Emily
* Uses doctests instead of should-not-typecheck because should-not-typecheck does not check custom errors * Gets rid of `RemoveAccessTo` * `getFields` should be faster now
Add an option to automatically clear the compilation buffer when a --file-watch rebuild is triggered
(It would also be rather nice to have a customization switch to make this "on" by default)
> Is fromString . show what you meant by using show and then encoding into UTF8? Yep. > while whnf would just do the spine e.g. [x,x,x,x] right? `whnf` doesn't...
> We're only interested in benchmarking the speed of serialization right? So we don't actually need the write-to-file part for this at all do we? Yes.
> bigTree = expoAlphaTree 15 is lazy right, so bigTree is just a thunk. is this fact skewing my benchmarks or not? This doesn't matter because criterion runs benchmarks several...
> Even if I am, "real-world" usage would include the lazy part so actually having that in the benchmark makes sense? Real-world usage likely will do something with the tree...
> Do you see anything fishy? I've never seen such a wide range before (`88.75 ms .. 612.6 ms`). It looks pretty fishy.
See my rather more short solution here: https://github.com/neongreen/haskell-ex/blob/master/week1/logic-brute/neongreen/minified.hs (and a commented version here: https://github.com/neongreen/haskell-ex/blob/master/week1/logic-brute/neongreen/Main.hs).
I kinda feel like the whole section ``` haskell -- from this isConfigPositionSat :: TTTConfig -> TTTPosition -> Bool -- till this calcWinningRow :: TTTPlay -> (TTTPlay -> [TTTRow]) ->...
Your minimax is slow because it's executed lots of times on the same input. For the same reason a naive implementation of `fib` is slow: ``` haskell fib 0 =...