When visiting the above URL http://areacode-info.com, the browser begins to build the page, and then crashes and prints the following message on the console: INTERNAL ERROR on Browser End: Pipe closed during read? State may be corrupt System error?:: Resource temporarily unavailable This happens every time. I tried it with Netscape 4.79, and it worked fine. Seen under Linux with Mozilla 0.9.6 (build ID 2001112012)

Confirming on Linux 2001121521, Java plugin 1.3.1_01. Seems to be Java related, no crash with Java disabled. In the third frame (http://areacode-info.com/navigate.htm, which by itself usually loads, but is not functional) Java console gives error java.lang.NullPointerException. Should be: Component: -> Java-something Keywords: -> Crash

OJI. likely a duplicate...

This is similar to 113957

still crasching with 2002021721.

Just to add another data point, since my original posting of this bug I've made the following changes: Upgraded from Redhat 6.2 to Redhat 7.2+errata (clean install) Upgraded from Java 1.3 to Java 1.4 Upgraded from Mozilla 0.9.6 to Mozilla 0.9.8 Still crashing, just like before.

Confirming on Linux and Solaris 2.8 with Mozilla 0.9.9+ and java 1.3.1

I think that the reason of the bug is last param tag of applet of http://areacode-info.com/navigate.htm. This parameter contains more than 900 string of characters. I'm going to check whether this is java plugin bug or this is bug of browser side.

I think this is java plugin bug. I load the changed http://areacode-info.com/navigate.htm by Netscape (I changed the 'applet' tag to the 'object' one). Netscape informs that 'java_vm has died'. When I reduced the last parameter both browsers worked fine. I filed bug #4657466 in Bugtraq against java plugin. I propose to close this bug as INVALID.

From bugtraq: ============= When we read data from the pipe we treate value passed as the length of a sting as a signed short instead of an unsigned short. Thus for long strings we get a negative index exception when we try and create the array. We can switch to using an unsigned short. This will fix this particular problem, but there will still be an inherent limit on the number of characters that a parameter can hold. The limit would be (2^16)-1 character. steve.katz@sun.com 2002-03-25 Fix is checked into 1.4.1 sources.

Chris Petersen is a new QA contact for oji component. His email is: petersen@netscape.com

fixing small error for pmac@netscape.com (filter with : SPAMMAILSUCKS)

reassign to me

->FIXED per comment 9

-- Verified in latest trunk build on Linux with JRE1.4.1_02. I am able to load the page fine without any crash. Marking Verified.