Community
Participate
Working Groups
I've registered the following listener in my code: public void eventFired(EventObject event) { if (event instanceof ProcessStateChangeEvent) { ProcessStateChangeEvent pscEvent = (ProcessStateChangeEvent) event; System.out.println(pscEvent.toString()); if (pscEvent.getEventId().equals( ProcessStateChangeEvent.EVENT_PROCESS_TERMINATED)) { if ((pscEvent.getSource() instanceof ProcessContext) && ((ProcessContext) pscEvent.getSource()).getID() .equals(context.getID())) { exitValue = pscEvent.getExitCode(); terminated = true; fireTerminateEvent(); } } } However, in my console the only event I am getting is the process creation event I am generating myself. Also, terminal window stays active forever.
Can you log the TCF communication for this case? One way of doing this is to enable "Enable logging" within the preferences (Window > Preferences > Target Explorer > Logging) and ZIP and attach the content of ${workspace_loc}/.metadata/.plugins/org.eclipse.tcf.te.tcf.log.core/.logs. The questions is if we get the terminated event from the agent.
Here is the log from the agent's command line(started as obj/GNU/Linux/i686/Debug/agent -l0x7C -L-): =================================================================== TCF 44:02.147: Starting channel 0x8100848 TCP:127.0.0.1:49402 TCF 44:02.147: channel server connecting TCF 44:02.175: Peer TCP:127.0.0.1:49402: Event: E Locator Hello ... TCF 44:02.175: channel server connected, remote services: TCF 44:02.175: Diagnostics TCF 44:02.175: Locator TCF 44:02.175: ZeroCopy TCF 44:02.212: Peer TCP:127.0.0.1:49402: Command: C 0 FileSystem stat ... TCF 44:02.223: Peer TCP:127.0.0.1:49402: Command: C 1 FileSystem open ... TCF 44:02.244: Peer TCP:127.0.0.1:49402: Command: C 2 FileSystem write ... TCF 44:02.296: Peer TCP:127.0.0.1:49402: Command: C 3 FileSystem write ... TCF 44:02.303: Peer TCP:127.0.0.1:49402: Command: C 4 FileSystem write ... TCF 44:02.310: Peer TCP:127.0.0.1:49402: Command: C 5 FileSystem fstat ... TCF 44:02.315: Peer TCP:127.0.0.1:49402: Command: C 6 FileSystem fsetstat ... TCF 44:02.318: Peer TCP:127.0.0.1:49402: Command: C 7 FileSystem close ... TCF 44:02.356: Socket is shutdown by remote peer, channel 0x8100848 TCP:127.0.0.1:49402 TCF 44:02.356: Closing channel 0x8100848 TCP:127.0.0.1:49402 TCF 44:02.356: channel 0x8100848 disconnected TCF 44:02.439: Deleting channel 0x8100848 TCF 44:02.499: Starting channel 0x8100848 TCP:127.0.0.1:49403 TCF 44:02.499: channel server connecting TCF 44:02.590: Peer TCP:127.0.0.1:49403: Event: E Locator Hello ... TCF 44:02.590: channel server connected, remote services: TCF 44:02.590: Diagnostics TCF 44:02.590: Locator TCF 44:02.590: ZeroCopy TCF 44:02.644: Peer TCP:127.0.0.1:49403: Command: C 8 Streams subscribe ... TCF 44:02.760: Peer TCP:127.0.0.1:49403: Command: C 9 ProcessesV1 getEnvironment ... TCF 44:02.842: Peer TCP:127.0.0.1:49403: Command: C 10 ProcessesV1 start ... TCF 44:02.978: waitpid: add pid 5516 TCF 44:03.069: waitpid: pid 5516 status 0, error 0 TCF 44:03.069: waitpid: pid 5516 exited, exit code 0 TCF 44:03.069: event: ctx not found, pid 5516, exit status 0, term signal 0 TCF 44:03.070: Peer TCP:127.0.0.1:49403: Command: C 11 Streams read ... TCF 44:03.111: Peer TCP:127.0.0.1:49403: Command: C 12 Streams read ... TCF 44:03.114: Peer TCP:127.0.0.1:49403: Command: C 13 Streams disconnect ... TCF 44:03.141: Peer TCP:127.0.0.1:49403: Command: C 14 Streams unsubscribe ... TCF 44:03.144: Socket is shutdown by remote peer, channel 0x8100848 TCP:127.0.0.1:49403 TCF 44:03.144: Closing channel 0x8100848 TCP:127.0.0.1:49403 TCF 44:03.144: Stream is left connected by client: VS0 TCF 44:03.144: channel 0x8100848 disconnected TCF 44:03.146: Deleting channel 0x8100848 ======================================================================
After launching my test run launch about 10 times in a row, I have to correct my statement: sometimes termination event gets sent and the terminal window becomes inactive as expected. However, it happened only once or twice. Also sometimes I see no output(in console there is a "Failed to read the output" message).
I am experiencing the same behavior with the TCF run launcher.
For me, this looks like an issue on the agent side. Probably the process is so short living that the agent cannot intercept the process properly before it dies again. Can you add a sleep at the end of the launched process (if it is one you have build by yourself)? And you may enable tracing for the processes listener. See the .options file in o.e.tcf.te.tcf.processes.core. It will print the tracing information to stdout. If you see "exited" called for the launched process ID and the terminal is still active, than this is something we can fix on the TE side. If you don't see "exited" called, than this is on the agent side.
I tried to reproduce it on Fedora 15 and OpenSUSE 11. The agent appears to generate correct event messages, including "ProcessesV1 exited". BTW, agent's log (-l option) is not a right way to see TCF channel traffic. In particular, -l log does not say anything about outgoing messages. To see all messages, you need to use a protocol logger, like TCF Trace view in Eclipse.
It works for me too on openSUSE 12.1. 15:27:43.721: ---> C 3 Streams subscribe "ProcessesV1" 15:27:43.723: <--- R 3 15:27:43.804: ---> C 4 ProcessesV1 getEnvironment 15:27:43.805: <--- R 4 [...] 15:27:43.809: ---> C 5 ProcessesV1 start "/home/uwe/tmp" "/home/uwe/tmp/HelloWorld" ["/home/uwe/tmp/HelloWorld"] [...] {"StopAtEntry":false,"StopAtMain":false,"UseTerminal":true,"Attach":false,"AttachChildren":false} 15:27:43.811: <--- E Streams created "ProcessesV1" "VS0" "P9240" 15:27:43.813: <--- E Streams created "ProcessesV1" "VS1" "P9240" 15:27:43.813: <--- R 5 {"Name":"[HelloWorld]","CanTerminate":true,"StdInID":"VS0","StdOutID":"VS1","ID":"P9240"} 15:27:43.818: <--- E Processes exited "P9240" 0 15:27:43.818: <--- E ProcessesV1 exited "P9240" 0 15:27:43.819: <--- E Processes exited "P9240" 0 15:27:43.820: <--- E ProcessesV1 exited "P9240" 0 # 2012-09-22 15:27:43.827: Closed channel 7e5f1c5e (error=null) 15:27:43.828: ---> C 6 Streams read "VS1" 1024 15:27:43.828: <--- R 6 (13)Hello World 0 true 15:27:43.836: ---> C 7 Streams disconnect "VS1" 15:27:43.840: <--- E Streams disposed "ProcessesV1" "VS1" 15:27:43.841: <--- R 7 15:27:43.843: ---> C 8 Streams unsubscribe "ProcessesV1" 15:27:43.843: <--- R 8 Anna, can you enable logging as described in my first comment (Comment #1)? Attach the generated log to the bugzilla so we can have a look.
Created attachment 221387 [details] logs After I pulled the latest from master, I fail to get any output in the terminal. Here is my console log: !ENTRY org.eclipse.tcf.te.tcf.processes.core 4 0 2012-09-23 18:42:57.170 !MESSAGE Failed to read data from stream with id 'VS13'. Possible Cause: Channel is closed !STACK 0 java.lang.Exception: Channel is closed at org.eclipse.tcf.util.TCFTask$4.run(TCFTask.java:97) at org.eclipse.tcf.EventQueue.run(EventQueue.java:95) at java.lang.Thread.run(Thread.java:619) !ENTRY org.eclipse.tcf.te.tcf.core 1 0 2012-09-23 18:42:57.184 !MESSAGE org.eclipse.tcf.core.ChannelTCP@104ee8: Channel closed. ; Possibly caused by: null ChannelStateChangeListener at trace/channelStateChangeListener (Mode 0) ChannelStateChangeListener.stateChanged( org.eclipse.tcf.core.ChannelTCP@104ee8, CLOSED ) ChannelTraceListener at trace/channelTraceListener (Mode 0) TraceListener.onChannelClosed ( java.lang.NullPointerException ) ChannelManager at trace/channelManager (Mode 0) Request to close channel. Target id = TCP:127.0.0.1:1534 ChannelManager at trace/channelManager (Mode 0) Closed channel. Target id = TCP:127.0.0.1:1534
In case you run it from a PDE environment, can you set a Java exception breakpoint for "NullPointerException" before you launch the process please? Send me the stacktrace and I'll take care of the NPE.
Here is the stack: Daemon Thread [TCF Event Dispatcher] (Suspended (exception NullPointerException)) TERunProcess.eventFired(EventObject) line: 161 EventManager$FireRunnable.run() line: 65 EventManager.fireEvent(EventObject) line: 344 ProcessProcessesListener.exited(String, int) line: 113 ProcessesProxy$1.event(String, byte[]) line: 128 ChannelTCP(AbstractChannel).handleInput(AbstractChannel$Message) line: 964 AbstractChannel.access$0(AbstractChannel, AbstractChannel$Message) line: 852 AbstractChannel$1$1.run() line: 254 EventQueue.run() line: 95 Thread.run() line: 619 And I'll attach the code.
Created attachment 221423 [details] code
Works for me now after replacing waiting for callback with listening for process creation event as Uwe suggested.
