If I developed a library during my PhD (in physics, not CS), but the topic is focused on what I used it for, rather than the code development itself, how much implementation detail should I give in my thesis? The entirety of the code was written by me (and my supervisor) and a significant portion of my time was spent on the development, but I'd also like to avoid going through every module and every class.
As the comment says, best talk to your adviser but some things to consider when writing about the implementation:
- Is there anything novel in the implementation?
- Do you want to provide information about how someone could run the same experiments as you?
- Did you use standard packages or algorithms which would help place your work in the correct context?
- Is the hardware you target relevant?
- Were there any difficulties in translating equations or algorithms into code?
If at all possible, avoid hiding hard work under a bushel!
It's common in pursing a PhD to have to do many things that appear only tangentially related to the "real" topic. However, frequently, those tangential things require ingenuity, creativity, insights and understanding, and just plain hard work. Unless the code is trivially obvious, and required no special understanding of the more central problem of your PhD research, I think that it would be a mistake to minimize it's importance. Moreover, by incorporating a discussion of the code into your thesis, you can show yet another aspect of the mastery you have of your subject matter.
It you cannot see an obvious way of putting the code into the body of the thesis, consider incorporating it into an appendix. Frequently, it is the unexpected, tangential things in one's research that end up being of more importance to the community of other researchers than the thing that was the primary focus. Consider, for example, Don Glaser, who was awarded the Nobel Prize in physics for the invention of the bubble chamber ... a side-show to his real interest which was investigating subatomic particles.