Back to Clones

Ancient History

When I last wrote about clones, I was 32 and still working at Showcase IDX. I never got around to finishing clones and in fact worked on rawbones with my dear friend James Dabbs for a spell while teaching at the Flatiron School. By my count I have something like 4 half-finished NES emulators now.

• famiclom - My original half-hearted attempt in CL when I had no idea what I was doing.
• nescavation - A NES emulator I hacked up while learning to teach Javascript.
• clones - A better thought out attempt in CL.
• rawbones - The ReasonML version I wrote with James.

I seem to write one whenever I get bored and with any luck I'll wind up finishing one of them sooner or later. Nescavation and Famiclom really never got close to running games, clones and rawbones both got much closer to playable territory but I never got background scrolling right. I still find it a bit funny that famiclom gets more attention than my later, improved efforts like clones or rawbones. (Probably because cl-6502 mentions it and achieved a little notoriety.)

Getting to a playable state has never been the point though. These projects have been part learning exercise, part avenue for exploring literate programming, and often just a fun project to noodle with for my own entertainment. I still like the idea that a fast and reasonably accurate emulator can be written in a concise, clear way with a garbage-collected language.

Present Day

Recently, I got the itch again and so I decided to start fresh with clones. There are a few interesting changes this time around. When I made cl-6502, creating a readable document from the program was a primary goal and resulted in a literate book. This ethos never quite made the transition from the CPU stage to the full system emulators. This time I'll be leaning heavily into that spirit using mgl-pax. I'll also be testing with try and relying as heavily as I can on CPU and PPU test roms.

This is all happening in the "once-more-with-feeling" branch on sourcehut. So far there isn't a lot there though I'm on vacation starting in 6 days so I'm hoping to get ROM parsing and a basic structure for stepping the CPU in place to crank through NEStest. I do have some nice automation set up though. Every push runs the test suite and deploys the docs. I also have a very basic twitch stream working in case I want to indulge in the silliness of coding on camera.

For now, here's a look at the .build.yml file that powers the CI on sourcehut. It really isn't harder to set up an automatation pipeline for a CL app than anything else. Here's to working on fun projects again. More soon. 👋

image: alpine/latest
oauth: pages.sr.ht/PAGES:RW
environment:
  site: clones.kingcons.io
packages:
  - sbcl
sources:
  - https://git.sr.ht/~kingcons/clones
tasks:
  - install-quicklisp: |
      curl -O https://beta.quicklisp.org/quicklisp.lisp
      sbcl --non-interactive \
           --eval "(load \"~/quicklisp.lisp\")" \
           --eval "(quicklisp-quickstart:install)" \
      mkdir -p ~/quicklisp/local-projects/
  - test: |
      ln -sf ~/clones ~/quicklisp/local-projects/clones
      sbcl --non-interactive \
           --eval "(load (merge-pathnames \"quicklisp/setup.lisp\" (user-homedir-pathname)))" \
           --eval "(ql:quickload '(clones clones/test))" \
           --eval "(unless (try:passedp (try:try 'clones.test:test-all)) (uiop:quit 1))"
  - build-site: |
      cd clones
      echo 'Building site'
      sbcl --non-interactive \
           --eval "(load (merge-pathnames \"quicklisp/setup.lisp\" (user-homedir-pathname)))" \
           --eval "(ql:quickload '(clones mgl-pax/document))" \
           --eval "(clones.docs::build-site)"
      mv ~/clones/site/clones.html ~/clones/site/index.html
      tar -C site -cvz . > site.tar.gz
      acurl -f https://pages.sr.ht/publish/$site -Fcontent=@site.tar.gz
      rm site.tar.gz