Running a Batch

Java Heap

Batch processing is a relatively resource intensive operation involving the creation and retention of large inheritance and dependency trees. The easiest way to increase the memory available to Eclipse is to edit the eclipse.ini file. As a guide, a medium sized Java project requires about 500 MB to run to completion.

The eclipse.ini file is located in the same folder as the eclipse executable on Windows. On Mac OS X right click on the Eclipse.app file and select Show Package Contents which will open a new finder window with the eclipse.ini located under the MacOS folder.

--launcher.XXMaxPermSize256m
-vmargs
-Xms40m
-Xmx512m

Note that is is important to avoid setting the -Xmx value greater than the physical memory available as this will only result in slow processing due to the use of paged memory and the postponement of Java garbage collection.

Running

A batch may be run directly from the wizard (see Creating a New Batch) or run, restarted or re-run.

The wizard creates a new folder called batch under the project or folder you specify. In that folder it creates a file batch.ajb which is read as the batch is run.

Right clicking on the batch.ajb file and selecting AgileJ StructureViews - Run/Resume invokes the batch processing.

If they do not already exist, folders are created in the batch folder as follows:

• descriptors
• diagrams
• logs
• web

descriptors

In the descriptors folder are kept the working files of the batch. These files hold the work-in-progress information which allows the batch to be resumed.

diagrams

The diagrams folder is used to contain the diagram files which may be opened by the AgileJ StructureViews diagram editor. This folder will remain empty until such time as the batch processing reaches the point where it is ready to start making diagrams.

logs

As the batch runs it opens a fresh log file (even if there have been previous runs). The name of the logfile is the current system time. The log files can be found in a folder called logs which is a child of the batch directory which also contains the batch.ajb file. The log is continuously updated with information regarding the progress of the batch run. Monitoring the log file is the normal way to watch the progress of a batch. Use of a logfile in this way rather than the usual Eclipse job progress monitoring creates a record which outlives the Eclipse session and makes it possible to monitor the progress of a batch remotely if the logfile is located in a shared directory.

web

The web folder contains the batch output which can be uploaded to a web server. This is the last of the folders to be populated.

However batch processing concludes, a message box announces that batch processing has stopped and provides a reminder to the location of the log file.

Cleaning a Batch

A batch can be cleaned and re-run without needing to start from scratch with the wizard. To clean a batch simply delete the folders which are siblings to the batch file leaving only the batch folder and the batch.ajb file. The folders to delete are called descriptorsToBeAssessed, descriptorsToBeIgnored, descriptorsToBeGenerated, diagrams, logs and web.