308802 - java applet slowand appears not the right way - sometimes crashes camino

Status: RESOLVED WORKSFORME

(Camino Graveyard :: Plug-ins, defect, P2)

Description

[TeeWorks]

User-Agent:       Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8b4) Gecko/20050914 Camino/1.0a1
Build Identifier: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8b4) Gecko/20050914 Camino/1.0a1

on this car configurator site, camino doesnt show the java popup-hints and does
sometimes crash.

Reproducible: Always

Steps to Reproduce:
1.
2.
3.

Comment 1

[Steven Michaud [:smichaud] (Retired)]

What version of OS X are you using?

Comment 2

[TeeWorks]

(In reply to comment #1)
> What version of OS X are you using?

Its OSX 10.4.2

Comment 3

[Steven Michaud [:smichaud] (Retired)]

I don't know that the "java popup-hints" are.  Please tell me exactly
(step-by-step) how to make them appear in Safari.

I didn't experience any crashes.

Please attach here the stack traces for one or more of your crashes.

(The stack traces can be found in
~/Library/Logs/CrashReporter/Camino.crash.log.  Copy each trace to a
separate file and choose "Create a New Attachment" in this bug.  Don't
just paste the traces into comments.)

Comment 4

[Steven Michaud [:smichaud] (Retired)]

Also, in my tests the applet didn't load any more slowly in Safari
than it did in Camino.

Comment 5

[Steven Michaud [:smichaud] (Retired)]

> Also, in my tests the applet didn't load any more slowly in Safari
> than it did in Camino.

Oops!

The applet didn't load any more slowly in Camino than it did in
Safari.

Comment 6

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

I can certainly confirm the slowness (Cm 2005091504 (v1.0a1), Mac OS X 10.3.9
without the JEP-breaking latest Java upate), and sometimes the applet doesn't
draw at all.  The Spanish version (mentioned in bug 288554) appears very slow as
well.

However, Steven is right...it's very slow in Safari 1.3.1 as well, and has
drawing problems there.  My conclusion is that BMW changed something in their
applet to the detriment of Mac Java.

Comment 7

[TeeWorks]

(In reply to comment #6)

> However, Steven is right...it's very slow in Safari 1.3.1 as well, and has
> drawing problems there.  My conclusion is that BMW changed something in their
> applet to the detriment of Mac Java.

yes, ure right... its always slow (in other browsers too) ...main problem is the
applet itself, like you said, its not workin well (concerning the popups, i mean
those are mouseovers to inform about details like pricing, options etc. - which
dont work) 
... i now tested the old IE5.5:mac ...and its workin well... so its not due to
the java engine itself. In fireFox 1.0.6 its also basically workin (some
features dont work like switching to the interior pic of the car), only Camino
does definitely not render like it should.

Crashlog from Camino lateron ;)

Comment 8

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

There are identical display problems in any browser using Java 1.4.x, so those
we can't do anything about.

The crash perhaps we can do something about, if you provide a crashlog.

Comment 9

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

Attachment: Camino crashlogs

Well, I just got this to crash twice in a row while the applet was beginning to
load.  The second time Camino crashed--displayed the OS crash reporter--but
hung around for quite some time before disappearing (and never called
Talkback).

The crash logs are somewhat different; attaching both as one file.

Camino 2005101004 (v1.0a1+), 10.3.9 with Java Security Update

Comment 10

[Steven Michaud [:smichaud] (Retired)]

Neither crash takes place in any part of the JEP or in Apple's JVM.  In both
cases the thread 0 stack (where the crash takes place) is very long, and
contains lots of repeats.  Stack overflow caused by an infinite loop?

I think these must be a Camino bug, possibly triggered by something the JEP
or Apple's JVM is doing (though nothing in the other threads' stacks gives any
indication what that might be).

> Crashlog from Camino lateron ;)

Reporter:  Do you have any Camino crash logs you can attach?

Comment 11

[Steven Michaud [:smichaud] (Retired)]

I believe that BMW has changed their site recently -- for example, the "applet
loading" graphic looks unfamiliar.

But this is definitely a Camino bug -- a fairly recent regression.

I see the same crashes (on OS X 10.3.9) with the 2005-10-10-04 and
2005-10-05-04 builds, but not with Camino 1.0 Alpha 1 or Camino 0.8.4 (the
last two using JEP 0.9.4+a).

If they're sufficiently different to be interesting, I'll post some of my logs
later today.

Comment 12

[Steven Michaud [:smichaud] (Retired)]

The new crash happens with Java disabled (to be sure it's completely
disabled, restart the browser after unchecking "Enable Java").

It happens only on the 1.0 branch.

It doesn't happen with the 2005-09-30-04-1.0 nightly, but does happen
with 2005-10-04-04-1.0 and later nightlies.  The regression must have
appeared between these two dates.

Comment 13

[Simon Fraser [no longer active]]

Looks like we're recursing in nsChildView::UpdateWidget().

Comment 14

[Steven Michaud [:smichaud] (Retired)]

The new crash should probably block the next Camino beta/alpha.

Comment 15

[Simon Fraser [no longer active]]

I filed bug 312563 on the sans-Java crash.

Comment 16

[Steven Michaud [:smichaud] (Retired)]

Thanks for letting me know.  I've removed my blocking request.

Comment 17

[Simon Fraser [no longer active]]

Even with the crash fixed, this site still does very bad things to Camino.

With Java on, the applets never draw. ThreadViewer seems to indicate that one of
the Java threads is blocked in some InternString method (don't know if this is
normal).

The site works much better in Firefox.

Comment 18

[Steven Michaud [:smichaud] (Retired)]

You get these results with 2005-10-04 and subsequent Camino branch (and
trunk?) nightlies with your patch from bug 312563?

I didn't see anything so bad with 2005-09-30 and prior Camino nightlies
... though I do remember a little wierdness (part of the window (though not
the applet) temporarily blanked out that shouldn't have been).

Comment 19

[Steven Michaud [:smichaud] (Retired)]

Sorry about the blocking flag -- it reappeared somehow.

Comment 20

[Simon Fraser [no longer active]]

I was testing with my mostly up-to-date debug branch build.

Comment 21

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

This site seems "much" better currently (1.0 and latest trunk)--there are stilll drawing/refresh glitches (which I don't see in Safari any more), but it doesn't seem to hang/lock (comment 17)/crash any more.

Comment 22

[Stuart Morgan]

The test case isn't there any more, but given comment 21 I'm closing this WFM.