The runnable jar / source is available here:
The changes are mostly internal:
- With non-random trail leaving enabled (it is by default), agents will no longer leave trails for longer than they were holding their pieces.
- Instead of having n different images for each n-sized puzzle, there is one image for each puzzle. Piece images are created from the image processor. This cut down on file size (~4MB to < 500 kb) and makes importing the project much easier.
- The program attempts to load a “masterInput.json” file in the current directory and on the desktop automatically, so it is no longer necessary to use the command line to load input files. If no file is available, the default image/agent selection screen appears. The masterInput.json file is available in the zip above.
- Agents no longer wonder off screen while leaving trails due to the reverse movement array being initialized with the wrong size.