Evaluation

When complete, there are a number of ways that my system could be objectively evaluated:

• Measure the number of polygons plotted per second. This is a very widely cited benchmark, though how meaningful it is in real terms is doubtful.
• Can programs that use OpenGL be ported from other platforms relatively easily? It should at least be possible to compile and run simple graphics demos written in C.
• Is my library generally faster than the RISC OS port of Mesa? This could be assessed in timed tests.
• Do fellow programmers consider the balance between speed and sophistication of rendering to be reasonable? (I would need to solicit opinions on this matter.)
• Can my library be used from BASIC? Ideally novice programmers should, with the aid of a tutorial, be able learn OpenGL and write 'toy' graphics demos.

However, at the current stage of development (e.g. about half way through), none of the above points are really appropriate.

In terms of the tools that I created for my own evaluation and debugging purposes during development, we have the following:

• A test harness that allows access to SWI routines and other module facilities (assuming that the module is statically linked) through APCS function definitions that can be safely called from test programs written in C.
• A moving starfield demo, with depth-shaded particles moving at different velocities towards or away from the projection reference point, to test perspective projection and handling of very large numbers of primitives.
• A test program that plots a number of triangles/points/lines at random positions and depths, defining those vertices further away in darker colours, and those closer to the viewer in brighter colours. This allowed many bugs in the painter's algorithm implementation to be picked up by eye.
• A test program that plots a single primitive of the type to be tested, whilst allowing the vertices to moved around by keyboard controls, and the vertex colours to be changed also. This allowed many bugs in the basic primitive rendering routines to be picked up by eye.