Instructions for installing and using ACP3, version 1.0 ------------------------------------------------------- Updated: Wed Apr 29 05:47:10 1998 by Jeff Dalton ACP3 is distributed as a gzipped tar file. The current version is available from http://www.aiai.ed.ac.uk/~arpi/ACP3/1.0/ There you will find links to a README file, and INSTALL file, and a tar file. The INSTALL file is the same as the instructions you are now reading. The name of the tar file, when you get it, should be acp3-1.0-29Apr98.tar.gz ACP3 requires JDK 1.1.2 or later, and it's assumed that the corresponding "java" command is in the PATH of anyone using ACP3. The system has been tested only with JDK 1.1.5 and only on Suns running SunOS 5.5.1 (Solaris). To install ACP3: 1. Make a new, empty directory to contain the system. 2. Get a copy of the file acp3-1.0-29Apr98.tar.gz and place it in the directory created in step 1. 3. Unpack the release tar file there, e.g. as follows: zcat acp3-1.0-29Apr98.tar.g | tar xpf - 4. Make sure the subdirectory "acp3-logs" is writable by anyone who will be running ACP3. The directory you created above must be your current directory when you run ACP3. When you are in that directory, the script bin/acp3 will start ACP3. The bin/acp3-viewer script runs a version that does not do any KQML monitoring. To test the installation: 1. Make sure you are in the right directory and that the subdirectory "acp3-logs" is writable by anyone who will be running ACP3. 2. Type: bin/acp3-viewer When ACP3 starts up, two new windows will appear, and some debugging output will be printed as the system runs. Some window managers may not automatically place the windows at their intended locations, in which case they may need to be placed manually. Here is the suggested layout: +--------------------------------+ +------------------+ | ACP3 Main Frame | | ACP3 Information | | | | | | | | | | | | | | | | | | | | | | | +------------------+ | | | | | | | | +--------------------------------+ 3. Step through the demonstration by clicking on the numbered buttons in the Main Frame window. The buttons in the information window can be used at any time. 4. Ask ACP3 to run a simulation. First, click on the "Monitor" button to put ACP3 in the mode in which it displays the results of KQML monitoring. Then select "Test COA 1" from the "Test" menu. This will cause ACP3 to send itself a series of messages of the sort it would receive in normal use, to interpret those messages, and to display the resulting status changes as it determines that actions have started, or finished, execution. When you run this simulation, the effects should be similar to those described in steps 7 and 8 in the TESTING-KQML file (except for those parts that mention the "SendKqml" program). The simulation tests everything except the code that would receive actual KQML messages, and it can be run even when no KQML ANS server is available. 5. Click on the "Quit" button to exit. 6. So far, everything has been tested except the KQML message-handling code. If you'd like to test that, there are further instructions in the file TESTING-KQML in this directory. As mentioned above, the bin/acp3-viewer script runs a version that does not do any KQML monitoring. The following versions are available: bin/acp3 The "default" version. It is equivalent to bin/acp3-with-watcher. (It is a symbolic link to bin/acp3-with-watcher.) bin/acp3-viewer The interactive graphical interface set up to be used on its own. bin/acp3-with-watcher The interface plus the KQML monitoring code, for use in the TIE 97-1. To use this, you may need to edit the line that sets the ANS_HOST, or you can set an ANS_HOST environment variable. It is possible to set the variable and run ACP3 in one command by using the "env" command: env ANS_HOST=somehost.somewhere bin/acp3-with-watcher When ACP3 is being used as a process monitor that is observing KQML traffic, it will not display the results of monitoring unless the "Monitor" button has been pressed. The button affects only what is displayed. Monitoring happens automatically even when one of the numbered buttons has been pressed instead. This allows the user to choose whether to display the monitoring results or one of the numbered demonstration screens, and to switch back and forth between the two cases. The "Test" menu can be used to run a simulation in which ACP3 processes messages recorded during an earlier run. The messages are treated as if they were actual Kqml messages, and the "Monitor" button must be pressed in order to observe the resulting status changes. The "Test" menu entry "Create finished COA 1" can be used to quickly create a completely developed COA-1 in cases where COA-1 was produced before ACP3 started running, and ACP3 will see only messages about later COA. The "Reset" button can be used to discard the status information derived from monitoring, in case the TIE 97 process that is being monitored is restarted from the beginning, or to remove the effects of a simulation. It will also stop a simulation that has already begun. Running a new simulation does not automatically clear out the effects of earlier simulations (or of KQML monitoring). To clear those effects, press "Reset". After a "Reset", it is necessary to click again on "Monitor" if you want to see the monitoring results.