I am checking back on Nim after 8 months or so. Previously I had tried to improve support for single precision and tried to make math.nim truly generic (like Julia), but I gave up after some effort.
I am curious as to the design philosophy of the designers of the language: do they envision supporting float32 and float64 on equal footing in Nim, or will float32 be the step-child always? Thanks.
Well it's hard to answer this question since I cannot recall you ever complain about the state of float32 vs float64 in Nim.
But given that both Urho3D and Unreal Engine 4 use float32 almost exclusively (and GPUs are much better at processing float32), I'm more than willing to improve Nim's float32 support.
Sorry, I did not mean to criticize. I edited some files a while ago to equalize support for float32, but I did not issue a pull request because I felt it was hackish. Moreover, major portions of math.nim are now defined to be non-generic and I wasn't sure exactly how to proceed. Should I start a new gmath.nim or did you envision at some point making math.nim generic? In any case my last query (many months ago) on this subject went unanswered.
In any case, from what I recall of the code, it is not probably a good idea for a newbie to mess with this part of the library, though I could help with testing as I would be using it. I am not a game developer, but a numerics person.