At the start of the project, it was decided to have free graphics that anyone could create and add. To make that a reality, the program needs to accept a known file format, such that a game extension can be stored in a single file (to simplify distribution).
The rcd/data_format.pdf file is the key-file here, as it contains an independent description of the format (separate of any implementation). It serves as the master description of the file format.
At the start of the project, a quick and dirty approach was taken to create the data files. With special-case Python code, and a lot of copy/paste, a solution was coded quickly.
This code has worked well for the past six months in the development of the C++ code, but it was already clear it needed further work at some point.
After finishing most of the path puzzles, I wanted to extend the data format. Doing that in the old and hacky code did not seem very good, thus the time had come to cleanup first.
The data-file format description has been moved from Python to XML. The
actual data is now also captured in XML format.
As an example of what the new format looks like for a graphics artist, consider the fragment from rcd/freerct.xml below (with some small additions to make it a complete XML file):
<rcdfiles>This describes the four arrows used for indicating the direction of path building. The data file is called 'buildarrows_8bpp64.rcd', and the data block containing arrows is called 'BDIR' (Build DIRection). It has five fields. The first field is the tile-width (so you can have different arrow sprites for different zoom-levels). Fields two to five are the four sprites, which are pulled from a sprite sheet, in the order north-east, south-east, south-west, and north-west (see the names attribute).
<file target="buildarrows_8bpp64.rcd" magic="RCDF" version="1">
<gameblock magic="BDIR" version="1">
<field name="width" value="64"/> <!-- Zoom-width -->
<sheet x-base="0" y-base="0" step="64" step="64"
names="ne,se,sw,nw" x-offset="-32" y-offset="-33"
It is still quite a lot of data, but I think it is a big step forward from coding Python to define both the file format and the actual data.