## Big changes

### cube type changess

* rename cube_t to cube_internal_t and cube_array_t to cube_t
* include only cube_t typedef in cube.h, remove ifdef from cube.h
* rework public functions: for many the simple implementation
  in the first section of cube.c is fine, other should first
  convert and then call the internal function
* for CO: move to bits 5 and 6, no need for padding bit

### Remove stuff from API, use more strings

* Remove move_t and trans_t
* Remove all functions related to trans, not useful for users
  (or maybe keep and let use transform? can see some use
  for it, in strange cases)
* Removes functions that read or write moves
* All functions should take strings instead of moves
* Performance is worse, more stuff must be done internally,
  expose only stuf that users are likely to use
* Benchmark: add some simple benchmarking functions to nissy.h,
  bench.c becomes very short

### More for moves

* keep move(cube_t, move), but prefer direct inline moves over it
* define macro to loop over moves e.g. #define FOREACHMOVE(action)

### API goals:

* manipulate move sequences (invert, unniss, cleanup, mirror / transform...)
* solvers (optimal, generic, coordinates)
* print cube (in various formats)
* print ptables (or layout data in such a way that can be printed
  easily, e.g. first bytes are null-terminated strig and can be
  printed by user)

### Solvers

* Actually do A*, no fixed depth
* Use threading (see below)
* Return strings, newline separated (see nissy_ffi)
* Instead of depth, I need the following parameters:
  int minmoves
  int maxmoves
  a parameter for all solutions / nmax / optimal / -O n
* How to make the above nicer? can it be done with a minimal
  amount of parameters (e.g. at most 2)?

### Rename to libnissy

* prefix public functions with libnissy_ or something similar
* move() that takes a string (alg) as input

## Solving

### Generic solver

* tests: solve full cube (max 7-8 moves?)
* more tests: eo and other stuff
* benchmarks

### Coordinates

TODO: specify in the comments that coordinates return 0 if solved

* [done] eo
* co
* ep
* epsep
* cp
* cpsep
* cphtr

What about symcoord?

### More solvers

* solve_light: first based on solve_generic, then optimize; benchmark
  to see up to what length it works best (7 moves? 10 moves?)

### Implement the following solvers:

* Slow: basic solver without any table.
* H48: one-bit-per-entry table + fallback, 48 symmetries and so on.
  See planner.
* nxopt31: mostly for comparison.
* other nxopt solvers: make generic and take the type as parameter.
* Step solver: take a coordinate function and a moveset as a parameter.

### New method:

* 48 symmetries, cocsep (or chtr, or similar) + epsep + some EO
* 1 bit per entry + fallback
* store necessary stuff (e.g. ttrep) all interleaved in the same table

### Other considerations:

* Reconsider going corners-first, so there is no need to sumco()

### Pruning tables

* ptable should contain some extra data at the beginning:
  an integer (size)
  a checksum
  some summary info, maybe even in text form

## Optimizations

### General things

* Trans: don't do full compose, for some trans composing perm is enough.
  Split out sumco() as a separate function and refactor, optimize.
* Use multi-move (up to 4/5 moves at once)
* CO is the worst part of moving, transforming and inverting. Try basing
  everything on representing the cube without CO and apply it only at the
  end to check that it is actually solved.
* see if vcube's method to flip all corners is better
* find a better way for computing the inverse?
* Improve avx2 instructions in general

### Threading

* THREADS build time option for the number of threads. If set to one,
  do not include any threading library or code. Try detecting at build
  time, or set to a sane default (e.g. 8? 16?) for generic builds.
* pthread or threads.h? I am more familiar with pthread, but threads.h
  is standard (from C11, so it requires switching to it from C99).
  Does using threads.h help in any way (e.g. building on Windows)?

## Improvements and other things

* NISS: Add mask to moves (e.g. U | NISS where NISS = 32 or something);
  adapt readmoves and writemoves.
* Consider adding centers and other moves (for avx2: centers in the
  same lane as corners, numbered from 9 to 14)
* More I/O:
  nissy
  ascii art (color = 1 letter)
  twizzle binary https://www.experiments.cubing.net/cubing.js/spec/binary/
  reid?

## "Front-end"

* Write adapter code for other languages:
  python
  hare
  rust, go
  dart ffi, js
  java
* add also example code (e.g. an optimal solver) in examples/