102371 - Visit this person's home page and watch Mozilla die horribly.

Status: RESOLVED WORKSFORME

(Core Graveyard :: Java: OJI, defect)

Description

[William Cattey]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.4) Gecko/20010913
BuildID:    2001091311

Just visit the site, and it kills Mozilla.

Reproducible: Always
Steps to Reproduce:
1.Visit web site
2.Mozilla dies before page finishes drawing
3.

Actual Results:  Mozilla death with an error log and traceback.

Expected Results:  Display of the page.

Comment 1

[William Cattey]

Attachment: Traceback supplied by Mozilla when it died #1 of 2

Comment 2

[William Cattey]

Attachment: Mozilla traceback #2 of 2

Comment 3

[Mike Young]

I can't confirm this crash, but From the logs it looks like this is with the
Java Plugin. Reassigning. (slap me if I'm wrong.)

Reporter, can you try removing and readding the Java Plugin and see whether that
helps. I really didn't see any Java on the page, but I guess it could've been in
an ad.

Over to OJI

Comment 4

[William Cattey]

Forgive my asking a stupid question.  I'm quite new to Mozilla, and the
nuances of installing and maintaining it.

By "Removing and readding the Java Plugin", do you mean that I should
fetch a new copy of jre1.3.1_01 from Sun and re-install it, and the
jre1.3.1_01/plugin/i386/ns600/libjavaplugin_oji.so that comes with it?

-wdc

P.S. Why can't I add a comment by replying to the Email I received?
I get:
    ... while talking to gila.mozilla.org.:
    >>> RCPT To:<[e-mail]>
    <<< 550 5.1.1 <[e-mail]>... User unknown
    550 <[e-mail]>... User unknown

Is this a tranient problem, or should I expect I just cant reply to bugs via email?

Comment 5

[William Cattey]

Ok.  I refreshed java and learned MANY things:

1. Trying 3 times following the link
"* Users of installer builds on Win32 and [x86 Linux] can install the Java Run
Time Environment (JRE) v1.3 Plug-in to add Java support to Mozilla. On
Macintosh, Mozilla uses the JRE already installed on your system." on the page
       http://www.mozilla.org/releases/
which takes you to
       ftp://ftp.netscape.com/pub/netscape6/english/6.0/unix/linux22/xpi/jre.xpi

The fetch just HANGS.

2. When I go to the relase notes: 
      http://www.mozilla.org/releases/mozilla0.9.4/notes
and follow the link:       
Windows and Linux: To run Java applets, you must install the  [Java Run Time
Environment (JRE)] plug-in. When using installer builds, everything should Just
Work without any help.

which takes me to:
       http://home.netscape.com/plugins/jvm.html

The download proceeds.  And then I get to the "waiting for installation to
complete, please restart" thing.  I don't know if it was elapsed time, or
running Mozilla as non-root and then rerunning it, but at any rate, the
install was successful as evidenced by my being able to go to
www.wheresgeorge.com and do java things.

3. I was not using that plugin.  I was using the SUN JRE, and the Sun JRE
was a LOT more robust.  I was using j2re-1_3_1_01-linux-i386.bin
and I believe I got it from:
       http://java.sun.com/j2se/1.3/jre/download-linux.html
(Retracing my steps, I see I opted for the tar not the RPM. Probably
because I wanted to control where it landed.)

Which brings us by TOO roundabout a path to the question you asked:

With the JRE pointed at by the release notes, the bug changes:
Symptom is Mozilla hangs before displaying any of the page.
The last think you see is "resolving adult.iamproud.com" in the status line
at the bottom of the window.

I REALLY don't like that JRE.  With it, my favorite streaming video site
does not work at all.
Go to:
        http://demo.emulive.com:81/CMD/HOMEPAGE/CH1
and click on any of the buttons to get a hang.  I tried "Tune In!" under
STANDARD/VIDEO.

Bottom line: Sorry.  I was using a better JRE that gave a graceful backtrace.
The correct bug report is "Mozilla hangs when viewing any web page with
complex Java"  The work around is to pull the plugin from Sun's JRE 
Standard Edition 1.3.1_01.  Unfortunately, that still does not remedy
the problems viewing
       http://adult.iamproud.com/~gifman/index1.htm

How do you wish to proceed from here?

Comment 6

[Mike Young]

OK, so presumably that means you are using JRE 1.3.0_01. That is old, and
JRE1.3.1 is the newer version. (Netscape however is still shipping with 1.3.0
and I don't know why.)

I just wanted to see exactly what happened when you reinstalled the JRE. I
cannot reproduce this on Win at all, with or without Java, so I can only assume
its linux related. (I'm adding it to my list of bugs to check on my linux box,
but I very rarely get over there.)

Reporter, could you provide what distribution of Linux you are running and any
"unusual" setups you may have. Also, are you running a Talkback enabled build?
If so could you tell us the TalkbackID of the crash you experienced. (If it
brought up a talkback window at all.) If not, could you provide a core dump if any. 

I'm inclined to believe this may be one of the many "Linux Java has problems"
bugs we have that will be fixed with JRE 1.4, however, if we get some more info
we'll be able to determine it for good.

Comment 7

[William Cattey]

I continually astounded by the number of web sites that work just fine
under Windows that do not work under UNIX.  I appreciate that there are
many sites, built with Microsoft Front Page that "optimize" for Internet
Explorer, but my experience is that Netscape, though available on both
platforms, demonstrates greater reliability under Windows. The thing that
has most excited me about Mozilla is that it alleges to be MUCH more platform
independent.  I'm REALLY surprised to find this is a "Works under Windows" bug.

Anyway...

Linux version:  Red Hat 7.1 from the commercially distributed CD.
There *IS* one unusual aspect to my installation:
It's VERY clean, and VERY vanilla. 
I installed 7.1 onto a brand new disk.  So there is NO legacy from previous
installs.

For what it's worth, I'm also running XFree86 version 3.3.6 (4.0 doesn't yet
fully support my video card.)

I am using the Talkback enabled build.  It did not run Talkback this time.
(It has run Talkback before and since. So I have a CLEAN test case on this
fact today too.  :-)  )

I see no core dumps.

Additional:  With the Netscape JRE there'd be one hung java_vm process 
persisting after I killed Mozilla.  If you like, I can run the test case
and try to send it a die and dump core signal.  (All I gotta do is remember
which signal that IS.)

Recognizing that this is probably a JRE bug, I'm surprised that Mozilla
has not developed immunity from completly hanging when the JRE mis-behaves.
This aspect to Netscape 4 is one of the things that drove me to Mozilla.
I mean GEE!  I thought UNIX had better sub-process support than Windows (95
certainly).  Is it the case that the Windows developers know they MUST 
debug the JRE sub-process code, whereas the UNIX developers just assume it works?

Comment 8

[Olivier Cahagne]

wfm using build 2001101103 on Win2k.

Reporter, can you try with latest Mozilla nightly build and report if problem
still  occurs ?
Build available here:
http://ftp.mozilla.org/pub/mozilla/nightly/latest/mozilla-i686-pc-linux-gnu-sea.tar.gz

Might have been fixed with bug 86591.

Comment 9

[William Cattey]

As instructed, I just this minute fetched the nightly build.
(ID 2001101208).

I went to the URL I reported.
Mozilla definitely did stay up longer, but it still died horribly.

I have attached the log file of the death.
Reminder:  I am running the Sun-supplied JRE j2re-1_3_1_01-linux-i386.bin
No other JRE is robust enough to keep running with some of the other
java-based sites I frequent.

Comment 10

[William Cattey]

Attachment: Back trace of horrible death using latest nightly build.

Comment 11

[Alex Vincent [:WeirdAl]]

Connection refused when trying to reach URL.

Mozilla/5.0 (Windows; U; Win98; en-US; rv:0.9.5+) Gecko/20011026

Comment 12

[Igor Nekrestyanov]

Site with testcase seems permanently down -> can not reproduce

Reporter: could you recommend any other testcase for this bug?

Comment 13

[William Cattey]

Alas, I cannot recommend other test case pages.
I was going through my old links when I stepped on this land mine.
I guess we'll have to close this one owing to lack of test case.
I'll periodically look to see if the site comes back online and if so,
I'll reopen the bug.  

Two favorable aspects:  The page won't blow up Mozilla any more  :-)
There was definitely some improvement to Mozilla before the site went offline.

I suspect there's something still not quite right in how Mozilla 
does sound that would have been easier to track down and fix if this
site had remained online.  Oh well...

I leave it to you guys to name the proper state for the bug.

Comment 14

[Mike Young]

while we do not have a page on which to reproduce this, we do have a talkback
trace to work with. However, this appears to be a problem with our
Java/sound/plugin/something implementation so I'd say lets mark this WORKSFORME
now in light of the work being done over there in Java Land. If this crops up
again, I'm sure there will be a bug filed. 

Marking WORKSFORME... but reopen it if there is any more data.

Comment 15

[Patty Mac]

Chris Petersen is a new QA contact for oji component. His email is:
[e-mail]

Comment 16

[Matthias Versen [:Matti]]

fixing small error for [e-mail] (filter with : SPAMMAILSUCKS)