lab notes
While moving forward through the implementation, it wasn't clear how the indexing worked in detail. How do you determine what segments were different between the live and storage-backed indices? By definition, aren't all segments in the live index novel? But more worryingly, what is the key
lab notes
A good part of the week was spent reading and watching talks on "The simple essence of automatic differentiation". I got the gist the first go around but much of the details were opaque to me. Reading it this second time around and actually working through the examples
lab notes
Reading more papers, mostly focused on Relaxed Radix Balanced Trees, since other trees, such as B-trees, Finger Trees, and HAMT trees seem to predate the research on this one. The premise is generally pretty simple, but the devil is in the details. The papers on it are fairly clear. I
lab notes
Watched lectures and read papers on persistent trees, and come away feeling like it's a rabbit hole. I initially found it difficult to implement a B+-tree in Rust, due to ownership rules, so I stopped. As much of an advantage it would be to have these things implemented, it's a commodity.
lab notes
I spent a bunch of time mapping out the architecture of Datomic in detail. Diving into all the details revealed all the different places my knowledge was spotty, or where Rich Hickey glossed over the details. * Are the internal nodes of the storage-backed index only in memory, or are they
lab notes
For the programming environment, I wanted a database embedded in it. I had stuided the architecture of Datomic, and it seems a convincing architecture to solve the problem of creating offline apps distributed online. However, Datomic is proprietary, and there isn't an official corresponding client side database either.