#+TITLE: Speedcubing in Emacs
#+AUTHOR: Vasilij Schneidermann
#+DATE: November 2023
#+OPTIONS: H:2
#+BEAMER_THEME: Rochester
#+BEAMER_COLOR_THEME: structure[RGB={87,83,170}]
#+LATEX_HEADER: \hypersetup{pdfauthor="Vasilij Schneidermann", pdftitle="Speedcubing in Emacs", colorlinks, linkcolor=, urlcolor=blue}
#+LATEX_HEADER: \setminted{fontsize=\footnotesize,escapeinside=||}
#+LATEX: \AtBeginSection{\frame{\sectionpage}}

* Intro

** About

- Vasilij Schneidermann, 31
- Cyber security consultant at CODE WHITE GmbH
- mail@vasilij.de
- https://depp.brause.cc/
- https://emacsninja.com/

** Motivation and challenges involved

- I picked up speedcubing again
- Cubing hardware has improved a lot
- Scrambling and timing essential for getting better
- Seemingly all good options are web/mobile (cstimer, twisty-timer)
- No decent options inside Emacs
- Case study how to do better
- Making use of new Emacs features (svg.el, transient, sqlite)
- Most importantly: Having fun

** Prior art

- https://github.com/rafoo/stimer (timer, shows stats)
- https://github.com/param108/rubiks.el (solver)
- https://github.com/deadblackclover/scramble-el (text-only scrambler)
- https://github.com/queyenth/rubik.el (scrambler, unfinished timer)
- https://github.com/Kurvivor19/rubik-mode (cube simulator)
- https://gitlab.com/funkenblatt/elisp/-/blob/master/rubik.el (group theory)
- https://github.com/lukasjuhrich/cubing (algorithm notation)
- https://codeberg.org/akib/emacs-cube (3D cube simulator)
- None of those does it all :(

* The wca-prep package

** WCA what?

- Coming up with a good name was difficult
- World Cube Association holds events for twisty puzzles
- They standardized the puzzles, rules and notation
- Goal of my package: Helping people prepare for a WCA competition
- Therefore the name wca-prep was chosen
- This requires: Scrambler, scramble visualization, timer, stats

** Non-goals

- To get somewhere, scope needs to be kept small
- Focus on 3x3x3 cube for now, expand later
- Out of scope:
  - Solving the cube (the speedcuber does that)
  - Simulating a cube (simulated cubes are awkward to use)
  - Non-WCA puzzles (there are a ton, many 3x3x3 variations, some are
    challenging to visualize)
  - Exotic WCA events (fewest moves challenge, blind solves)
  - Specialized scrambles (helping with algorithm practice)

** Demo time

#+ATTR_LATEX: :height 5cm :caption \caption{User interface}
[[./img/screenshot.png]]

** Challenge: Puzzle representation

- Many possible representations for cube state
- There is no obvious best representation
- Since programmatic solving is out of scope, I picked the easiest
  possible one (array of numbered facelets)
- Awkward to do scrambles on, but otherwise good enough
- Future: Learn group theory, pick better representation

** Challenge: Scrambling

- After deciding on a representation, this should be easy
- A turn ends up swapping several facelets with other facelets
- Determining which facelets was very confusing
- One helpful suggestion: Make a paper cube with numbers on it
- ~cl-rotatef~ was highly useful for in-place swapping
- Future: Make use of group theory and better representation

** Challenge: Scrambling

#+ATTR_LATEX: :height 6cm :caption \caption{Two very different cubes}
[[./img/papercube.jpg]]

** Challenge: Scramble algorithm

- WCA: 3x3x3 cube is scrambled by generating a random cube, solving it
  and emitting the sequence of solve moves
- WCA: Solve moves are canonicalized (folding subsequent moves when
  possible)
- Example: Rotate 90° CW, rotate 90° CW -> rotate 180° CW
- Seemingly all other scramblers: Generate random moves, some
  canonicalize
- This scrambler: Generates random move, canonicalizes with previous
  move, repeats until desired amount of moves has been generated

** Challenge: Scramble visualization

- Mapping between every facelet and unfolded puzzle view was generated
- Out of scope: Facelet orientation (non-WCA puzzles would require
  this)
- Initial prototype used colored text
- svg.el turned out to be easy enough to use
- Future: Eliminate hardcoded puzzle face/color mapping for theming

** Challenge: Intuitive UI with transient

- Idea: Self-documenting, magit-style UI
- Transient has become part of Emacs, so why not
- Issue: Documentation is difficult to understand, unclear what best
  practices are or how things should be done
- transient-showcase helps with examples, they feel half-done though
- Issue: Unhelpful error messages when using transient wrong
- Future: Find more packages using it and try to make better use of it

** Challenge: SQLite book-keeping

- SQLite support landed in Emacs recently (29.1)
- It's still early days (found a bug in transaction macro)
- API is very basic, convenience helpers are missing
- The biggest challenge was figuring out a decent schema
- Finally no more persistence issues

* Outro

** Lessons learned

- There's still plenty packages for Emacs to be written
- The 3x3x3 cube does not require advanced mathematics
- However: It proves there's finally a real-world use for group theory
- The new Emacs features are good enough, but need more testing
- Call for action: Please make use of SQLite, report bugs, ...

** What next?

- More testing and UI improvements
- Better stats
- More puzzles
- ...

** Questions?