139531 - back button can't reload

Status: RESOLVED WORKSFORME

(Core :: DOM: Navigation, defect)

Description

[Fred Trinkoff]

I just downloaded Mozilla 1.0RC1 a couple days ago and have pretty much had this
problem ever since. Sometimes, when I try to load http://elf.hockey.cornell.edu
(which redirects me to list.php?f=1) I get an error message "The file
/list.php?f=1 cannot be found. Please check the location and try again." I
thought it was server errors at first, but then realized that it was working
fine in other browsers and for other people.

The way for the error message to go away is to quit the browser and open it back
up. However, if I use the page for long enough, it will soon happen again. I
have only seen this error first happen when I use the back button to attempt to
return to the list.php?f=1 page. However, after it happens by using the back
button, it will continue whenever I try to access the page in any manner until I
relaunch Mozilla.

I have discussed with owner of the site and he was able to recreate the error
after a while (his email address is [e-mail] if you wish to contact him,
but I don't believe he is currently registered w/ Bugzilla). We're both running
Mac OS X 10.1.3 while someone running 10.1.4 has reported no problems, but this
is only out of a sample of about 3 people. It appears to be some sort of back
button/cache error, from my limited knowledge. Thank you for your attention.

Comment 1

[Boris Zbarsky [:bzbarsky]]

To session history, for starters.

Comment 2

[Radha on family leave (not reading bugmail)]

I tried this  on latest trunk build in linux and W2K and I can not reproduce it.
I can go to several links of that redirected site, click back dorward or reload
and the page loads just fine. 

Comment 3

[Radha on family leave (not reading bugmail)]

I still cannot reproduce this bug in windows and linux. Unless it is a problem
only with Mac, I would like to mark this WFM. 

Comment 4

[Fred Trinkoff]

Though I don't know a lot about your system, I was under the impression that an
initial email which changed the status from UNCONFIRMED to NEW, as well as
changing the value of everconfirmed from 0 to 1, implied that it had been
confirmed on Mac OS X.

I have since upgraded to Mac OS X 10.1.4 and can confirm that the bug is still
around and completely legit. It is still happenening regularly. Thanks.

-Fred

Comment 5

[Fred Trinkoff]

I'd like to confirm that it still exists under Mozilla 1.0RC2 for OS X - please
take care of this as soon as you guys have a chance - it's quite annoying.

Comment 6

[Radha on family leave (not reading bugmail)]

Claudius: Can you confirm that this is indeed a problem with Mac OS X.

Comment 7

[Claudius Gayle]

I'll give this a whirl on OSX (I use 10.1.5).
Fred can you please let me know if any of your networking/cache/proxy preference
settings are anything other than the defaults?

Comment 8

[Fred Trinkoff]

Claudius-

I highly doubt I've changed any of those settings.

Cache:
Memory cache: 4096 K
Disk cache: 50000 K (looking at this, I might have raised this, could that be a
problem?)
Compare the page in the cache: When the page is out of date

Proxy:
No proxy, just 'direct connection to the internet'

HTTP Networking:
Use HTTP 1.1
Enable Persistent Connections: checked
Enable Pipelining: not chekced
Accept-Encoding: "gzip, deflate, compress;q=0.9"

Since I've reporting all this a while ago, I received this problem on one other
page. Interestingly, it was linked to from the forum I initially reported this
with, but when the problem occured I believe I was only auto-completeing the
URL. It was a forum or another thought, just a text page with a graduation
speech
(http://www.harvard-magazine.com/on-line/features/comm2002-classday-text.html).

These are the only two pages I've ever seen this problem with. Quitting Mozilla
and opening it back up always fixes it temporarily. And FYI, not surprisingly,
it occurs in Netscape 7 PR 1 too. Last, I am now also running 10.1.5, in case I
never mentioned that.

-Fred, [e-mail]

Comment 9

[Claudius Gayle]

as it stands I can't reproduce this bug with 20020716(or any) builds. On MacOS
10.1.5 I've been surfing through that forum all day (and previously) and have
not seen one instance of this failure.

If you have steps to reproduce that you can make consistently happen (ideally on
a different machine, or at least a new build and profile) then present them.
Otherwise, i'm forced to conclude tis bug 'worksforme'

Comment 10

[Fred Trinkoff]

Hmmmm, well, that's disappointing to hear because it's been plaguing me for at
least a couple months. Before you give up on this, is there anyway I could do
something by turning on some bug tracing system and send you the state of the
program when this happens? Some way for you guys to do some bug fixing from my
system? I'd be willing to help as much as possible, short of giving you access
to my computer :-)... plus my router is flaky with port forwarding anyhow.

Finally, if nothing else, could you at least recommend some troubleshooting I
can try before you discard the bug report? Thanks for anything we can do.

-Fred, [e-mail]

P.S. I'll be out of town next week, so if you respond later than tomorrow night,
please be patient for a week.

Comment 11

[Radha on family leave (not reading bugmail)]

marking this WFM since our QA is not able to reproduce this and it seems to be
working from other machines at the reporter's site.

You can turn on the logging for the network layer by setting the following
environment variables:

NSPR_LOG_MODULES=nsHttp:5 
NSPR_LOG_FILE=c:\http.log 

This should spit out messages from the network module and you can find out if
the redirected site is indeed being fetched from the server.