Using 2/29 build on Windows NT 1. Install Seamonkey without Java 2. Read warren's digitalMASS email 3. Boom. Talkback ID 6082790 shows nsPluginInstanceOwner::GetParameter() on the tip of the stack trace.
Will try to get to it soon, but right now, verifying that the Java VM is working well is the top priority.
cc email@example.com on the chance that this is a plugins issue rather than an OJI issue.
I crashed again on http://www.widespreadpanic.com trying to load a page with a Java applet. Same stack trace. I'm getting the feeling that we just don't work without Java. If that's the case, I think we should consider this bug for beta1.
Reassign to av, cc amusil
phil, I have trouble reproducing it. What exactly should I do?
1. Install Seamonkey without Java 2. Load http://www.widespreadpanic.com 3. Click on image to enter site 4. boom
http://climate/reports/reporttemplate.cfm?style=0&reportID=1090 found 5.0nsPluginInstanceOwner::GetPar (170 Incidents) in the database. looks like some form of this crash happens on several sites with several m14 and post m14 users
phil, I still see no boom. Is there any chance that SeaMonkey is trying to pick up older Java plugin from 4.x installation, which is known to be incompatible? I don't have it on my machine maybe this is a reason I don't crash. Could you try to, say, rename plugin folder in your 4.x install and see if you still crash?
Yep, that seems to be the root of the problem. I renamed the Plugins directory from 4.72 and didn't crash. Then I renamed it back again, and we crashed again. Not sure what you mean by "known to be incompatible" though. Seems like Seamonkey has to run on machines which have 4.x installed on them...
Right. But is not it going to be distributed with Java?
Let me re-state: Seems like Seamonkey without Java has to run on machines which have 4.x installed on them. There are several installation options which do not include Java. If you believe that crashing when Java is not installed is ok for beta1, you can remove the PDT+ and the PDT will reevaluate the bug in light of what we've learned.
Hm... As per current implementation, SeaMonkey looks for plugins in 4.x installation too. Option of 'not including Java' sounds weird to me now. Adding ekrock. Adding firstname.lastname@example.org to the cc list. George, I think new Java plugin works with 4.x. Would it make sense to replace the older one in 4.x installation during SeaMonkey install?
av: I'll ask the JDK product folks for their opinion on this. It's not really my call, although your idea is an interesting one. I don't know if it fixes this bug, but I do like the idea that people would have a single Java plug-in for both versions of the browser.
Can't we fix Seamonkey so it only accepts a compatible Java plugin? I'd sure hate to have Seamonkey potentially disable 4.x installations if whacking the 4.x Java with a new Java didn't work. Unless that's had a ton of testing I'm not aware of, it seems pretty high risk.
Agree with Phil that we should definitely not incur the added risk for beta1 of having the Seamonkey installer overwrite whatever Java-support plug-in is present in the Nav4 installation. There is simply too high a risk of hosing someone's Nav4 installation and no time to test this. Instead, we need to make sure that the plug-in detection code (which currently scans the mozilla plugins directory first, then the Nav4 plugins directory) ignores any Java plug-in found in Nav4. Can this be easily done?
Yes, it's sorta one-liner. I can ignore npjava*.dll's in 4.x dir by their names. Alternatively, it can be done by mime types.
I have a fix. Can check it in as soon as I get it reviewed and approved.
does this also affect linux, cause the javaplugin from blackdown causes a crash on startup. I thought the plugin arch was backwards compatible. sPluginFile(/home/sford/mozilla/dist/bin/plugins/javaplugin.so) LoadPlugin() /home/sford/mozilla/dist/bin/plugins/javaplugin.so returned 826a358
As of 3/9, this was listed as fix in hand, waiting for review. Who is doing the review and not responding? We need to get this landed... the beta train needs to chug along. Thanks, Jim
I will check it in today.
av - let me know if you need a review
The fix is in both the branch and the tip.
Verified. No longer crashes in build 2000031506 beta build.
Works for me now too, using the 3/16 beta1 branch build.