would like to keep this bug open for a a little while longer. Bugs about the layout issues on aftonbladet have been resolved as worksforme so many times before because the bug has temporarily vanished, just to come back later.
Strike my comments in comment 48 about the page getting very long. That was in fact a problem with the graphics drivers, and reinstalling Windows solved that. Also, "100% reproducible testcase" only generates the original error in Mozilla 1.3 final, not in the latest nightly. So maybe this bug is fixed after all?
Re: Comment 61 Indeed, a new set of NVIDIA drivers fixed my problem as well... Sorry for all the problems regarding this bug, it should however be known to others who get this type of bug that a newer version of your Graphics Drivers could solve the problem
Blocks: 200047
David: > Strike my comments in comment 48 about the page getting > very long. That was in fact a problem with the graphics > drivers, and reinstalling Windows solved that. happen to know what bug # that might be? (see bug 194254)
[Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4a) Gecko/20030327 Phoenix/0.5] As far as I know, the build above should have the fix that people hoped would fix this bug. Unfortunately, today I ran into the poor layout that can be seen on the attached screenshot. After a reload the layout was good.
Attachment #119608 - Attachment description: Screenshot showing very bad first-time rendernig on new build → Screenshot showing very bad first-time rendering on new build
I notice that in the test case (attachment 114886 [details]), there are 200+ images with wrong filenames (.fig instead of .gif). I've corrected them all in this attachment. Reporter (André Dahlqvist) and everyone else who has trouble w/ the site, can you try both this attachment and attachment 114886 [details] and see if there is any difference.
Attachment #98449 - Attachment is obsolete: true
This bug still exists in 1.3 . I think it is more common in 1.3 then in 1.2.1 and today it does not go away with reload. http://www.aftonbladet.se/vss/nyheter/story/0,2789,291047,00.html
Re comment 66: The fix for bug 186593 is _not_ in 1.3, please test with 1.4a or a current nightly.
adt: nsbeta1-
Keywords: nsbeta1nsbeta1-
Happened today with 2003-04-25 of Mozilla Firebird.
André, am I correct in assuming that you only see this with Phoenix/Firebird and not "vanilla" Mozilla?
This seems to be fixed in 1.4beta (I saw this a lot in 1.3).
I'm still seeing it occasionally with 2003-05-05 Mozilla Firebird. Torben, since I am basically only running Mozilla Firebird these days, that's what I notice it with. It's possible that this only happens on Mozilla Firebird (maybe because it has a different delay time until rendering starts?), but I am not sure about that.
Happened again with plain Mozilla 2003-06-06, from the 1.4-branch.
can anyone who experience this problem produce a reduced testcase (see http://www.mozilla.org/newlayout/bugathon.html#testcase) ?
Keywords: qawanted
background image for a testcase
A flash banner needed for a testcase.
This is a very reduced testcase which should be 100% reproducable. The problem occurrs at least with Mozilla 1.4 and a recent nightly of Firebird (tested on windows2000). The problem seems to be related to the width and hight of the flash banner. The scripts below is what I found was causing the problem. <script language="JavaScript"> document.write('<OBJECT classid="clsid:D27CDB6E-AE6D-11cf-96B8-444553540000" ID=banner WIDTH="250" HEIGHT="120">'); document.write('<PARAM NAME="movie" VALUE="http://bugzilla.mozilla.org/attachment.cgi?id=130246&action=view">'); // remove the width and height in the next line and the bug dissappears document.write('<EMBED SRC="http://bugzilla.mozilla.org/attachment.cgi?id=130246&action=view" WIDTH="468" HEIGHT="60" ></EMBED>'); document.write('</OBJECT>'); </script> This was the first time a had this problem with 1.4, in 1.3 it was _very_ common. The bug wasn't visible anymore after a reload, but having it saved on local disk it seems to be present all the time. I think the bug didn't show up on reload since then it was anther set of banners there.
Tried the testcase, and it does look similar to how this bug usually presents itself. However, I tried the same with Opera 7.11 on Linux, and it also rendered the page with an overflow.
The overlap is present in opera 7.11 on windows also, both opera and mozilla uses the <embed> but IE just looks at the <object> -parameters. Moz supports both (http://www.mozilla.org/newlayout/testcases/printing/misc/embed_obj_tag.html) but not the same way is IE. So basically it comes down to en error in aftonbladet's coding?
> So basically it comes down to en error in aftonbladet's coding? seems so to me. Magnus Melin, thanks for the testcase -> TE
Assignee: jerry.tan → spanish
Status: REOPENED → NEW
Component: Layout: Block & Inline → Spanish
Keywords: qawantedtestcase
Product: Browser → Tech Evangelism
QA Contact: moied → spanish
Version: Trunk → unspecified
-> Swedish not Spanish
Component: Spanish → Other
Havent seen this bug in ages. Visit Aftonbladet every day. Changing from spanish@evangelism.bug...
Assignee: spanish → other
QA Contact: spanish → other
I see it almost every time I visit Aftonbladet and have done so for a couple of weeks (and builds).
Yeah, it has entered one of those "bad periods" again where it happens almost all the time. Can anyone who has a good understanding of the bug contact Aftonbladet at webmaster@aftonbladet.se and try to work with them to reach a solution? I currently do not have the time to do it, nor do I know the exact code bug.
From Catarina Smedshammar at aftonbladet.se, at 2003-11-28 14:09: > Hej David > Vi ger upp den gamla risiga html-koden, och släpper en helt ny (och > förhoppningsvis felfri) sida i nästa vecka. Då ska vår sida se ok ut även > i Mozilla - hoppas du har tålamod till dess! This means that they'll probably change the html code soon which should solve the problem. However, that doesn't mean the alignment bug in Gecko is automatically fixed, but aftonbladet.se will render properly.
Maybe the alignment bug is not fixed, but its great news that they are redesigning, cause usually when you recommend Mozilla to anyone (in sweden) they complain about aftonbladet...
The redesign of aftonbladet.se is complete. The site now renders without problems. I'm not sure what to do with this bug though, since it's about a gecko bug that still exists.
It might be complete, but it still have prolems. Yesterday the blocks of the page was layed out badly, it looked like blocks that should have been next to each other was under each other. When I reloaded the page it looked fine again. This happend twice yesterday. So the text is not overlapping as before, but it's still broken. Reloading a page should not render it differently, no matter what.
They had problem with their ISP yesterday. It was also stated on their web site for a while. So it was probably that.
Unfortunately, I see the text overlapping problem mentioned in comment 88 too. First the frontpage renders fine, but a reload (or shift-reload) later there is text on top of one of the images. I saved the HTML of the badly layed out version and the correctly layed out version, and the only difference between them (apart from a comment from their publishing system, which puts in the current date) is this: Correct rendering ----------------- <map name="mellanplus4"> <area shape="rect" coords="2,2,118,43" href="http://plus.aftonbladet.se/matvin/"> <area shape="rect" coords="123,2,243,43" href="http://plus.aftonbladet.se/tvtabla/"> <area shape="rect" coords="248,2,373,43" href="http://plus.aftonbladet.se/dujag/"> Bad rendering ------------- <img src="http://www.aftonbladet.se/annons/plus/plus/mitten375_2.gif" width="375" height="45" border="0" usemap="#mellanplus2"> <map name="mellanplus2"> <area shape="rect" coords="2,2,118,43" href="http://plus.aftonbladet.se/plusmail/"> <area shape="rect" coords="123,2,243,43" href="http://plus.aftonbladet.se/webbspel/"> <area shape="rect" coords="248,2,373,43" href="http://plus.aftonbladet.se/spotlight/"> So, the cause for bad rendering once again seams to be an ad.
How about adding a testcase?
Attachment #139335 - Attachment description: html page that shows text on an image → testcase that shows text on an image
I can reproduce last added testcase, reloading fixes the problem. I guess this shouldnt be tech envagelism then?
The original issue for this bug seems to be gone since aftnobladet started using thier new code... Although it might have been an evangelism bug at some point, I agree its not anymore. Also, most (all?) old attachemnts seem obsolete. The testcase I created (comment #93) shows the text overlapping an image at first, but after reload it displays correctly. Sometimes it does not even need a reload, changing getween tabs may also help... So clearly it's something with moz. Clear your cache to make sure you can reproduce, then it works always for me atleast.
Attachment #84943 - Attachment is obsolete: true
Attachment #84945 - Attachment is obsolete: true
Attachment #87726 - Attachment is obsolete: true
Yes, this has somehow becomed a "all bugs on aftonbladet" bug. The original problem that occured with the old design is gone (fixed?). I suggest closing this one and opening a new one. This is already very confusing.
I'm resolving this bug as fixed, since their redesign fixed the original issue. Someone please file the remaining problems (which I see too) in a new bug.
Status: NEW → RESOLVED
Closed: 23 years ago22 years ago
Resolution: --- → FIXED
Remaining problem filed in bug 231498.
Status: RESOLVED → VERIFIED
Product: Tech Evangelism → Tech Evangelism Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: