51972 - Mozilla gets into infinite redirect loop

Status: VERIFIED INVALID

(Core :: Networking, defect, P3)

Description

[J. Paul Reed [:preed]]

From Bugzilla Helper:
User-Agent: Mozilla/4.74 [en] (X11; U; Linux 2.2.16 i686; Nav)
BuildID:    2000090806

When loading the above page, mozilla gets itself into an infinite loop of
redirects; older versions of netscape render the page upon going to the above
URL.

Reproducible: Always
Steps to Reproduce:
1. Go to the above page...
2. Watch your webshells count yoyo; page never gets rendered.

Actual Results:  Mozilla gets caught up in redirects (I think), and sits there,
with its webshell count going up and down by three, never rendering the page...

Expected Results:  Should load the page...NS 4.74 has no problems doing so.

...general startup info...

Document about:blank loaded successfully
->>>>>>>>>>>>>> Write Clipboard to memory
->>>>>>>>>>>>>> Read Clipboard from memory
->>>>>>>>>>>>>> Write Clipboard to memory
WEBSHELL+ = 4
WEBSHELL+ = 5
WEBSHELL+ = 6
WEBSHELL+ = 7
WEBSHELL+ = 8
WEBSHELL- = 7
WEBSHELL- = 6
WEBSHELL- = 5
WEBSHELL+ = 6
WEBSHELL+ = 7
WEBSHELL+ = 8
WEBSHELL- = 7
WEBSHELL- = 6
WEBSHELL- = 5
WEBSHELL+ = 6
WEBSHELL+ = 7
WEBSHELL+ = 8
WEBSHELL- = 7
WEBSHELL- = 6
WEBSHELL- = 5
WEBSHELL+ = 6
WEBSHELL+ = 7
WEBSHELL+ = 8
...etc...

Comment 1

[Tobias Burnus]

This bug is INVALID (suggest as such, since I can mark it as such)

The reason for the reload is, that there is in the <header>:

<noframe><meta http-equiv="reload">

Since noframe in the header is *not* valid and http-equiv cannot be escaped, the
reload is done.

Comment 2

[J. Paul Reed [:preed]]

If this is *really* against the RFC's or W3, then that makes sense; but I
suppose *my* question is why NS 4.7x renders the page, if it's so horribly
standards-incompliant.

Well...maybe you don't need to answer that. :-)

But, this sort of html will...DOS mozilla of sorts (the renderer basically stops
while all of this yo-yo-ing is going on); would there be any way to either a)
just make it work, or (if that's horrendously unacceptable) b) make mozilla go
"I'm in a loop, and I shouldn't be...ignore this stupid html(tm)."

I actually figured it was something that the html-writer was doing that wasn't
standards compliant, aka stupid, but I still think it shouldn't throw mozilla
into a tizzy, and NS 4.7x *does* get it right.

Comment 3

[Eugene Savitsky]

See it on 2000 too. Marking as NEW for future decision.

Comment 4

[Phil Schwartau]

Browser, not engine. Changing component to Browser-General for
further assignment (not sure of the exact component on this one) -

Comment 5

[Asa Dotzler [:asa]]

->networking

Comment 6

[Jesse Ruderman]

related: bug 46115.  cc'ing owner of that bug.

Comment 7

[Doron Rosenberg (IBM)]

really over to networking now

Comment 8

[Gagan]

->ruslan

Comment 9

[ruslan]

Gagan,

why is this bug assigned to networking?

Comment 10

[J. Paul Reed [:preed]]

It should be noted the the page actually does display with build 2000091908 (a
bit old, I know).

The infinite redirect loop is still there, and fortunately, is a bit more
visible (the left content window keeps trying to get loaded in the main content
window) in these later builds.

Comment 11

[Gagan]

Hmmm... Now that I read more about it, I guess I have to mark it INVALID.

Comment 12

[J. Paul Reed [:preed]]

What was the reasoning for marking it invalid?

I'd be the first to say it's bad javascript or something else wacky on the part
of the site owner, BUT NS 4.74 DOES display this page w/o problems, and I think
that should be noted.

Isn't Mozilla supposed to be able to display everything the 4.x series can?

Comment 13

[Gagan]

No. Mozilla is not expected to repeat the same mistakes as NS4.* Unless you can 
convince me that this style of usage is of any use, I'd like to leave this 
invalid. 

Comment 14

[Tom Everingham]

verified INVALID