3 ms·
That’s a separate issue. The paper is concerned with the data structure not having gotchas (i.e. it is performant across a wide, continuous range of configurati
by bradleyjg 5y ago
That’s a separate issue. The paper is concerned with the data structure not having gotchas (i.e. it is performant across a wide, continuous range of configuration and input values) whereas nn3 is concerned about personally understanding the design of the data structure.
- ithkuil 5y agoThe two aspects are intimately tied. A library that implements an algorithm that has few gotchas and can be used intuitively requires less troubleshooting/debugging and understanding. Sure you need to trust the authors. I'm sure we regularly do such leap of faith all the time we use various software components we surely don't review on a daily basis, not sure why this particular tool should be judged a different standard (provided that is indeed straightforward to operate)