Closed Bug 140730 Opened 22 years ago Closed 22 years ago

Odd behaviour on page (appearing and disappearing)

Categories

(Tech Evangelism Graveyard :: English US, defect, P2)

defect

Tracking

(Not tracked)

RESOLVED INVALID

People

(Reporter: gabriel, Assigned: doronr)

References

()

Details

(Keywords: testcase)

Attachments

(2 files)

Linux 2002041711

Clicking on above link shows a blank page. If you keep clicking reload,
sometimes the page appears for a second and then disappears.
Also I just noticed that keyboard shortcuts do not work with the page loaded,
e.g. ctrl-T does not open a new tab.
-> LAyout and confirming with win2k build 20020427..
Assignee: Matti → attinasi
Status: UNCONFIRMED → NEW
Component: Browser-General → Layout
Ever confirmed: true
QA Contact: imajes-qa → petersen
confirming using build 2002042803 on Win2k.
Crashing when right clicking once page loaded (and has become blank): TB5699598Y.
Might be then dupe of bug 129980.
Keywords: crash, stackwanted
OS: Linux → All
Hardware: PC → All
Sev -> critical
Severity: major → critical
In case anyone was wondering, 216.239.39.100 is Google (not that that probably
makes a difference).
Also confirming, did not crash for me - other tabs were responsive.
Did not crash using 2002050108 build on WinXP, but it did display the strange
"appears when clicking but only for a moment" behavior
Priority: -- → P2
Target Milestone: --- → Future
Still crashing for me using build 2002050108 on Win2k.
Attached file Page source
View source of page.
Note that there is no </body> or </html> in the page. I wonder if that might be
significant ?
Using build 2002050904 on Win2k (trunk), I don't crash anymore since fix for bug
129980.
However, page still remain blank, bad HTML code: HTML code before doctype, no
</body>, no </html>... might be Tech Evangelism ?
IE6 also give a blank page if view source then go back to page.
The original page is ok, has proper HTML code.

Propose to move to TE because of google adding some code, and removing closing tags.
Severity: critical → normal
Keywords: crash, stackwanted
I just checked something - I see exactly the same appearing/disappearing
behaviour in Konqueror ! So there's presumably something broken with the page.

Agree some tech evangelism may be needed - I've never known google cache to
screw up before.
Updating product to Tech Evangelism. Please change the assignee if necessary.
Component: Layout → US General
Product: Browser → Tech Evangelism
Version: other → unspecified
`Bad HTML' does not mean this bug goes to Tech Evangelism. If that were the
case, most of the web would end up there. Let's continue to call this a layout
bug until we decide that the site is so broken that we ought not fix it. A
minimized testcase would help.
Component: US General → Layout
Product: Tech Evangelism → Browser
Version: unspecified → other
Let me be a little more clear. First we understand what the specific problem is,
_then_ we decide whether this is a technical issue or an evangelism issue.
Keywords: qawanted
My apologies for jumping the gun.
makingtest - deathwnd@yahoo.com
Attached file Testcase
Deleting any more code causes problem to disappear - Using Build ID: 2002061808
under Windows XP
Been playing around with this...adding the proper <HTML>,<BODY>,</BODY>, and
</HTML> tags don't fix the problem in the testcase.

Another interesting nugget is that doing View Source on the testcase in IE6
gives us:
<layer clip="470,62"
src="http://ad.doubleclick.net/adl/about.com/hobbies_chess;svc=;site=chess;kw=;chan=hobbies;syn=about;pos=slot1;sz=468x60;ord=10.770918768206975"
target="_top" visibility="hide"
onload="moveToAbsolute(adslot1.pageX,adslot1.pageY); visibility='show';"></layer>

Instead of the source. Odd.
Yet another oddity is that the non-google-cached version of this page displays
fine in Mozilla.

So I decided to compare the two around the area where the testcase code is.
There's:

<script language="JavaScript" src="http://f.about.com/z/js/cj19.js"></script>

in the google-cached, and

<script language="JavaScript" src="http://f.about.com/z/js/cj004.js"></script>

in the non-cached...quite a few variables change (in the preceding javascript
sections which I didn't copy over for reasons of clarity) and it's referring to
two different .js files.

Copying in the correct variable code from the line above the cj004.js file and
changing cj19.js to cj004.js in the testcase fixes the page in moz.
Interestingly enough, this _also_ fixes View Source under IE6 so it displays the
correct sourcecode. This surely points to code in cj19.js as the root of our
problems here.

In cj19.js there's a line:
document.writeln('<layer clip="',j,',',k,'" src',b,'adl',a,zat,'
visibility="hide" onload="moveToAbsolute(',u,'.pageX,',u,'.pageY);
visibility=\'show\';"></layer>')}

Which seems to be the _true_ root of our problems.

cj004.js has the same line just a little differently:
document.writeln('<layer clip="',j,',',k,'" src="',s,zat,'" visibility="hide"
onload="moveToAbsolute(',u,'.pageX,',u,'.pageY);visibility=\'show\';"></layer>')}

Seemed to me the cj19.js had the src= code written incorrectly so I stored it
locally and changed the link in the testcase to reflect this then hacked it
around a bit so the src= part of the document.writeln line looked more like it
does in cj004.js giving us:
document.writeln('<layer clip="',j,',',k,'" src="',b,adl,a,zat,'"
visibility="hide" onload="moveToAbsolute(',u,'.pageX,',u,'.pageY);
visibility=\'show\';"></layer>')}

Which, coincidentally, works perfectly. Mozilla doesn't go crazy when the .js is
modified to that code, and IE6 behaves normally when doing View Source. I have
this all zipped if someone needs it attached.

So, in conclusion, it seems to be a JS coding error on the part of a)
doubleclick or b) about.com. Somewhere along the line it got fixed, but the
Google cache doesn't reflect this (although, cj19 is still out there, could
still be on random banner ad rotation for about, a simple google search reveals
though that cj19.js only seems to reside on about.com). My suggestion for this
bug: make mozilla handle this error better. Could be Tech Evangelism though, as
it does deal partially with layers and bad code.

Sorry for the (exceedingly) long comment, and thanks for reading it all.  Anyone
else have any ideas? Now that you have "the facts," is this a technical issue or
an evangelism issue?
Good job !!
->Tech Evang. by botching that attribute, I think cj19.js is trying to pump
something *very* bizarre into the source...
Assignee: attinasi → doron
Component: Layout → US General
Keywords: qawantedtestcase
Product: Browser → Tech Evangelism
QA Contact: petersen → zach
Target Milestone: Future → ---
Version: other → unspecified
Keywords: evang500
not a clear evang issue->invalid
Status: NEW → RESOLVED
Closed: 22 years ago
Resolution: --- → INVALID
Product: Tech Evangelism → Tech Evangelism Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: