New OB release 2009-07-07
Tim Hutton
Changes since last release:
- bug fix: level 19 (cell) was crashing when caustic atom acted.
(thanks Calcyman for reporting, Bertrand for fixing)
- bug fix: level 14 (ladder) mis-reported 'middles must have 3 bonds'
error on success. (thanks Calcyman for reporting)
- several text typos fixed (thanks Calcyman for reporting)
- new 'reset simulation' button in settings dialog, useful for when
you change the number of atoms. (thanks Bertrand)
- internal changes (thanks Bertrand): testing framework, configuration file.
Remaining issues:
- Occasional (non-fatal, safely ignored) exceptions in updateImage at
max speed (multithreading issue caused by bonds changing during
draw?).
- Reloading the page containing the applet stops it working. This was
also true on the old version. I suspect we're not implementing the
full applet life cycle correctly.
- Logging of solutions isn't working. Sourceforge's changes broke
this. https://sourceforge.net/apps/trac/sourceforge/wiki/Project%20web%20and%20developer%20web%20platform#Read-onlywebspace
I need to retrieve the old persistent data and connect to the new
location.
- The separate language links on the "Get it, Test it" page loads an
applet that still asks the user what language they want. I just need
to update the php files to pass the language.choice parameter.
Some ideas for future directions:
- Move simulation.* parameters (currently in configuration.properties)
and initial world setup (currently in Java code) into a configuration
file (xml?) for each level. A new "world builder" interface allows the
user to design their own configuration and initial conditions, and to
share them. This idea is leading us towards allowing users to create
and share their own challenges/experiments, although the evaluation
step is a problem at the moment.
- New physics type: square lattice, with bond neighborhood of Moore or
larger and option to forbid bonds becoming crossed. While not all
levels will work in this physics (e.g. depending on options, cell
walls sometimes not easy to make) it shows an important point of ACs
which is that the physics is changeable. Should run faster too.
- New feature for continuous physics: static charge forces to repel
(and attract?) over a distance of N radii. The idea is to get phase
separation, and lipid bilayers too as these affect the chemistry by
making new structures possible: recognisable protocells. See
lipids.exe here: http://www.sq3.org.uk/Evolution/Squirm3/development/
All thoughts welcome,
Tim