At 04:25 PM 1/19/00 -0800, Sarwate, Sandip wrote:
>Hi,
>
>My configuration:
>WinNT
>Emacs 20.4.1
>JDE 2.1.6beta16
>
>I am running a simple client that talks over an ORB (Visigenic Visibroker)
>to my Server object. I want to debug the server side code. Everything is
>running on the same machine.
>
>After running the orb, naming service and Interface repository, I invoke my
>server object within Emacs. I then start my debugger and then attach it to
>the the server process via shared memory - I tried using sockets too (I
>followed the steps outlined by Paul sometime earlier). The debugger connects
>to the server process. I then setup breakpoints in the appropriate files.
>The debugger tells me that breakpoints have been set.
>
>I then invoke my java client outside of emacs - on a cmd shell, and am
>expecting control to stop when the flow reaches the method where I have set
>the breakpoints. But this is not happening. The client runs through fine,
>without stopping atall.
>
Hi Sandip,
Setting a breakpoint requires both that the debugger be able to "resolve"
the breakpoint, that is, verify that the specified location actually
exists. Your *JDEbug* buffer shows that the breakpoints were specified but
never resolved, that is, in addition to this message,
Setting breakpoint at line 736 in STARServerImpl.java.
you should also see
Resolved breakpoint set in STARServerImpl.java at line 736.
I'm not sure why the breakpoints were not resolved. Make sure that you
compile your server with all debug information. Also, if you used jikes to
compile your server, recompile it with javac.
Another suggestion is to start your server from the debugger rather than
attaching to it.
But please note I have tested JDEBug's attach feature successfully on
MATLAB, a commercial application that embeds a Java vm and thousands of
lines of multithreaded Java code. I am able to attach to a running instance
of MATLAB, set breakpoints, and step through the code without any problem.
- Paul
