Community
Participate
Working Groups
DW (7/26/01 2:23:19 PM) Use of the shared memory transport may give improved debugger performance when debugging locally. Results on WIN98 - not too promising. IBM1.2.2 - VM GPF'd when speficied dt_shmem JDK1.2.2 - Failed to init transport - FATAL ERROR in native method. JDK1.3.0 - D:\workspaces\ws1\Test>\jdk1.3\bin\java -cp . -Xdebug -Xnoagent -Djava.compiler= NONE -Xrunjdwp:transport=dt_shmem,address=8000 Test Java HotSpot(TM) Client VM warning: Setting of property "java.compiler" is ignor ed Error accessing shared memory, rc = -1 Transport dt_shmem failed to initialize, rc = -1. FATAL ERROR in native method: No transports initialized JDK 1.4 D:\workspaces\ws1\Test>\jdk1.4\bin\java -cp . -Xdebug -Xnoagent -Xrunjdwp:transp ort=dt_shmem,address=8000 Test Transport dt_shmem failed to initialize, rc = 509. FATAL ERROR in native method: No transports initialized JGS (8/2/01 6:23:41 PM) The dt_shmem transport seems to work fine on 98, NT and 2000 on JDK1.2.2 & JDK1.3.0. JDK1.4 fails, but may be because of misplaced JDI DLLs. JGS (8/6/01 9:53:49 AM) The JDI DLLs are misplaced in the JDK1.4 beta. They ship in \jre\bin but should be moved to \bin. When this is done, dt_shmem works fine. JGS (8/6/01 2:16:40 PM) JDI Implementation does not currently support shared memory transport, so we cannot play with this in Eclipse or do timings to compare to socket transport.
Deferred
*** Bug 27955 has been marked as a duplicate of this bug. ***
In the past when I have used shared memory transport for debugging locally, I have had to use the classic vm instead of hotspot. Each time that I have started up a session where I have specified shared memory debugging if I did not specify the classic vm it would state something like the following: Java HotSpot(TM) Client VM warning: Setting of property "java.compiler" is ignored
This sounds to me like a warning - not an error. It just means that the command line option -Djava.compiler=NONE has been ignored.
Currently have no plans to implement shared memory transport.
Marking as "wont fix".
*** Bug 57963 has been marked as a duplicate of this bug. ***
Nowadays both NetBeans and JDeveloper support shmem debugging. Maybe this bug should be reconsidered to fix? I would like to stick in Eclipse but I need to persuade other team members to give up NetBeans or JDeveloper. Many thanks!
Currently, we do not have the resources to implement shared memory transport. Contributions welcome. However, how does using the shared memory transport change or improve the debugging experience?