lateralus's Journal: End of project notes

I've decided to end development of my "big number" math library. It was a very cool project that taught me a lot about math and the C language. While the library is pathetically slow, it's mathematically sound. That is given enough time it would arrive at the correct answer. I've no intention on trying to be fast. There are enough "big number" math libraries out there that do that.

Looking back at the development process here is what I've noticed:

Understanding or at least estimating the big O complexity of your algorithms before you write them is useful.

Write everything properly the first time and you will not have to come back to it again and again.

No function is to small to write correctly.

Perform brave paradigm shifts and do not be afraid of major rewrites, thats what CVS is for.

Turning on all GCC's flags is essential.

Do the math -> abstract -> [ write -> debug -> profile -> debug... ] -> Stop when it's good enough.

GNU/Emacs rules.

Build test harnesses, plot and analyze your data.

Stop when you are not having fun anymore.